702 Street, NW, Suite 300, Washington, DC 20001 Tel: 202-462-1177 - Fax: 202-462-4507 20 February, 2015 The Honorable Lamar Smith, Chairman United States House Committee on Science, Space, and Technology 2321 Rayburn House Of?ce Building Washington, DC 20515 The Honorable Eddie Bernice Johnson, Ranking Member United States House Committee on Science, Space, and Technology 2468 Rayburn Of?ce Building Washington, DC 20515 Dear Chairman Smith and Ranking Member Johnson: I am writing to apprise you ofa situation concerning the United States House Committee on Science, Space, and Technology (House Committee on Science) inquiry into transparency and con?icts ofinterest in science. Last year, the House Committee on Science sent a letter to the US Environmental Protection Agency (EPA). That letter states that transparency in science is ?necessary to assure Congress and the American people? that government decisions are based on ?the best available science and not a predetermined regulatory agenda.? 1 share those concerns and would like to inform you ofa serious matter concerning con?icts of interest and lack of transparency at the Harvard-Smithsonian Center for Astrophysics According to documents made public to Greenpeace by the Smithsonian Institution under the Freedom of Information Act, it appears that the has been producing scienti?c studies for companies that fund their efforts, failing to disclose that funding to the public, and speci?cally failing to disclose the corporate funding ofthese studies in a conflict ofinterest statement when the studies are published in peer-reviewed literature. I will outline these concerns in greater detail, below. In January 2008, the Harvard-Smithsonian submitted a proposal to Southern Companies Services. That proposal was for a $60,000 grant to fund scienti?c studies to be published by Dr. Willie Soon. According to the proposal, ?expected outcomes? included ?Publication of both original and review papers on solar variability and climate change and various environmental impacts ofthat related change in leading scienti?cjournals for the advancement ofclimate and meteorological sciences.? ATTACHMENT Southern Company Services is a subsidiary ofthe Southern Company based in Atlanta, Georgia, one ofthe nation?s largest generators and distributors of electricity and a top corporate emitter of greenhouse gases. The agreement signed by the Harvard-Smithsonian and Southern Company Services states that published studies were part ofthe ?deliverables? that the Harvard-Smithsonian would provide to Southern Company Services (SCS). ATTACHMENT Speci?cally, the ?fth clause ofthe signed agreement reads: Deliverables. In consideration to SCS for its one (1) year funding contributions to the Project, Smithsonian will deliver to SCS a progress report ofthe ?ndings including a detailed summary and analysis ofthe results and ?ndings at the end ofthe one-year period. SCS shall be entitled to a no-cost, non?exclusive irrevocable license to utilize the data and results ofthe Project for its internal purposes. Other language in the agreement creates the appearance that the Harvard-Smithsonian Center agreed to hide this funding from public view; one clause states: Publicity. Smithsonian shall not publish and utilize the name or otherwise identify SCS or its af?liate companies in any publications or other advertisements without the express written consent of SCS. As further consideration to SCS, Smithsonian shall provide SCS an advance written copy of proposed publications regarding the deliverables for comment and input, if any, from SCS. In January of 2009, Harvard-Smithsonian sent Southern Company a one-year update on the progress ofDr. Soon?s scienti?c studies. That report states, ?The goals ofthis research proposal have been completely and successfully executed with the following list of deliverables.? ATTACHMENT The report to Southern Company then lists six different studies that Dr. Soon published in peer- reviewed sciencejournals. When we checked those studies, we discovered that funding by Southern Services Companies was not disclosed in any ofthe studies? acknowledgement sections. In some cases, those journals have policies that require disclosure of con?icts ofinterest and the sources of funding. In other instances, those journals do not appear to require disclosure of con?icts ofinterest, running counter to accepted practices in scienti?c publishing. For instance, in their report to Southern Company, the Harvard-Smithsonian listed Dr. Soon?s publication titled ?Validity of climate change forecasting for public policy decision making? as a deliverable to Southern Company. The published study in the International Journal of Forecasting does not acknowledge that it was created for Southern Company. This lack of disclosure is in apparent violation ofthejournal?s con?ict ofinterest policy. Speci?cally, thatjournal?s conflict ofinterest policy states: All authors are requested to disclose any actual or potential conflict ofinterest including any ?nancial, personal or other relationships with other people or organizations within three years of beginning the submitted work that could inappropriately in?uence, or be perceived to in?uence, their work. While we do not know if Dr. Soon disclosed his support from Southern Company to the International Journal of Forecasting, it is clear that he did not publicly disclose the support in the published piece. In a separate case, Dr. Soon published "Polar Bear Population Forecasts: A Public-Policy Forecasting Audit,? which Harvard Smithsonian listed as a deliverable to Southern Company Services. This study was published in the journal Interfaces, and Dr. Soon does not appear to have acknowledged that it was funded as a ?deliverable? to Southern Company Services. When we went to thejournal?s submission guidelines, we could ?nd no con?ict of interest policy. This is also a problem: many journals do not have published con?ict ofinterest policies that ensure transparency in science. We have provided you with a copy of each ofthe studies that the Harvard-Smithsonian Center provided as a ?deliverable? to Southern Services Company. Along with each study, we are also providing the con?ict ofinterest policy for the journal where the study was published. 1. "Polar Bear Population Forecasts: A Public-Policy Forecasting Audit." Interfaces vol. 38, 382- 404. J. Scott Kesten C. Green, and Willie Soon (2008) Acknowledgement: We thank Don Esslemont, Milton Freeman, Paul Goodwin, Benny Peiser, Orrin Pilkey, Tom Stewart, Mitchell Taylor, and two anonymous reviewers for the comments on earlier drafts. Janice Down and Kelly Jin provided editorial assistance. COMMENT: No apparent con?ict ofinterest policy. ATTACHMENT 2. ?Reply to response to et al. (2007) on polar bears and climate change in western Hudson Bay by Stirling et (2008)? Ecological Complexity, vol. 5, 289-302 by Dyck, Soon et. al. (2008) Acknowledgements: We thank Kesten Green, Mitchell Taylor, and Martina for important contribution oftheir expertise and knowledge in preparing this reply. Conflict of Interest: All authors are requested to disclose any actual or potential con?ict of interest including any ?nancial, personal or other relationships with other people or organizations within three years ofbeginning the submitted work that could inappropriately in?uence, or be perceived to in?uence, their work. ATTACHMENT 3. ?Centennial Variations ofthe Global Monsoon Precipitation in the Last Millennium: Results from ECHO-G Model,? Journal of Climate, vol 22, 2356?237 1 by Jian Liu, Bin Wang, Qinghua Ding, Xueyuan Kuang, Willie Soon, and Eduardo Zorita. Acknowledgements: Jian Liu and Bin Wang acknowledge the ?nancial supports from the Innovation Project ofthe Chinese Academy ofSciences (Grant the National Basic Research Program of China (Grant 2004CB720208), and the National Natural Science Foundation ofChina (Grant 40672210). Bin Wang and Qinghua Ding acknowledge the support received from the National Science Foundation (NSF) climate dynamics group (ATM06-295331) and American Meteorological Society, Author Disclosure Obligations: All funding sources should be identi?ed in the manuscript. Authors should disclose to the editor any ?nancial arrangement with a research sponsor that could give the appearance ofa con?ict of interest. ATTACHMENT 4. ?Validity of climate change forecasting for public policy decision making,? Kesten C. Greena, J. Scott Willie Soon, International Journal of Forecasting Volume 25, Issue 4, October?December 2009, Pages 826?832. Acknowledgements: We thank the nine people who reviewed the paper for us at different states of its development and the two anonymous reviewers for their many helpful comments and suggestions. We also thank Michael Guth for his useful suggestions on the writing. Official Publication ofthe International Institute of Forecasters Con?ict ofinterest: All authors are requested to disclose any actual or potential con?ict of interest including any ?nancial, personal or other relationships with other people or organizations within three years of beginning the submitted work that could inappropriately influence, or be perceived to in?uence, their work.1 ATTACHMENT 5. ?Multiple and changing cycles of active stars II. Results,? Astronomy Astrophysics, Volume 501, Number 2, July 112009 K. Olah, Z. Kollath, T. Granzer, K.G. Strassmeier, A.F. Lanza, S. Jarvinen, H. Korhonen, S.L. Baliunas, W. Soon, S. Messina, G. Cutispoto Acknowledgements: K.O. acknowledges support from the Hungarian Research Grants OTKA T- 048961 and T-0468626. S.B. acknowledges support from the Smithsonian Institution Restricted Endowment Funds and NASA NNX07AI356. KGS appreciates the continuous support ofthe Vienna-AIP APTs in southern Arizona through the State of Brandenburg MWFK. Comment: No apparent con?ict ofinterest policy. ATTACHMENT 6. ?Solar Arctic?Mediated Climate Variation on Multidecadal to Centennial Timescales: Empirical Evidence, Mechanistic Explanation, and Testable Consequences,? Physical Geography, Volume 30, Issue 2, 2009. Acknowledgments: I thank two referees for their constructive comments and edits that improved the paper. I also thank all colleagues whose works are cited here, and especially those who have allowed access to their hard-earned data series: Nicola Scafetta, Karin Boessenkool, Igor Yashayaev, Igor Polyakov, Mihai Dima, Lars Smedsrud, JeffKnight, Rob Allan, Daniel Hodson, David Holland, Mads Ribergaard, Frank Kauker, and John Fasullo. I thank Scott Gene Avrett, Sallie Baliunas, Dan Botkin, Bob Carter, Shaun Cheok, Susan Crockford, Bob Ferguson, Dave Fettig, Kesten Green, Joe Kunc, Keith Lockitch, Christopher Monckton, Lubos Motl, Jane Orient, Eric Posmentier, Art Robinson, Mitch Taylor, Bin Wang, and the late Robert Jastrow for their encouragement, and Gene Avrett and Steve Cranmer for their editorial help. I further thank Than, Lien and Julia Pham, Chiew-See Chua, as well as Benjamin and Franklin Soon for motivation. This new was based on a presentation at the 33rd International Geological Congress held in Oslo, Norway, August 6?14, 2008, as well as by another presentation at the International Symposium on Climate and Weather ofthe Sun?Earth System held at Jakarta, Indonesia, November 24-26, 2008. The views expressed herein are solely those of the author and are independent of sources providing support. Further information and an example ofa Conflict ofInterest form can be found at: Taylor and Francis Publishing Ethics: What constitutes a con?ict of interest and how to declare them A con?ict ofinterest can occur when an author or an author's employer or sponsor has a ?nancial, commercial, legal, or professional relationship with other organizations, or with the people working with them, that could in?uence that author?s research. Such a con?ict can be actual or potential. Full disclosure by the author, whether actual or potential, is required at the point of submission to ajournal. Once disclosed, the editor will use this information to inform editorial decisions and may publish such disclosures ifthey believe them to be important to readers in evaluating the article. Additionally, a decision may be made by the editor or peer reviewers not to publish, on the basis ofany declared con?ict. Personal con?icts of interest Potential con?icts ofinterest in relation to the submitted manuscript could include: Consultancies Employment Grants Fees Honoraria Patents Royalties Stock or share ownership If necessary, please describe any potential con?icts ofinterest in a covering letter, indicating funding ifit is more than per year. All funding sources supporting the work should also be fully acknowledged. Institutional con?icts of interest Ifyou are aware of your employer having any ?nancial interest in, or con?ict with, the subject matter or materials discussed in your manuscript please provide additional detail in your covering letter to the editor. Disclosure statement Authors should also include a relevant disclosure statement along with the text oftheir article, in conjunction with any acknowledgments and details offunders. Con?ict ofinterest: sample disclosure statements In accordance with Taylor Francis policy and my ethical obligation as a researcher, I am reporting that I [have a ?nancial and/or business interests in] [am a consultant to] [receive funding from] (delete as appropriate) a company that may be affected by the research reported in the enclosed paper. I have disclosed those interests fully to Taylor Francis, and have in place an approved plan for managing any potential con?icts arising from [that involvement]. OR This research is sponsored by [company and may lead to the development of products, which may be licensed to [company in which 1 have a business and/or financial interest. I have disclosed those interests fully to Taylor Francis, and have in place an approved plan for managing any potential con?icts arising from this arrangement. ATTACHMENT We appreciate your attention to this matter and applaud your attention to con?icts ofinterest and transparency in science. We are certain you will take this issue seriously on behalfofthe American public. Because transparency and conflicts of interest are such extremely important topics in science, we are providing copies of this letter to Chair and Ranking Members ofthe Senate Committee on Environment and Public Works, as well as the Chair and Ranking Members of the Senate. Should you need any further information, please do not hesitate to contact Jesse Coleman at iesse.coleman@greenpeace.org. Sincerely, 5mm Ann Leonard Executive Director Greenpeace cc: The Honorable Harry Reid, Minority Leader, United States Senate The Honorable James lnhofe, Chair, Senate Committee on Environment and Public Works The Honorable Barbara Boxer, Ranking Member, Senate Committee on Environment and Public Works The Honorable Rob Bishop, Chair, House Committee on Natural Resources The Honorable Raul Grijalva, Ranking Member, House Committee on Natural Resources ATTACHNIEN A ?13 Smithsonian Astrophysical Observatory Sponson Programs and Procurement Department 30 January 2008 Mr. Robert P. Gehn? Principal Research Specialist Research and Environmental Affairs Southern Company Services 600 North 13?? Street Birmingham AL 3529] Dear Mr. Gehri: The Smithsonian Astrophysical Observatory (SAO), per your request, is pleased to submit the attached Proposal P68 82-] -08 for a one (1) year Research Grant with Nonpro?t Organizations in the amount of $60,000 for Understanding Solar Variability and Climate Change: Signals from Temperature Records of the United States that could commence on 15 January 2008 and continue through 31 December 2008. The program will be conducted by the Smithsonian Astrophysical Observatory in Cambridge, Massachusetts. The program will be performed mder the direction of Dr. Willie Soon, as the Principal Investigator, within the Solar, Stellar, and Planetary Sciences Division, with Dr. Nancy Brickhouse as the Associate Director of the Division. inquiries of a technical nature should be directed to Dr. Willie Soon, Mail Stop 16, Smithsonian Astrophysical Observatory, 60 Garden Street, Cambridge, Massachusetts 02138-1516, telephone (617) 495-7448, e-mail wsmn@cfa.f harvardedu. Inquiries and documents of a contractual nature should be directed to Mr. William 3. Ford, Contract and Grant Specialist, Mail Stop 23, same address, telephone (617) 495-7317, e-mail Mo: rd@gfa.harvgd.edu. Sincerely yours, Charles Alcoclt Director Enclosure SMITHSONMN INSTITUTION 60 Garden Street Cambridge MA 02130-1516 6! 7.4917000 Telephone SI-000039 PROPOSAL TO SOUTHERN COMPANY SERVICES FOR UNDERSTANDING SOLAR VARIABILITY AND CLIMATE CHANGE: SIGNALS FROM TEMPERATURE RECORDS OF THE UNITED STATES P6882-1-08 at the period 15 January 2008 through 31 December 2008 January 2008 Smithsonian Institution Astrophysical Observatory Cambridge. Massachusetts 02138 '1 The Smithsonian Astrophysical Observamry is a member of the Harvard-Smithsonian Center for PROPOSAL TO SOUTHERN COMPANY SERVICES FOR UNDERSTANDING SOLAR VARIABILITY AND CLIMATE CHANGE: SIGNALS FROM TEMPERATURE RECORDS OF THE UNITED STATES PP6882- l-08 For the period 15 January 2008 through 3 December 2008 Funds Requested: $60,000 Princing Investigator Agog-[mm Solar, stellar. and Piggy! Diyisign Dr. Willie Soon Dr. Nancy S. Briekhouse January 2008 Smithsonian institution Astrophysical Observatory Cambridge, Massachusetts 02138 Director: Dr. Charles Alcock The Smithsonian Atrophysical Observatory is a member of the Harvard-Smithsonian Center for Astrophysics Si-000041 1" a} Smithsonian Astrophysical Observatory Understanding Solar Variability and Climate Change: Signals from Temperature Records of the United States A Proposal to The Southern Company Dr. Willie Soon, Principal Investigator Smithsonian Astrophysical Observatory Solar, Stellar and Planetary Sciences Division (617-495-7488; wsoon@cfa.harvard.edu) January, 2008 Research Target and Proposal: This proposal seeks $60,000 from The Southern Company for year one of this two-year project, ?Understanding Solar Variability and Climate Change: Signals ?~om Temperature Records of the United States. I propose to conduct an intensive up-to-date science review of solar variability and climate change (see Soon 2007a), with emphasis on the signals from temperature records of the U.S., that will be a clear improvement of previous studies. The goals for the ?rst year are to collect and assess the scienti?c quality of the available temperature records from the United States, aggregated into four inter-related spatial domains: l) a rural city a city that is minimally disturbed by urban development), 2) an individual state, 3) regional U.S. area, and 4) the whole contenninous U.S. The goals for the second year are to study any plausible connection of these U.S. temperature records with estimated solar irradiance history for the past I l2 years from 1895 to 2006. The previously published research paper by Soon (2005) identi?es both the multidecadal variation in total solar irradiance and the l-year solar UV irradiance forcings to be important in explaining the observed Arctic surface air temperature change over the past [30 years or so. The overall goal for this 2-year program is to extend our basic understanding on how the variable solar irradiance outputs could be physically connected to the Earth climate system. The ability to con?rm or reject the statistical correlations shown in Figure I will be of enormous scienti?c importance. The ultimate physical understanding will arise from detailed assessments on how the solar irradiance is related to the cloud ?eld as well as how the solar irradiance may systematically and persistently modulate the land surface heat ?uxes sensible and latent heats) on multidecadal to centennial time scales. A parallel hypothesis regarding the role of rising atmospheric carbon dioxide (see Soon 2007b) in warming the surface temperatures of the United States on these 4 spatial scales will also be evaluated. Sl-000042 rem solar Irradlence Total Solu- In'adlanu 1m I 'Ih'il 1 (1) mat-am. Its . 1357 a; I "Hg l36li .3. was it. an; 13cc - '5 554:1 I954 i i .I. .I. 1353: I I 1890 1900 1920 194-0 1900 1980 2000 1880 1900 1920 1940 1900 1980 2000 lm I v-f?T-i? I?r'r?l a. Cantu-l 0:5. Cant-ruinou- U.S. 136'! 5 i - . 136? . A ?1 I. I lead . nae.- . 1386. . .g 5 a: as: 1364 1364 5 1383 1m 1030 1900 1920 1940 1900 1900 3000 1000 1900 1920 [940 1900 1950 3000 Year Year Figure l: A plausible connection of the solar irradiance (red curves in all four plots; based on Hoyt and Schatten l993-rescaled to the mean absolute valueI measured by the ACRIM radiometers) compared with U.S. temperature records in 4 spatial domains (the blue dotted curves are for l) Atchinson, KS, 2) state of Kansas, 3) Central region of the U.S.. and 4) contenninous U.S.). These results extend the previous relation found for the Arctic shown in Soon (2005). The scienti?c hypothesis for this sun-climate relation will be care?illy formulated and examined in the proposed project. [Temperature Data Source: U.S. National Climatic Data Center, hilly [lwf.ncdc.noaa.gov( 0a 1 climate researchz cag3?cag3.html I. 1 Soon (2007a) calls for the solar physics community to firmly establish this value emphasizing its great importance in establishing the mean climatology in climate models. The mean climatology in climate models can be subjected to a rather arbitrary tuning given that the absolute level of total solar irradiance is not determined to any level of con?dence, with values ranging from 1372 to 1360 m2. 2 Sl?000043 Expected Outcomes: (1) Publication of both original and review papers on solar variability and climate change and various environmental impacts of that related change in leading scienti?c joumals for the advancement of climate and meteorological sciences. (2) DeveIOpment of tools, including power-point presentations and concise scienti?c essays, for unbiased and more accurate science accounting that will more powerfully serve informed public policy making. (3) Better public education with active participations by the Pl of this research proposal in all national and international forums interested in promoting the basic understanding of solar variability and climate change. Research Team: Dr. Willie Soon at the Smithsonian Astrophysical Observatory, which is part of the Harvard-Smithsonian Center for Astrophysics, will lead and direct this scienti?c research program. in addition, the Pl may solicit interests for collaborative effort from interested colleagues at no additional cost to the proposal. Funding Request: The funding is primarily to support approximately 3.5 months of the full-time research work of Dr. Willie Soon at the Smithsonian Astrophysical Observatory and a small amount of travel to a scientific meeting or publication costs. This research proposal requests $60,000 from the Southern Company for work to start January, 2008, extending for a duration of about one year. References Hoyt D. and Schanen K. H. (I993) A discussion or plausible solar irradiance variations, 11004992. Journal of Geophysical Research 98 (Al I). l8895-I8906 [with updates from Dr. Nicola Scafena, Duke University. private communication May 31. 2007]. Soon W. (2005) Variable solar irradianee as a plausible agent for multideeadal variations in the Arctic-wide surface air temperature records of the past 130 years. Geophysical Research Letters 32'. Ll67l1. Soon W. (2007a) Some Issues of Solar lrradiance Variability and Climatic Responses: A Brief Review. lnivited Talk GC42A-05 at the American Geophysical Uni0n Fall Meeting (December IO-M, 2007). Soon (2007b) Implications ofthe secondary role of carbon dioxide and methane forcing in climate change: Past, present. and future Physical Geography 28, 97425. Sl-000044 ESTIMATE OF COST Period of Performance: January 15, 2008 through December 31, 2008 Productive Labor: Dollars Dr. Vl?ilie Soon. Pl 494 $25,209 Program Administration 8 $495 Secretary 20 Total Productive Labor _5_22 26.311 Leave 19.5% 5.191 Total Direct Labor 31.442 Fringe Bene?ts 26.5% Direct Operating Overhead Base 39.774 Direct Operating Overhead 30% 11.932 Travel schedule 1.789 Printing and Reproduction - see schedule 1.050 8. A Base 54.545 A 10% 5.455 TOTAL ESTIMATED COST $0.900 81-000045 TRAVEL SCHEDULE NO more: RATE Tor AIR TOT TOTAL TRIPS Imp PER DIEM PER DIEM FARE AIR FARE cosr Sdan??c Mutiny-Sun Francine 1 1 204 31.020 3500 5500 $299 51,789 TOTAL TRAVEL 31.029 55.40 5259 $1.139 'lncludas local Inasmuch wall and moallng lass PRINTING AND REPRODUCTION SCHEDULE COST TOTAL BASIS OBJECT GLASS QEBORIPBON CQST Ell Plan Charon Aauuphyalcal Journal 31,050 Plans 10 call Per Page 103 TOTAL PRINTING AND REPRODUCTION $1.050 81-000046 CONTRACTUAL AND COST INFORMATION INCLUDING CERTIFICATIONS The Smithsonian Institution, an independent trust establishment was created by an act of the Congress of 1846 to carry out the terms of the will of James Smithson of England, who had bequeathed his entire estate to the United States of America ?to found at Washington, under the name of the Smithsonian Institution, an Establishment for the increase and diffusion of knowledge among men." After accepting the trust property for the United States, Congress vested responsibility for administering the trust in a Smithsonian Board of Regents. The Smithsonian performs research, educational and other special projects supported by grants and contracts awarded under the cost principles of the Federal Acquisition Regulation, Subpart 31.7 Contracts with Nonprofit Organizations. It is audited by the Defense Contract Audit Agency, Landover, Maryland. The Charter of the Smithsonian Institution carries a mandate for the "increase and diffusion of knowledge among men.? Therefore, any grant or contract that may be awarded as a result of this proposal must be unclassified, in order not to abridge the Institution's right to publish, without restriction, findings that result from this research project. Considering the nature of the proposed effort, it is requested that a Research Grant with reimbursement via electronic funds transfer be awarded to cover the proposed project in accordance with Subpart Section .22(e) of OMB Circular No. A-llU dated 30 September 1999. Pursuant to Subpart C, Section .33 and .34 of OMB Circular No. A-llO dated 30 September 1999, it is requested that title to all exempt property and equipment purchased or fabricated under the proposed contract be vested irrevocably in the Institution upon acquisition. In accordance with an agreement between the Office of Naval Research and the Smithsonian, the Institution operates with predetermined fixed overhead rates with carryeforward provisions. For Fiscal Year 1996 and beyond, the Indirect Cost and Fringe Benefits Rates are developed in accordance with the Office of Management and Budget Circular (OMB) A-122: Cost Principles for nonprofit organizations. The following approved rates, provided by ONR Negotiation Agreement dated 2 November 2007, shall be used for forward pricing and billing purposes for Fiscal Year 2008. The Fringe Benefits Rate will be applied to the Total Direct Labor Costs. The Material Overhead Rate will be applied to the cost of materials, equipment and subcontracts. The Direct Operating Overhead Rate will be applied to the Direct Labor and Benefits costs. The Rate will be applied to the base consisting of total costs except the costs associated with the materials, equipment and subcontracts. SF000047 The following Approved Rates shall be used for forward pricing and billing purposes for Fiscal Year 2008: Material Burden Rate 5.4% (Cost of Materials, equipment and subcontracts) Personnel Leave Rate 19.5% (Total Direct Labor Costs less paid leave and training (Productive Labor}) Fringe Benefits Rate (Full/Part Time Employees) 26.5% (Total Direct Labor Costs) Fringe Benefits Rate (Intermittent Employees) 8.5% (Total Direct Labor Costs) Direct Operating Overhead Rate 30.0% (Total Direct Labor and Fringe Benefits Costs) General and Administrative Rate (Gen) 10.0% (Base consists of Direct Operating Activities less Net Costs Associated with materials, subcontracts and equipment) Central Engineering Overhead Rate 23.9% (Central Engineering Direct Labor and Benefits Costs) Rate Verification can be made by contacting Ms. Linda Shippa-Office of Naval Research, Indirect Costs/ONE 242, 800 N. Quincy Street, Room 704, Arlington, Virginia 22217, telephone (703) 696-8559, or e-mail Engineering services are provided by the Central Engineering Department as a Cost Center. Charges by the department to research projects are inclusive of Direct Labor, Fringe Benefits, and Central Engineering Overhead. Pursuant to Executive Order 12549 and implementing rule (FAR the Smithsonian Institution certifies that it presently is not debarred, suspended, proposed for debarment, declared ineligible or voluntarily excluded from covered transactions by any Federal department or agency. Pursuant to Section 1352, Title 31, United States Code (USC) and implementing rule (FAR the Smithsonian Institution certifies that no Federal appropriated funds have been paid or will be paid to any person for influencing or attempting to influence an officer or employee of any agency, a Member of Congress, an officer or employee of Congress, or an employee of a Member of Congress on his or her behalf in connection with the awarding of any Federal contract, the making of any Federal grant, the making of any Federal loan, the entering into of any cooperative agreement, and the extension, continuation, renewal, amendment or modification of any Federal contract, grant, loan or cooperative agreement. SL00004B ATTACHMENT AGREEMENT FOR FUNDING A GRANT TO SMITHSONIAN ASTROPHYSICAL OBSERVATORY THIS AGREEMENT is entered into by and between the Smithsonian Astrophysical Observatory, located at 60 Garden Street, Cambridge. MA 02138.15 16. hereinafter referred to as ?Smithsonian"). and Southern Company Services, lnc., having its principal place of business at 600 North 18th Street. Birmingham. Alabama 35203. on behalf of itself. its parent and its af?liate companies. (collectively referred to as WHEREAS, the Smithsonian is interested in conducting an intensive science review of solar variability and climate change, as provided in the attached Proposal 1-08 (referred to as the ?Project"); and. WHEREAS, SCS. on behalf of itself, its parent and its af?liate companies is interested in furthering the research on the Project and in obtaining advance information and is therefore willing to make a grant to fund this research. NOW, THEREFORE, Smithsonian and SCS hereby agree as follows: 1. Scope of Work. The Scope of Work for this Project shall be conducted in accordance with the attached Proposal P68824403 entitled ?Understanding Solar Variability and Climate Change: Signals from Temperature Records of the United States". which is incorporated and made a part of this Agreement. In consideration of the Research to be provided by Smithsonian. SCS agrees to make an advance payment in the sum of Sixty Thousand Dollars ($60.000.) and to reimburse Smithsonian for its costs in accordance with the Proposal in an amount nor to exceed the advance sum. 2. Limig Natum of Parties Obligations. The obligations of SCS and the Smithsonian hereunder shall be limited to payment of the amounts and the Project effort as speci?ed in Article 1 above. SCS assumes no other obligation or responsibility of any kind to the Smithsonian or any other participants or sponsors. if any. SCS makes no warranties or Representations, Express or implied. of any kind. 3. Termination. Smithsonian understands and agrees that in the event the Project is terminated prior to completion or is not in accordance with the attached Proposal. SCS shall be entitled to a refund of the uneJtpended funds. 4. No lloint Venture. This Agreement is not intended to create nor shall it be construed to-create any partnership. joint ventur?. employment or agency relationship between or among the parties. and no party shall be liable for the payment or performance of any debts. obligations. or liabilities of any other party. unless expressly assumed in writing. 5. Deliver-ables. in consideration to SCS for its one (1) year funding contribution to the Project. Smithsonian shall deliver to SCS a prOgress report of the ?ndings including a detailed summary and analysis of the results and ?ndings at the end of the one year period. SCS shall be entitled to a no-cost. non-exclusive irrevocable license to utilize the data and results of the Project for its internal purposes. ft. Authority. Each party represents and warrants to the other that as of the effective date of this Agreement: it has all requisite power and authority to enter into and perform its obligations under this Agreement, and there are no actions. suits. or proceedings pending. or to the best of its knowledge threatened. which may have a material adverse effect on its ability to ful?ll its obligations under this Agreement or on its operations. business. properties. assets or condition. Sl-000036 9. 10. 11. 12. 13. 14. Assignment and Subcontracting Prohibited. This Agreement shall not be assigned by Smithsonian nor its obligations subcontracted without the prior written consent of SCS, which shall not be unreasonably withheld. Any assignment or subcontracting in violation of this provision shall be deemed null and void and SCS shall be entitled to a refund of its contribution in full. Subscgucnt Changes in Agreement. This Agreement may be modi?ed only by an amendment executed in writing by a duly authorized representative for each party. Eartial Invalidig. If any provision of this Agreement is found to be unenforceable then. notwithstanding such unenforceability. this Agreement shall remain in effect and there shall be substituted for such unenforceable provision a like but enforceable provision which most nearly effects the intention of the parties. If a like but enforceable provision cannot be substituted, the unenforceable provision shall be deemed to be deleted and the remaining provisions shall continue in effect, provided that the performance. rights. and obligations of the parties hereunder are not materially adversely affected by such deletion. and Assigns. This Agreement shall inure to the bene?t of and be binding upon the respective successors and permitted assigns, if any. of the parties, provided that this provision shall not be construed to permit any assignment which would be unauthorized or void pursuant to any other provision contained herein. Non-Waiver. No provision of this Agreement shall be deemed waived and no breach shall be deemed excused unless such waiver or consent is in writing and signed by the party claimed to have waived or consented. No consent by either party to, or waiver of, a breach by the other, whether express or implied. shall constitute a consent to. waiver of. or excuse for any different or subsequent breach. Forge Mgieure. Neither party shall be deemed to be in default of any provision ofthis Agreement or liable for failures in performance resulting from acts or events beyond the reasonable cottuol of such party. Such acts shall include but not be limited to acts ot'God. civil or military authority, civil disturbance, war. strikes, ?res, other catastrophes, or other 'force majeure' events beyOnd a party's reasonable control. Survival of Representatio?. The provisions contained in this Agreement that by their sense and context are intended to survive the performance hereof by either or both parties shall so survive the completion of performance and termination of this Agreement. including the making of any and all payments due hereunder. Notices. All notices permitted or required to be given under this Agreement shall be in writing and shall be deemed duly given upon personal delivery (against receipt) or on the fourth day following the date on which each such notice is deposited postage prepaid in the United States Mail. registered or certi?ed, return receipt requested. All notices shall be delivered or sent to the other party at the address(es) shown below or to any other address(es) as the party may designate by ten (10) days prior written notice given in accordance with this provision. if to Smithsonian: Smithsonian Institution Astrophysical Observatory 60 Garden Street Cambridge, MA 02138- [5 l6 Attention: Dr. Willie Soon (for technical matters) Attention: Mr. William J. Ford (for contractual matters) 81-00003? 15. 16. 17. SOUTHERN COMPANY SERVICES, INC. [f to SCS: Southern Company Services, Inc. 600 North 18?? Street Bin i4N-8195 Birmingham, Alabama 35203 Attention: Robert P. Gehri (for technical matters) Attention: Joseph L. Coker (for contractual matters) Publicity. Smithsonian shall not publish and utilize the name or otherwise identify SCS or its af?liate companies in any publications or other advertisements without the express written consent of SCS. As further consideration to SCS. Smithsonian shall provide SCS an advance written copy of preposed publications regarding the deliverables for comment and input, if any, from SCS. Duglicate Originals. Duplicate originals of this Agreement shall be executed, each of which shall be deemed an original but both of which together shall constitute one and the same instrument. Entige Agreement. This Agreement contains the entire agreement of the parties and there are no oral or written representations. understandings or agreements between the parties respecting the subject matter of this Agreement which are not fully expressed herein. Di WITNESS WHEREOF, each of the parties hereto acknowledge that they have caused this Agreement to be executed in duplicate originals by its duly authorized representative on the respective dates entered below. THE SMITHSONIAN INSTITUTION OBSERVATORY gs?: (?Smithsonian?) By: . By: {Sig are) (Signature) Name: Ergo?g?uiil?l? Name: (Typed or printed) (Typed or printed) Title: ?lings: Title: Contrast and grant ?mcigli?t E?Q/Z/zof Date: Sl-000038 ATTACHMENT 6 March 2009 Dr. Robert P. Gehri Southern Company Services, Inc. 600 North 181h Street Bin 14N-8195 Birmingham, AL 35207 Reference: Agreement for SAD Proposal P6882-1-08 Understanding Solar Variability and Climate Change: Signals from Temperature Records of the United States Subject: Year 1 Report Dear Dr. Gehri: Transmitted herewith is one (1) copy of the subject report for the period 15 January 2008 through 14 January 2009, in accordance with the provisions of the above referenced Agreement. Very truly yours, William J. Ford Contract and Grant Specialist Enclosure cc: Mr. Joseph Coker, Southern Co, w/encl. ebc: C. Alcock, w/encl. N. Brickhouse, w/encl. W. Soon, w/encl. N. Rathle. w/encl. P. Sozanski, w/encl. File: Southco-OOI, w/encl. Sl-000051 UNDERSTANDING SOLAR VARIABILITY AND CLIMATE CHANGE: SIGNALS FROM TEMPERATURE RECORDS OF THE UNITED STATES YEAR 1 REPORT For the Period 15 January 2008 to 15 January 2009 Principal Investigator: Dr. Willie Soon January 2009 Prepared for Southern Company Atlanta, GA 30308 I The Smithsonian Astrophysical Observatory is a member of the Harvard-Smithsonian Center for Astrophysics The Southern Company contact for this grant is Robert Gehri, Southern Company, 30 Ivan Allen Jr. Blvd. NW, Atlanta, GA 30308 Sl-000052 Year 1 Report "Understanding Solar Variability and Climate Change: Signals from Temperature Records of the United States" For the Southern Company Period of performance: 1/15/08 to 1/15/09 by Willie Soon, Principal Investigator Smithsonian Astrophysical Observatory Solar, Stellar and Planetary Sciences Division (617-495-7488; The goals of this research proposal have been completely and successfully executed with the following list of deliverables: (1) The publication of: ?Polar bear population forecasts: A public-policy foecasting audit? Interface, vol. 38, 382-405 by Scott Kesten Green and Willie Soon (2008) [with comments and replies] Calls to list polar bears as a threatened species under the United States Endangered Species Act are based on forecasts of substantial long-term declines in their p0pulation. Nine government reports were written to help US Fish and Wildlife Service managers decide whether or not to list polar bears as a threatened species. We assessed these reports based on evidence-based (scienti?c) forecasting principles. None of the reports referred to sources of scienti?c forecasting methodology. Of the nine, Amstrup et al. [Amstrup, S. C., B. G. Marcot, D. C. Douglas. 2007. Forecasting the rangewide status of polar bears at selected times in the let century. Administrative Report, USGS Alaska Science Center, Anchorage, and Hunter et al. [Hunter, C. M., H. Caswell, M. C. Runge, S. C. Amstrup, E. V. Regehr, l. Stirling. 2007. Polar bears in the Southern Beaufort Sea ll: Demography and population growth in relation to sea ice conditions. Administrative Report, USGS Alaska Science Center, Anchorage, were the most relevant to the listing decision, and we devoted our attention to them. Their forecasting procedures depended on a complex set of assumptions, including the erroneous assumption that general circulation models provide valid forecasts of summer sea ice in the regions that polar bears inhabit. Nevertheless, we audited their conditional forecasts of what would happen to the polar bear population assuming, as the authors did, that the extent of summer sea ice would decrease substantially during the coming decades. We found that Amstrup et al. properly applied 15 percent of relevant forecasting principles and Hunter et al. 10 percent. Averaging across the two papers, 46 percent of the principles were clearly contravened and 23 percent were apparently contravened. Consequently, their forecasts are unscienti?c and inconsequential to decision makers. We recommend that researchers apply all relevant principles properly when important public policy decisions depend on their forecasts. 1 Fin al 81-000053 (2) The publication of "Reply to response to et al. (2007) on polar bears and climate change in western Hudson Bay by Stirling et al. (2008)" Ecological Complexity, vol. 5, 289-302 by Dyck, Soon et al. (2008) We address the three main issues raised by Stirling et al. [Stirling, 1., Derocher, A.E., Gough, W.A., Rode, K., in press. Response to et a1. (2007) on polar bears and climate change in western Hudson Bay. Ecol. Complexity]: (1) evidence of the role of climate warming in affecting the western Hudson Bay polar bear population, (2) responses to suggested importance of human? polar bear interactions, and (3) limitations on polar bear adaptation to projected climate change. We assert that our original paper did not provide any ?alternative explanations [that] are largely unsupported by the data? or misrepresent the original claims by Stirling et a1. [Stirling, I., Lunn, N.J., Iacozza, L, 1999. Long-term trends in the p0pulation ecology of polar bears in western Hudson Bay in relation to climate change. Arctic 52, 294?306], Derocher et al. [Derocher, A.E., Lunn. N.J., Stirling, 1., 2004. Polar bears in a warming climate. Integr. Comp. Biol. 44, 163?176], and other peer-approved papers authored by Stirling and colleagues. In sharp contrast, we show that the conclusion of Stirling et al. [Stirling, I., Derocher, A.E., Gough, W.A., Rode, K., in press. Response to et al. (2007) on polar bears and climate change in western Hudson Bay. Ecol. Complexity] suggesting warming temperatures (and other related climatic changes) are the predominant determinant of polar bear population status, not only in western Hudson (WH) Bay but also for populations elsewhere in the Arctic is unsupportable by the current scienti?c evidence. The commentary by Stirling et al. [Stirling, Derocher. All. Gough. W.A.. Rode, K., in press. Response to et al. (2007) on polar bears and climate change in western Hudson Bay. Ecol. Complexity] is an example oi'uni-dimensional, or reductionist thinking, which is not useful when assessing effects ot?climatc change on complex ecosystems. Polar bears of WH are exposed to a multitude of environmental perturbations including human interference and factors unknown seal population size, possible competition with polar bears from other populations) such that isolation of any single variable as the certain root cause climate change in the form of warming spring air temperatures), without recognizing confounding interactions. is imprudent. unjusti?ed and ol?questionablc scienti?c utility. et al. [Dye-k, M.G., Soon, W., Baydack, R.K., Legates, D.R., Baliunas, 3., Ball, Hancock, L.O., 2007. Polar bears of western Hudson Bay and climate change: Are warming spring air temperatures the ?ultimate? survival control factor? Ecol. Complexity, 4, 73?84. 2007.03.002] agree that some polar bear populations may be negatively impacted by future environmental changes; but an oversimplification of the complex ecosystem interactions (of which humans are a part) may not be beneficial in studying external effects on polar bears. Science evolves through questioning and proposing hypotheses that can be critically tested, in the absence of which, as Krebs and Borteaux [Krebs, C.J., Borteaux, D., 2006. Problems and pitfalls in relating climate variability to population dynamics. Clim. Res. 32, 143?149] observe, ?we will be little more than storytellers.? (3) The publication of the scienti?c manuscript "Centennial variations of the global monsoon precipitation in the lastmillennium: Results from ECHO-G model" by ian Liu, Bin Wang, Qinghua Ding, Xueyuan Kuang, Willie Soon and Eduaordo Zorita (2009) in press for the peer-reviewed journal Journal of Climate. We investigate how the global monsoon (GM) precipitation responds to the external and anthrOpogenic forcing in the last millennium by analyzing a pair of control and forced millennium simulations with the ECHO-G coupled ocean?atmOSphere model. The forced run, which includes the solar, volcanic and greenhouse gas forcing, captures the major 2 81-000054 modes of precipitation climatology comparably well when contrasted with those captured by the NCEP reanalysis. The strength of the modeled GM precipitation in the forced run exhibits a signi?cant quasi-bi-centennial oscillation. Over the past 1000 years, the simulated GM precipitation was weak during the Little Ice Age (145 0-1 850) with three weakest periods occurring around 1460, 1685, and 1800, which fell in, respectively, the Sporer Minimum, Maunder Minimum, and Dalton Minimum periods of solar activity. Conversely, strong GM was simulated during the model Medieval Warm Period (ca. 1030?1240). Before the industrial period, the natural variations in the total amount of effective solar radiative forcing reinforce the thermal contrasts both between the ocean and continent and between the northern and southern hemispheres resulting in the millennium-scale variation and the quasi-bi-centennial oscillation in the GM index. The prominent upward trend in the GM precipitation occurring in the last century and the notable strengthening of the global monsoon in the last 30 years (1961?1990) appear unprecedented and owed possibly in part to the increase of atmospheric carbon dioxide concentration though our simulations of the effects from recent warming may be overestimated without considering the negative feedbacks from aerosols. The simulated change of GM in the last 30 years has a spatial pattern that differs from that during the Medieval Warm Period, suggesting that global warming that arises from the increases of greenhouse gases and the input solar forcing may have different effects on the characteristics of GM precipitation. We further note that GM strength has good relational coherence with the temperature difference between the northern and southern hemispheres, and that on centennial timescale, the GM strength responds more directly to the effective solar forcing than the concurrent forced response in global mean surface temperature. (4) The publication of the scienti?c manuscript "Validity of Climate Change Forecasting for Public Policy Decision Making" by Kesten Green, Scott and Willie Soon (2009) in the peer-reviewed journal International Journal of Forecasting [Status: accepted; subject to further revision] Policymakers need to know whether prediction is possible and if so whether any proposed forecasting method will provide forecasts that are substantively more accurate than those from the relevant benchmark method. Inspection of global temperature data suggests that it is subject to irregular cycles on all relevant time scales and that variations during the late-20th Century were not unusual. In such a situation, a ?no change? extrapolation is an apprOpriate benchmark forecasting method. We used the UK. Met Of?ce Hadley Centre?s annual average thermometer data from 1850 through 2007 to examine the performance of the benchmark method. The accuracy of forecasts from the benchmark is such that even perfect forecasts would be unlikely to help policymakers. For example, mean absolute errors for 20- and 50-year horizons were and We nevertheless evaluated the Intergovernmental Panel on Climate Change?s 1992 projected long-term linear warming rate of 0.03?C?per?year. We used the IPCC projection for our demonstration of benchmarking because it has influenced important policy decisions. The small sample of errors from ex ante projections for 1992 through 2008 was practically indistinguishable from the benchmark errors. Validation for long? term forecasting, however, requires a much longer horizon. We illustrate proper 3 Sl-000055 validation procedures by projecting the IPCC warming rate successively over a period analogous to that envisaged in their 21St Century warming scenario in which CO: levels are expected to grow exponentially. Namely 1851 to 1975. The errors from the projections were more than seven times greater than the errors from the benchmark method. Relative errors were larger for longer forecast horizons. Our validation exercise illustrates the importance for policymakers of determining predictability before making expensive decisions. (5) Preparation of the scienti?c manuscript ?Multiple and changing cycles of active stars II. Results? by K. Olah, Z. Kollathl, T. Granzer, K.G. Strassmeier, A.F. Lanza, S. Jarvinen, H. Korhonen, S.L. Baliunas, W. Soon, S. Messina, and G. Cutispoto (2009) for publication in the peer-reviewed journalAstronomy Astrophysics (Status: submitted) ABSTRACT Aims. We study the time variations of the cycles of 20 active stars based on decades-long photometric or spectroscopic observations. Methods. A method of time?frequency analysis, as discussed in a companion paper, is applied to the data. Results. Fifteen stars de?nitely show multiple cycles; the records of the rest are too short to verify a timescale for a second cycle. The cycles typically show systematic changes. In three stars we found 2-2 cycles that are not harmonics, and which vary in parallel, indicating that a common physical mechanism arising from a dynamo construct. The positive relation between the rotational and cycle periods is con?rmed for the inhomogeneous set of active stars. Conclusions. Stellar activity cycles are generally multiple and variable. (6) Preparation of the scientific manuscript ?Solar Arctic-Mediated Climate Variation on Multidecadal to Centennial Timescales: Empirical Evidence, Mechanistic Explanation, and Testable Consequences? (2009) by Willie Soon for publication in the peer-reviewed journal Physical Geography (Status: submitted) The abstract of this new paper says: ?Soon (2005) showed that the variable total solar irradiance (TSI) could explain, rather surprisingly, well over 75% of the variance for the decadally-smoothed Arctic-wide surface air temperature over the past 130 years or so. The present paper provides additional empirical evidence for this physical connection, both through several newly published high-resolution paleo-proxy records and through robust climate-process modeling outputs, and proposes a mechanistic explanation, involving 1) the variable strength of the Atlantic meridional overturning circulation (MOC) or thermohaline circulation (THC), 2) the shift and modulation of the Inter- Tropical Convergence Zone (ITCZ) rainbelt and tropical Atlantic ocean conditions, and 3) the intensity of the wind-driven subtropical and subpolar gyre circulation, across both the North Atlantic and North Paci?c. A unique test of this proposed solar TSI?Arctic thermal-salinity-cryospheric coupling mechanism is the 5-to-20-year delayed effects on 4 81-000056 the peak Atlantic MOC ?ow rate centered near and (2) sea surface temperature (SST) for the tropical Atlantic. The solar Arctic-mediated climate mechanism on multidecadal to centennial timescales presented here can be compared with and differentiated from both the related solar T81 and UV irradiance forcing on decadal timescale. The ultimate goal of this scienti?c research is to gain suf?cient mechanistic details so that the proposed solar-Arctic climate connection on multidecadal to centennial timescales can be con?rmed or falsi?ed. A further incentive is to expand this physical connection to longer, millennial-scale variability as motivated by the multiscale climate interactions shown by Braun et al. (2005), Weng (2005) and Dima and Lohmann (2009).? (7) The prominent participation of PI in the following list of scientific talks and discussion at both national and international forums of professional scientists: All power-point talks are available upon request January 4-6, 2008: Awakening 2008 Conference, Sea Island, Georgia "The secondary role of C02 radiative forcing in climate change: Real facts you are not even supposed to ?nd out!" March 2?4, 2008: International Climate Conference, New York City, NY "Global Warming 101: Al Gore?s C02 Theory" March 15, 2008: Good Neighbor Forum, Cheyenne, WY "Global Warming Explained!" (co-panelist Lyle Laverty, Assistant Secretary of Fish, Wildlife and Park Services) March 31, 2008: Deliberative Polling Event at California University of California, PA "Global Warming Explained: The importance of getting the science right!? April 3, 2008: Department of Physics Colloquium, University of Buffalo, NY "The secondary role of C02 and CH4 forcing on climate change: Past, present and future" April 24, 2008: Sutherland Institute Global Warming Panel (with Roy Spencer as co-panelist) "Future of Utah", Salt Lake City, Utah June 19-22, 2008: Annual "Winning Ideas Weekend" of the Free to Choose Network, New York City, NY "The Sun, C02 and Global Warming" (with Dave Legates as co-panelist) (among other speakers: John Fund of and John Stossel of ABC News) June 23?28, 2008: Nice France Special session for the ISF. Session Title: "Climate Forecasting and Public Policy.? 5 SI-000057 "Do the Forecasts by the US. Government Provide Valid Evidence for the Decision to Classify Polar Bears as an Endangered Species?" J. Scott The Wharton School, U. of Philadelphia, PA, Kesten. C. Green, Business and Economic Forecasting, Monash University, Vic 3800, Australia and Willie Soon, Harvard-Smithsonian Center for Astrophysics, Cambridge MA July 11-13, 2008: Annual Meeting of Doctors for Disaster Preparedness, Phoenix, Arizona "Endangering the Polar Bears: How environmentalists kill" August 6-14, 2008: the 33rd International Geological Congress, Oslo, Norway co-chairing, with Professor Bob Carter of James Cook University, the science session CGC-03: ?Solar drivers of climate change and the stratigraphic record? (ii) selected by Professor David Gee of Uppsala University, the IGC SciCom Chairman, to be one of the speakers for the August 8's Theme of the Day of IGC on "Climate" and the title of my talk: ?Solar and Climate Variability: Past, present and future? invited speaker for CGC-03 session: "Solar irradiance variability and climatic responses: A brief review" (iv) contributing author for session: ?Relationship between the global monsoon intensity and the effective solar radiation in the last millennium? by Jian Liu, Bin Wang and Willie Soon September 15, 2008: Marshall Institute Climate Discussion Group, "The Sun-Climate Connection" (I) September 23, 2008: University of Southern California, Ayn Rand Institute Global Warming and Policy Panel (with Keith Lockitch as co-panelist), "On the science of global climate change" September 25, 2008: University of California Berkeley, Ayn Rand Institute Global Warming and Policy Panel (with Keith Lockitch as co-panelist), ?On the science of global climate change" September 29, 2008: Columbus, Ohio, Annual Meeting of the Managers? Association, "On the science of global climate change" (0) November 24-26, 2008: Jakarta, Indonesia, Invited speaker at the International Symposium on Climate and Weather of the Sun-Earth System hosted by Indonesia?s National Agency for Meteorology Geophysics (as part of the scoping processes for the upcoming UN IPCC ARS reports). 6 Sl-000058 ATTACHMENT hirerfaces Vol. 38, No. 5, September-October 2008, pp. 382-405 EISSN 1526-551X 08 3805 0382 Inlm D01 10.1287/ inte.1080.0383 ?2008 INFORMS Polar Bear Population Forecasts: A Public?Policy Forecasting Audit 1. Scott The Wharton School, University of Philadelphia, 19104, Kesten C. Green Business and Economic Forecasting, Monash University, Victoria 3800, Australia, kesten@kestencgreen.com Willie Soon Harvard-Sn?thsonian Center for Astrophysics, Cambridge, Massachusetts 02138, wsoon@cfa.harvard.edu Calls to list polar bears as a threatened species under the United States Endangered Species Act are based on forecasts of substantial long-term declines in their population. Nine government reports were written to help US Fish and Wildlife Service managers decide whether or not to list polar bears as a threatened species. We assessed these reports based on evidence?based (scienti?c) forecasting principles. None of the reports referred to sources of scienti?c forecasting methodology. Of the nine, Amsth et al. [Amstrup, S. C., B. G. Marcot, D. C. Douglas. 200?. Forecasting the rangewide status of polar bears at selected times in the 2'lst centtuy. Administra? tive Report, USES Alaska Science Center, Anchorage, and Hunter et al. [I-Iunter, C. M., H. Caswell, M. C. Rouge, S. C. Amstrup, E. V. Regehr, I. Stirling. 200?. Polar bears in the Southern Beaufort Sea 11: Demography and population growth in relation to sea ice conditions. Administrative Report, USGS Alaska Science Center, Anchorage, were the most relevant to the listing decision, and we devoted our attention to them. Their forecasting procedures depended on a complex set of assumptions, including the erroneous assumption that general circulation models provide valid forecasts of summer sea ice in the regions that polar bears inhabit. Nev- ertheless, we audited their conditional forecasts of what would happen to the polar bear population assuming, as the authors did, that the extent of summer sea ice would decrease substantially during the coming decades. We found that Ame-trap et al. properly applied 15 percent of relevant forecasting principles and Hunter et al. 10 percent. Averaging across the two papers, :16 percent of the principles were clearly contravened and 23 per- cent Were apparently contravened. Consequently, their forecasts are and inconsequential to decision makers. We recommend that researchers apply all relevant principles properly when important public-policy decisions depend on their forecasts. Key words: adaptation; bias; climate diange: decision making; endangered species; expert opinion; extinction; evaluation,- evidence?based principlea; expert judgment; forecasting methods; global warming; habitat loss; mathematical models; scienti?c method; sea ice. History: This paper was refereed. Despite Widespread agreement that the polar bear population increased during recent years follow- ing the imposition of stricter hunting rules (Prestrud and Stirling 1994), new concerns have been expressed that climate change will threaten the survival of some subpopulations in the let century. Such concerns led the US Fish and Wildlife Service to consider listing polar bears as a threatened species under the United States Endangered Species Act. To list a species that is currently in good health must surely require valid 382 forecasts that its population would, if it were not listed, decline to levels that threaten the viability of the species. The decision to list polar bears thus rests on long-term forecasts. The US Geological Survey commissioned nine administrative reports to satisfy the request of the Secretary of the Interior and the Fish and Wildlife Service to conduct analyses. Our objective was to determine if the forecasts were derived from accepted scientific procedures. We first examined the references et al.: Polar Bear Population Forecasts: A Public?Policy Forecasting Audit Interfaces 38(5), pp. 382?105, @2008 INFORMS 383 in the nine government reports. We then assessed the forecasting procedures described in two of the reports relative to forecasting principles. The forecast? ing principles that we used are derived from evidence obtained from scienti?c research that has shown the methods that provide the most accurate forecasts for a given situation and the methods to avoid. Scienti?c Forecasting Procedures Scientists have studied forecasting since the 19303; (1978, 1985) provide summaries of impor- tant ?ndings from the extensive forecasting literature. In the mid-19903, Scott established the Forecasting Principles Project to summarize all use- ful knowledge about forecasting. The evidence was codi?ed as principles, or condition?action statements, to provide guidance on which methods to use under different circumstances. The project led to the Princi? ples of Forecasting handbook 2001}. Forty internationally recognized forecasting?method experls formulated the principles and 123 reviewed them. We refer to the evidence-based methods as scientific fore- casting procedures. The strongest evidence is derived from empirical studies that compare the perfonnance of alternative methods; the weakest is based on received wisdom about proper procedures. Ideally, performance is assessed by the ability of the Selected method to pro- vide useful ex ante forecasts. However, some of the principles seem selfwevident "provide complete, simple, and clear explanations of methods") and, as long as they were unchallenged by the available evi- dence, were included in the principles list. The principles were derived from many fields, in? cluding demography, economics, engineering, ?nance, management, medicine, politics, and weather; this ensured that they encapsulated all rele vant evidence and would apply to all types of forecast- ing problems. Some reviewers of our research have suggested that the principles do not apply to the physical sciences. When we asked them for evidence to support that assertion, we did not receive useful responses. Readers can examine the principles and form their own judgments on this issue. For example, does the principle, "Ensure that information is reli- able and that measurement error is low," not apply when forecasting polar bear numbers? The forecasting principles are available at W. forecastingprinciples.com, a website that the Interna- tional Institute of Forecasters sponsors- The directors of the site claim that it provides ?all useful knowl- edge about forecasting? and invite visitors to submit any missing evidence. The website also provides fore- casting audit software that includes a summary of the principles (which currently number 14.1.0) and the strength of evidence for each principle; met-s posted on the website provide details. General Assessment of Long-Term Polar Bear Population Forecasts We examined all references cited in the nine US Geo? logical Survey Administrative Reports posted on the Internet. The reports, which included 444 unique ref- erences, were Amstrup et al. (2007), Bergen et al. (2007), DeWeaver (2007), Durner et al. (2007), Hunter et al. (2007), Obbard et al. (2007), Regehr et al. (2007), Rode et al. (2007), and Stirling et al. (2007). We were unable to find references to evidence that the fore- casting methods described in the reports had been validated. Forecasting Audit of Key Reports Prepared to Support the Listing of Polar Bears We audited the forecasting procedures in the reports that we judged provided the strongest support forecasts) for listing polar bears. We selected Amstrup et al. (2007), which we will refer to as because the press had discussed their forecast widely. We selected Hunter et al. (2007), which we will refer to as H6, because the authors used a substantially different approach to the one reported in The reports provide forecasts of polar-bear popu- lations for 45, 75, and 100 years from the year 2000 and make recommendations with respect to the polar- bear?listing decision. However, their recommenda- tions do not follow logically from their research because they only make forecasts of the polar bear population. To make policy recommendations based on forecasts, the following assumptions are necessary: (1) Global warming will occur and will reduce the amount of summer sea ice; et Polar Bear Population Forecasts: A Public-apolle Forecasting Audit 384 Interfaces 38(5), pp. 382?405, @2003 (2) Polar bears will not adapt; thus, they will obtain less food than they do now by hunting from the sea- ice platform; (3) Listing polar bears as a threatened or endan- gered species will result in policies that will solve the problem without serious detrimental effects; and Other policies would be inferior to those that depend on an Endangered Species Act listing. Regarding the first assumption, both AMD and H6 assumed that general circulation models (GCMs) pro? vide scientifically valid forecasts of global tempera- ture and the extent and thickness of sea ice. AMD stated: "Our future forecasts are based largely on information derived from generalcirculation model projections of the extent and spatiotemporal distribution of sea ice? (AMI): p. p. 83, Figure 2). H6 stated, ?We extracted forecasts of the availability of sea ice for polar bears in the Southern Beaufort Sea region, using forecasts of sea?ice concentra- tions from IPCC Fourth Assessment Report (AIM) fully?coupled general circulation models" 11). (Note: IPCC is the intergovernmental Panel on Cli? mate Change.) "that is, the forecasts of both AMD and H6 are conditional on longaterm global warming lead- ing to a dramatic reduction in Arctic sea ice during melt-back periods in spring, late summer, and fall. Green and (200?) examined long-term climate-forecasting efforts and were unable to find a single forecast of global warming that was based on Scientific methods. When they audited the GCM climate modelers? protedures, they found that only 13 percent of the relevant forecasting principles were followed properly; some contraventions of princi? ples were critical. Their findings were consistent widi earlier cautions. For example, Soon et al. [2001) found that the current generation of GCMs is unable to meaningfully calculate the effects that additional atmospheric carbon dioxide has on the climate. This is because of the uncertainty about the past and present climate and ignorance about relevant Weather and cli- mate processes. Sorne climate modelers state that the GCMs do not provide forecasts. According to one of the lead authors of the ABA (Trenberth 2007), are no predictions by at all. And there never have been. The instead proffers "what if" projections of future Climate that correspond to certain emissions scenarios. There are a number of assumptions that go into these emissions scenarios. They are intended to cover a range of possible self con? sistent "story lines? that then provide decision makers with information about Whid'l paths might be more desirable. AMD and H6 provided no scientific evidence to support their assumptions about any of the four issues that we identified above. Thus, their forecasts are of no value to decision makers. Nevertheless, we audited their polar-bear?population forecasting pro- cedures to assess if they would have produced valid forecasts If the Lmderlying assumptions had been valid. In conducting our audits, we read AMD and and independently rated the forecasting procedures described in the reports by using the forecasting audit software mentioned above. The rating scale ranged from ?2 to the former indicated that the pro- cedures contravene the principle; the latter signified that it is properly applied. Following the initial round of ratings, we examined differences in our ratings to reach consensus. When we had dif?culty in reach? ing consensus, we moved ratings toward Princi- ple 1.3 (Make sure forecasts are independent of politics) is an example of a principle that was contravened in both reports (indeed, in all nine). By politics, we mean any type of organizational bias or pressure. It is not unusual for different stakeholders to prefer par- ticular forecasts; however, if forecasters are in?uenced by such considerations, forecast accuracy could suf? fer. The header on the title page of each of the nine reports suggests how the authors interpreted their task: Science Strategy to Support US Fish and Wildlife Service Polar Bear Listing Decision. A more neutral statement of purpose might have read ?Fore- casts of the polar bear population under alternative policy regimes.? While it was easy to code the two reports? pro- cedures against Principle 1.3, the ratings were sub- jective for many principles. Despite the subjectivity, our ratings after the ?rst rormd of analyses for each report were substantially in agreement. Furthermore, we readily achieved consensus by the third round. The two reports did not provide suf?cient detail to allow us to rate some of the relevant principles. As a result, we contacted the report authors for addi- tional information. We also asked them to review the et al.: Polar Bear Population Forecasts: A Public-Policy Forecasting Audit Interfaces 38(5), pp. 382?405, @2003 INFORMS 385 ratings that we had made and to provide comments. In their replies, the report authors refused to provide any responses to our requests. (See #2 in the Author Comments section at the end of this paper.) In December 2007, we sent a draft of this article to all authors whose works we cited substantively and asked them to inform us, if we had misinterpreted their ?ndings. None objected to our interpretations. We also invited each author to review our paper but received no reviews from our requests. Audit Findings for AMD In auditing Alva?s forecasting procedures, We first agreed that 24 of the 140 forecasting principles were irrelevant to the forecasting problem they were try- ing to address. We then examined principles for which our ratings differed. The process involved three rounds of consultation; after two rounds, we were able to reach consensus on ratings against all 116 relevant principles. We were unable to rate AMD's procedures against 26 relevant principles (Table A.3) because the paper lacked the necessary information. Tables A.1, A.2, A3, and A4 provide full disclosure of our AMD ratings. Overall, we found that AMD definitely contravened 41 principles and apparently contravened an addi- tional 32 principles. The authors provided no justi- fications for the contraventions. Of the 116 relevant principles, we could ?nd evidence that AMD properly applied only 17 (14.7 percent) (Table A4). In the remainder of this section, we will describe some of the more serious problems with the AMD forecasting procedures by listing a selected principle and then explaining how AMD addressed it. Principle 6.7: Match the forecasting method(s) to the situation. The AMD forecasts rely on the opinions of a sin? gle polar bear expert. The report authors transformed these opinions into a complex set of formulae without using evidence-based forecasting principles. In effect, the formulae were no more than a codification of the expert?s unaided judgments, which are not appropri? ate for forecasting in this situation. One of the most counterintuitive ?ndings in fore- casting is that judgmental forecasts by experts who ignore accepted forecasting principles have little value in complex and uncertain situations 1978, pp. 91?96; Tetlock 2005). This ?nding applies whether the opinions are expressed in words, spreadsheets, or mathematical models. In relation to the latter, Pilkey and Pilkey?Iarvis (2007) provide examples of the fail- ure of domain experts' mathematical models when they are applied to diverse natural science problems including ?sh stocks, beach engineering, and invasive plants. This finding also applies regardless of the amount and quality of information that the experts use because of the following: (1) Complexity: People cannot assess complex rela- tionships through unaided observations. (2) Coincidence: People confuse correlation with causation. (3) Feedback: People making judgmental predic? tions typically do not receive unambiguous feedback that they can use to improve their forecasting. (4) Bias: People have dif?culty in obtaining or using evidence that contradicts their initial beliefs. This problem is especially serious among people who view themselves as experts. Despite the lack of validity of expert unaided fore? casts, many public-policy decisions are based on such forecasts. Research on persuasion has shown that peo- ple have substantial faith in the value of such fore- casts and that faith increases when experts agree with one another. Although they may seem convincing at the time, expert forecasts can, a few years later, serve as important cautionary tales. Cerf and Navasky?s (1998) book contains 310 pages of examples of false expert forecasts, such as the Fermi award-winning .scientist John von Neumann?s 1956 prediction that few decades hence, energy may be free." Exam- ples of expert climate forecasts that turned out to be wrong are easy to ?nd, such as UC Davis ecolo- gist Kenneth Watt?s prediction during an Earth Day speech at Swarthmore College (April 22, 1970) that ?If present trends continue, the world will be about four degrees colder in 1990, but eleven degrees colder in the year 2000. This is about twice what it would take to put us into an ice age.? Tetlock (2005) recruited 284 people whose profes- sions included "commenting or offering advice on political and economic trends." He picked topics (geo? graphic and substantive) both within and outside of their areas of expertise and asked them to forecast the et al.: Polar Bear Population Forecasts: A Public-Policy Forecasting Audit 386 lnlerfaces 38(5), pp. 382?405, @2008 INFORMS probability that various situations would or would not occur. By 2003, he had accumulated more than 82,000 forecasts. The experts barely, if at all, outper- formed nonexperts; neither group did well against simple rules. Despite the evidence showing that expert forecasts are of no value in complex and uncertain situations, people continue to believe in experts? forecasts. The first author?s review of empirical research on this problem led him to develop the ?seer-sucker theory,? which states that "No matter how .much evidence exists that seers do not exist, seers will find suckers? 1980). Principle 7.3: Be conservative in situations of high uncertainty or instability. Forecasts should be conservative when a situation is unstable, complex, or uncertain. Being conservative means moving forecasts towards "no change? or, in cases that exhibit a well-established, long?term trend and where there is no reason to expect the trend to change, being conservative means moving forecasts toward the trend line. A long-term trend is one that has been evident over a period that is much longer than the period being forecast. Conservatism is a fun- damental principle in forecasting. The interaction between polar bears and their envi- ronment in the Arctic is complex and uncertain. For example, AMD associated warmer temperatures with lower polar bear survival rates; yet, as the follow- ing quote illustrates, colder temperatures have also been found to be associated with the same outcome: "Abnormally heavy ice covered much of the eastern Beaufort Sea during the Winter of 1973?1974. This resulted in major declines in numbers and productiv- ity of polar bears?aFdTi?ngEeals in 1975" et al. 1936, p. 249). Stirling (2002, pp. 68, 72) further expanded on the complexity of polar bear and sea-ice interactions: In the eastern Beaufort Sea, in years during and fol? lowing heavy ice conditions in spring, we found a marked reduction in production of ringed seal pups and consequently in the natality of polar The effect appeared to last for about three years, after which productivity of both seals and bears increased again. These clear and major reductions in produc- tivity of ringed seals in relation to ice conditions occurred at decadal?scale intervals in the mid~1970s and on the basis of less complete data, probably in the mid?19605 as . Recent analy- ses of ice anomalies in the Beaufort Sea have now also confirmed the existence of an approximately 10-year cycle in the is roughly in phase with a similar decadal-scale oscillation in the runoff from the Mackenzie . However, or whether, these regional-scale changes in ecological conditions have affected the reproduction and survival of young ringed seals and polar bears through the 19905 is not clear. Regional variability adds to uncertainty. For exam- ple, Antarctic ice mass has been increasing while sea and air temperatures have also been increas- ing (Zhang 2007). At the same time, depth?averaged oceanic temperatures arormd the Southeastern Bering Sea (Richter-Menge et a1. 2007) have been cooling since 2006. Despite the warming of local air tem- peratures by 1.6 d: 0.6 c?C, there was no consistent mid-September (the period of minimal ice extent) ice decline in the Canadian Beaufort Sea over the con- tinental shelf, which had been ice-covered for the 36 years between 1968 and 2003 (Melling et al. 2005). In their abstract, AMD predicted a loss of . . 2/3 of the world?s current polar bear population by mid? century." The 2/3 figure is at odds with the output from the authors' "deterministic model? as they show in Table 6 in their report. The model?s "ensemble mean? prediction is for a more modest decline of 17 percent in the polar bear population by the year 2050. Even the GCM minimum ice scenario, which the authors used as an extreme input, provides a fore- cast decline of 22 percent?-much less than the 2/3 figure they state in their abstract. We believe that the authors derived their 2/3 ?gure informally from the outputs oftheiLBayesian network-modeling exercise: 7 The Bayesian network output of interest is in the form of probabilities (expressed as percentages) for each of ?ve possible population states: "larger," "same as now," "smaller," "rare," and "extinct" (AMD, Table 8, pp. 66?67). There is, however, no clear link between the sets of probabilities for each population state for each of the authors? four Arctic ecu-regions and the dramatic 2/3 population-reduction ?gure. AMD made predictions based on assumptions that we View as questionable. They used little historical data and extreme forecasts rather than conservative ones. et Polar Bear Population Forecasts: A Public-Policy Forecasting Audit interfaces 38(5), pp. 382?105, @2008 INFORMS 387 Principle 8.5: Obtain forecasts from heterogeneous experts. polar bear population forecasts were the product of a single expert. Experts vary in their knowl- edge and in how they approach problems. A will? ingness to bring additional information and different approaches to bear on a forecasting problem improves accuracy. When researchers use information from a single source only, the validity and reliability of the forecasting process is suspect. In addition, in situa- tions in which experts might be biased, it is important to obtain forecasts from experts with different biases. Failing to follow this principle increases the risk that the forecasts obtained will be extreme when, in this situation, forecasts should be conservative (see Prin- ciple 7.3 above). Principle 10.2: Use all important variables. et a1. (2007) noted that scenarios of polar bear population decline from changing sea-ice habi- tat alone grossly oversimplify the complex ecologi- cal relationships of the situation. In particular, AMD did not adequately consider the adaptability of polar bears. They mentioned that polar bears evolved from brown bears 250,000 years ago; however, they appear to have underrated the fact that polar bears probably experienced much warmer conditions in the Arctic over that extended period, including periods in which the sea-ice habitat was less than the amoruit pre- dicted during the 21st century by the GCM projec? tions that AMD used. A dramatic reduction of sea ice in both the northwest Alaskan coast and north- west Greenland part of the Arctic Ocean during the very warm interglacial of marine isotope stage 5e ca. 130,000 to 120,000 years ago was documented by Hamilton and Brigham-Grette (1991), Brigham-Grette and Hopkins (1995), and Norgaard-Pedersen et al. (2007). Brigham-Grette and Hopkins (1995, 159) noted that the "winter sea-ice limit was north of Bering Strait, at least 800 km north of its present posi- tion, and the Bering Sea was perennially ice?free? and that "[the more saline] Atlantic water may have been present on the shallow Beaufort Shelf, suggesting that the Arctic Ocean was not stratified and the Arctic sea- ice cover was not perennial for some period.? The nature and extent of polar bear adaptability seem cru? cial to any forecasts that assume dramatic changes in the bears? environment. Audit Findings for H6 H6 forecast polar bear numbers and their survival probabilities in the Southern Beaufort Sea for the let century. Of the 140 forecasting principles, we agreed that 35 were irrelevant to the forecasting problem. We found that H6?s procedures clearly contravened 61 princi- ples (Table A5) and probably contravened an addi? tional 19 principles (Table A6). We were unable to rate H6?s procedures against 15 relevant principles (Table A7) because of a lack of information. Per- haps the best way to summarize H6?s efforts is to say that the authors properly applied only 10 (9.5 percent) of the 105 relevant principles (Table A.8). Many of the contraventions in H6 were similar to those in AMD. We describe some of the more seri? ous problems with the H6 forecasting procedures by examining their contraventions of 13 important prin- ciples that differed from contraventions discussed in AMD. Principles 1.1?1.3: Decisions, actions, and biases. The H6 authors did not describe alternative deci- sions that might be taken (as Principle 1.1 requires), nor did they propose relationships between possi? ble forecasts and alternative decisions (as Principle 1.2 requires). For example, what decision would be implied by a forecast that predicts that bear numbers will increase to where they become a threat to existing human settlements? Principle 4.2: Ensure that information is reliable and that measurement error is low. H6 relied heavily on five years of data with unknown measurement errors. Furthermore, we ques- tion whether the capture data on which they relied provide representative samples of bears in the South- ern Beaufort Sea given the vast area involved and difficulties in spotting and capturing the bears. Bears wander over long distances and do not respect administrative boundaries (Amstrup et a1. 2004). The validity of the data was also compromised because H6 imposed a speculative demographic model on the raw capture-recapture data (Amstrup et a1. 2001, Regehr et a1. 2006). Principle 4.4: Obtain all important data. et al.: Polar Bear Population Forecasts: A Public-Policy Forecasting Audi! 388 Interfaces 33(5). pp. 332-405, @2003 INFORMS H6 estimated their key relationship?between ice? free days and the polar bear population?by using data that appear to be unreliable primarily because of the difficulty of estimating the polar bear population, but also because of the measurements of ice. Experts in this field, including the authors of the nine reports, are aware of these problems. In addition, they rely on only ?ve years of data with a limited range of climate and ecology combinations. They might, for example, have independently estimated the magnitude of the relationship by obtaining estimates of polar bear pop- ulations during much warmer and much colder peri- ods in the past. The supplementary informatiOn in Regehr et al. (2007, Figure 3) shows that 1987, 1993, and 1998 were exceptional seasons with more than 150 ice-free days substantially above the 135 ice? free days documented for 2004?2005) in the Southern Beaufort Sea. Yet, there were no apparent negative impacts on the polar bear population and well-being (Amstrup et a1. 2001). Because they used only ?ve observations, the above points are moot. It is impossible to estimate a causal relationship in a complex and uncertain situation by using only ?ve data points. Principle 7.3: Be conservative in situations of high uncertainty or instability. The situation regarding polar bears in the South- ern Beaufort Sea is complex and uncertain. On the basis of ?ve years of data, H6 associated warmer temperatures (and hence more ice-free days) with lower polar bear survival rates. Yet, as we noted in relation to AMD, cold temperatures have also been found to be associated with the same outcome. In addition, regional variability sea ice increases while sea and air temperatures increase) adds to uncertainty. There is general agreement that polar bear pop ulations have increased or remained stable in the Alaska regions in recent decades (Amstrup et a1. 1995, Angliss and Outlaw 2007). H6 assumed that there are downward forces that will cause the trend to reverse. However, studies in economics have shown little suc- cess in predicting turning points. Indeed, and Collopy (1993) proposed the principle that one should not extrapolate trends if they are contrary to the direction of the causal forces as judged by domain experts. They tested the principle on four data sets involving 723 long-range forecasts and found that it reduced forecast error by 43 percent. Therefore, even if one had good reason to expect a trend to reverse, being conservative and avoiding the extrap- olation of [my trend will increase the accuracy of forecasts. Principle 9.2: Match the model to the underlying phenomena. Because of the poor spatial resolution of the GCMs, it is important that readers know the meaning of the "Southern Beaufort Sea? (SB) in the H6 report. H6 states: Because GCMs do not provide suitable forecasts for areas as small as the SB, we used sea ice concen- tration for a larger area composed of 5 IUCN (Inter- national Union for Conservation of Nature) polar bear management units (Aars et al. 2006) with ice dynamics similar to the SB management unit (Barents Sea, Beaufort Sea, Chukchi Sea, Kara Sea, and Laptev Sea; see Rigor and Wallace 2004, Durner et al. 2007). We assumed that the general trend in sea ice availabil- ity in these 5 units was representative of the general trend in the Southern Beaufort region. 12). Given the unique ecological, geographical, meteo- rological, and climatological conditions in each of the ?ve circumpolar seas, this assumption by H6 is not valid or convincing. Principle 9.5: Update frequently. When they estimated their model, H6 did not include data for 2006, the most recent year that was then available. From the supplementary information that Regehr et al. (2007, Figure 3) provide, one finds that the number of ice-free days for the 2006 sea- son was approximately 105??close to the mean of the "good" ice years. Principle 10.2: Use all important variables. When using causal models, it is important to incor- porate policy variables if they might vary or if the ptupose is to decide which policy to implement. H6 did not include policy variables, such as seasonal pro- tection of bears? critical habitat or changes to hunting rules. Other variables, such as migration, snow, and wind conditions, should also be included. For example, Holloway and Sou (2002), Ogi and Wallace (2007), et al.: Polar Bear Population Forecasts: A Public?Policy Forecasting Audit Interfaces 33(5), pp. 382405, @2003 INFORMS 389 and Nghiem et a1. (2007) suggested that large-scale atmospheric winds and related circulatory and warm? ing and cooling patterns play an important role in causing?in some situations with substantial time delays?both the decline in extent and thinning of Arctic sea ice. The GCM forecasts of sea ice did not correctly include those effects; hence, the forecasts of the quality of the polar bear habitats also did not. In addition, as et al. (2007) noted, forecasts of polar bear decline because of dramatic changes in their environment do not take proper account of the extent and type of polar bear adaptability. Principle 10.5: Use different types of data to mea- sure a relationship. This principle is important when there is uncer- tainty about the relationships between causal variables (such as ice extent) and the variable being forecast (polar bear population), and when large changes are expected in the causal variables. In the case of the latter condition, H6 accepted the GCM model predictions of large declines in summer ice throughout the let century. Principle 10.7: Forecast for alternate interventions. H6 did not explicitly forecast the effects of differ? ent policies. For example, if the polar bear population came under stress because of inadequate summer food, what would be the costs and benefits of pro- tecting areas by prohibiting marine and land-based activities, such as tourism, capture for research, and hunting at critical times? In addition, what would be the costs and benefits of a smaller but stable popu- lation of polar bears in some polar subregions? And how would the net costs of such alternative policies compare with the net costs of listing polar bears? Principle 13.8: Provide easy access to the data. The authors of the reports that we audited did not include all of the data they used in their reports. We requested the missing data, but they did not provide it. Principle 14.7: When assessing prediction intervals, list possible outcomes, and assess their likelihoods. To assess meaningful prediction intervals, it is help? ful to think of diverse possible outcomes. The H6 authors did not appear to consider, for example, the possibility that polar bears might adapt to terrestrial life over summer months by ?nding alternative food sources (Stempniewicz 2006, and Romberg 2007) or by successfully congregating in smaller or local- ized ice-hunting areas. Consideration of these and other possible adaptations and outcomes would have likely led the H6 authors to be less con?dent provide wider prediction intervals) about the out? come for the bear population. Extending this exer- cise to the forecasts of climate and summer ice extent would have further widened the range of possible outcomes. Discussion Rather than relying on untested procedures to forecast polarbear populations, the most appropriate approach would be to rely upon prior evidence of which fore? casting methods work best under which conditions. Thus, one could turn to empirical evidence drawn from a wide variety of forecasting problems. This evi- dence is summarized in the Forecasting Method Selec- tion Tree at (1985) provided an early review of the evidence on how to forecast given high uncertainty. Schnaars (1984) and Schnaars and Bavuso (1986) con- cluded that the random walk was typically the most accurate model in their comparative studies of hun- dreds of economic series with forecast horizons of up to ?ve years. This principle has a long history. For example, regression models ?regress? towards a no- change forecast when the estimates of causal relation- ships are imcertain. Because of the enormous uncertainty involved in long-term forecasts of polar bear populations, the lack of accurate time-series data on these populations, and the complex relationships that are subject to much uncertainty, prior evidence from forecasting research calls for simple and conservative methods. Therefore, one should follow a trend if such a trend is consistent and if there are no strong reasons to expect a change in the trend. Even then, however, it is wise to dampen the trend towards zero given the increasing uncer- tainty as the forecast horizon is extended. Empirical evidence supports this notion of ?dampng trends? 2001). Lacking a trend, forecasters should et al.: Polar Bear Population Forecasts: A Public?Policy Forecasting Audit 390 Interfaces 38(5), pp. 382?405, @2008 INFORMS turn to the so-called "random walk" or nochange model. Given the upward trend in polar bear num- bers over the past few decades, a modest upward trend is likely to continue in the near future because the apparent cause of the trend (hunting restrictions] remains. However, the inconsistent long-term trends in the polar bear population suggest that it is best to assume no trend in the long?term. Summary We inspected nine administrative reports that the US government commissioned. Because the current polar bear population is not at a level that is causing con? cern, the case for listing depends upon forecasts of serious declines in bear numbers in future decades. None of these reports included references to scienti?c works on forecasting methods. We found that the two reports that we judged most relevant to the listing decision made assump- tions rather than forecasts. Even if these assumptions had been valid, the bear population forecasting pro- cedures described in the reports contravened many important forecasting principles. We did forecasting audits of the two key reports (Table Decision makers and the public should require sci? entific forecasts of both me polar bear population and the costs and bene?ts of alternative policies before making a decision on whether to list polar bears as threatened or endangered. We recommend that important forecasting efforts such as this should prop- erly apply all relevant principles and that their proce? be audited to ensure that they do so. Failure to apply any principle should be supported by evidence that the principle was not applicable. Principles AMD H6 Contravened 41 51 Apparently contravened 32 19 Not auditable 26 15 Properly applied 17 10 Totals TE Table 1: We summarize our forecasting audit ratings at the AMD and H6 reports against relevant inrecasling principles. Author Comments 1. Our interest in the topic of this paper was piqued when the State of Alaska hired us as consultants in late September 2007 to assess forecasts that had been prepared "to Support US Fish and Wildlife Service Polar Bear Listing Decision.? We received $9,998 as payment for our consulting. We were impressed by the importance of the issue; therefore, after providing our assessment, we decided to continue work on it and to prepare a paper for publication. These latter efforts have not been funded. We take responsibility for all judgments an?d for any errors that we might have made. 2. On November 27, 2007, we sent a draft of our paper to the authors of the US Geological Survey administrative reports that we audited; it stated: As we note in our paper, there are elements of sub- jectivity in making the audit ratings. Should you feel that any of our ratings were incorrect, we would be grateful if you would provide us with evidence that would lead to a different assessment. The same goes for any principle that you think does not apply, or to any principles that we might have overlooked. There are some areas that we (:0de not rate due to a lack of information. Should you have information on those topics, we Would be interested. Finally, We would be interested in peer review that you or your colleagues could provide, and in suggestions on how to improve the accuracy and clarity of our paper. We received this reply from Steven C. Amstrup on November 30, 2007: "We all decline to offer preview comments on your attached manuscript. Please feel free, however, to list any of us as potential referees when you submit your manuscript for publication.? 3. We invite others to conduct forecasting audits of et al., Hunter et al., or any of the other papers prepared to support the endangered?species listing, or any other papers relevant to long-term fore- casting of the polar bear population. Note that the audit process calls for two or more raters. The audits can be submitted for publication on pubicpolicyfore? castingcom with the auditors? bios and any informa- tion relevant, potential sources of bias. et al.: Polar Bear Population Forecasts: A Public?Policy Forecasting Audit Interfaces 33(5), pp. 332-405, @2003 moms 391 Appendix: Full Disclosure of the Codings Table A.1: Principles contravened in Amslrup et al. (AMD). Setting objectives 1.2 Prior to forecasting, agree on actions to take assuming different possible forecasts. 1.3 Make sure forecasts are independent of politics. 1.4 Consider whetherthe events or series can be forecasted. 1.5 Obtain decision makers' agreement on methods. Identifying data sources 3.5 Obtain information from similar (analogous) series or cases. Such information may help to estimate trends. Collecting data 4.2 Ensure that information is reliable and that measurement error is low. Selecting methods 6.1 List all the important selection criteria before evaluating methods. 6.2 Ask unbiased experts to rate potential methods. 6.7 Match the forecasting methodis) to the situation. 6.8 Compare track records of various forecasting methods. 6.10 Examine the value of alternative forecasting methods. Implementing methods: General 7.3 Be conservative in situations of high uncertainty or instability. implementing judgmental methods 8.1 Pretest the questions you intend to use to elicit judgmental forecasts. 8.2 Frame questions in alternative ways. 8.5 Obtain forecasts from heterogeneous experts. Obtain forecasts from enough respondents. 8.8 Obtain multiple forecasts of an event from each expert. Implementing quantitative methods 9.1 Tailor the forecasting model to the horizon. 9.3 Do not use "fit" to develop the model. 9.5 Update models frequently. implementing methods: Quantitative models with explanatory variables 10.6 Prepare forecasts for at least two alternative environments. 10.8 Apply the same principles to forecasts of explanatory variables. 10.9 Shrink the forecasts of change if there is high uncertainty for predictions of the explanatory variables. Combining forecasts 12.1 Combine fOrecasts from approaches that differ. 12.2 Use many approaches (or forecasters), preferably at least five. 12.3 Use formal procedures to combine forecasts. 12.4 Start with equal weights. Evaluating methods 13.6 Describe potential biases of forecasters. 13.10 Test assumptions for validity. 13.32 Conduct explicit cost-benefit analyses. Assessing uncertainty 14.1 Estimate prediction intervals (Pis). 14.2 Use objective procedures to estimate explicit prediction intervals. 14.3 Develop prediction intervals by using empirical estimates based on realistic representations of forecasting situations. 14.5 Ensure consistency over the forecast horizon. 14.7 When assessing Pls. list possible outcomes and assess their likelihoods. 14.8 Obtain good feedback about forecast accuracy and the reasons why errors occurred. 14.9 Combine prediction intervals from alternative forecasting methods. 14.10 Use safety factors to adjust for overconfidence in the Pie. 14.11 Conduct experiments to evaluate forecasts. 14.13 incorporate the uncertainty associated with the prediction of the explanatory variables in the prediction intervals. 14.14 Ask for a judgmental likelihood that a forecast will fall within a predefined minimum-maximum interval. Table A2: Principles apparently contravened in AMD. Structuring the problem 2.1 Identify possible outcomes prior to making forecasts. 2.7 Decompose time series by level and trend. Identifying data sources 3.2 Ensure that the data match the forecasting situation. 3.3 Avoid biased data sources. 3.4 Use diverse sources of data. Collecting data 4.1 Use unbiased and systematic procedures to collect data. 4.3 Ensure that the information is valid. Selecting methods 6.4 Use quantitative methods ratherthan qualitative methods. 6.9 Assess acceptability and understandability of methods to users. Implementing methods: General 7.1 Keep forecasting methods simple. Implementing quantitative methods 9.2 Match the model to the underlying phenomena. 9.4 Weight the most relevant data more heavily. Implementing methods: Quantitative models with explanatory variables 10.1 Rely on theory and domain expertise to select causal (or explanatory) variables. 10.2 Use all important variables. 10.5 Use different types of data to measure a relationship. Combining forecasts 12.5 Use trimmed means. medians. or modes. 12.7 Use domain knowledge to vary weights on component forecasts. 12.8 Combine forecasts when there is uncertainty about which method is best. 12.9 Combine forecasts when you are uncertain about the situation. 12.10 Combine forecasts when it is important to avoid large errors. Evaluating methods 13.1 Compare reasonable methods. 13.2 Use objective tests of assumptions. 13.7 Assess the reliability and validity of the data. 13.8 Provide easy access to the data. et al.: Polar Bear Population Forecasts: A Public-Policy Forecasting Audit 392 Interfaces 38(5), pp. 382?405, @2008 INFORMS 13.17 Examine all important criteria. 13.16 Specify criteria for evaluating methods priorto analyzing data. 13.27 Use ex post error measures to evaluate the effects of policy variables. Assessing uncertainty 14.6 Describe reasons why the forecasts might be wrong. Presenting forecasts 15.1 Present forecasts and supporting data in a simple and understandable form. 15.4 Present prediction intervals. Learning to improve forecasting procedures 16.2 Seek feedback about forecasts. 16.3 Establish a formal review process for forecasting methods. Table A.3: Principles not rated because of lack of information in AMD. Structuring the problem 2.5 Structure problems to deal with important interactions among causal variables. Collecting data 4.4 Obtain all of the important data. 4.5 Avoid the collection of irrelevant data. Preparing data 5.1 Clean the data. 5.2 Use transformations as required by expectations. 5.3 Adjust intermittent series. 5.4 Adjust for past events. 5.5 Adjust for systematic events. 5.6 Use multiplicative seasonal factors fortrended series when you can obtain good estimates fer seasonal factors. 5.7 Damp seasonal factors for uncertainty. Selecting methods 6.6 Select simple methods unless empirical evidence calls for a more complex approach. implementing methods: General 7.2 The forecasting method should provide a realistic representation of the situation. implementing judgmental methods 8.4 Provide numerical scales with several categories for experts' answers. implementing methods: Quantitative models with explanatory variables 10.3 Rely on theory and domain expertise when specifying directions of relationships. 10.4 Use theory and domain expertise to estimate or limit the magnitude of relationships. integrating judgmental and quantitative methods 11.1 Use structured procedures to integrate judgmental and quantitative methods. 11.2 Use structured judgment as inputs to quantitative models. 11.3 Use prespecified domain knowledge in selecting. weighting, and modifying quantitative methods. 11.4 Limit subjective adjustments of quantitative forecasts. Evaluating methods 13.4 Describe conditions associated with the forecasting problem. 13.5 Tailor the analysis to the decision. 13.9 Provide full disclosure of methods. 13.11 Test the client's understanding of the methods. 13.19 Assess face validity. Assessing uncertainty 14.12 Do not assess uncertainty in a traditional (unstructured) group meeting. Learning to improve forecasting procedures 16.4 Establish a formal review process to ensure that forecasts are used properly. Table AA: Principles properly applied or apparently properly applied (italics) in AMD. Setting objectives 1.1 Describe decisions that might be affected by the iorecasts. Structuring the problem 2.2 Tailor the level of data aggregation (or segmentation) to the decisions. 2.3 Deccmpose the problem into parts. 2.6 Structure problems that involve causal chains. identifying data sources 3.1 Use theory to guide the search for information on explanatory variables. Collecting data 4.6 Obtain the most recent data. Preparing data 5.3 Use graphical displays for data. Selecting methods 6.3 Use structured ratherthan unstructured forecasting methods. 6.5 Use causal methods rather than naive methods it feasible. implementing methods: General 7.5 Adjust for events expected in the future. 7.6 Pool similar types of data. 7.7 EnSUre consistency with forecasts of related series and related time periods. Implementing judgmental methods 8.3 Ask experts to justify their forecasts in writing. implementing methods: Quantitative models with explanatory variables 10.7 Forecast for alternate interventions. Presenting forecasts 15.2 Provide complete. simple. and clear explanations of methods. 15.3 Describe your assumptions. Learning to improve forecasting procedures 16.1 Consider the use of adaptive torecasting models. et al.: Polar Bear Population Forecasts: A Forecasting Audit Interfaces 33(5), pp. 332?405, 2008 INFORMS 393 Table 11.5: Principles contravened in Hunter et al. (H6). Setting objectives 1.3 Make sure forecasts are independent of politics. 1.4 Consider whether the events or series can be iorecasted. Structuring the problem 2.6 Structure problems that involve causal chains. Identifying data sources 3.4 Use diverse sources oi data. 3.5 Obtain information from similar (analogous) series or cases. Such information may help to estimate trends. Collecting data 4.4 Obtain all of the important data. Preparing data 5.2 Use transformations as required by expectations. 5.4 Adjust for past events. 5.5 Adjust for systematic events. Selecting methods 6.1 List all the important selection criteria before evaluating methods. 6.2 Ask unbiased experts to rate potential methods. 6.6 Select simple methods unless empirical evidence calls for a more complex approach. Match the forecasting method(s) to the situation. 6.8 Compare track records of various forecasting methods. 6.10 Examine the value of alternative forecasting methods. Implementing methods: General 7.1 Keep forecasting methods simple. 7.2 The forecasting method should provide a realistic representation of the situation. 7.3 Be conservative in situations of high uncertainty or instability. 7.4 Do not forecast cycles. lmplemeuting quantitative methods 9.1 Tailor the forecasting model to the horizon. 9.2 Match the model to the underlying phenomena. 9.3 Do not use "lit" to develop the model. 9.5 Update models frequently. Implementing methods: Quantitative models with explanatory variables: 102 Use all important variables. 10.5 Use different types of data to measure a relationship. 10.7 Forecast for alternate interventions. 10.9 Shrink the forecasts of change if there is high uncertainty for predictions of the explanatory variables. Integrating judgmental and quantitative methods 11.1 Use structured procedures to integrate judgmental and quantitative methods. 11.2 Use structured judgment as inputs to quantitative models. 11.3 Use domain knowledge in selecting, weighting. and modifying quantitative methods. Combining forecasts 12.1 Combine forecasts from approaches that differ. 12.2 Use many approaches (or forecasters), preferably at least five. 12.3 Use formal procedures to combine forecasts. 12.8 Combine forecasts when there is uncertainty about which method is best. 12.9 Combine forecasts when you are uncertain about the situation. 12.10 Combine forecasts when it is important to avoid large errors. Evaluating methods 13.1 Compare reasonable methods. 13.2 Use objective tests ofassumptions. 13.3 Design test situations to match the forecasting problem. 13.5 Tailorthe analysis to the decision. 13.6 Describe potential biases of forecasters. 13.7 Assess the reliability and validity ofthe data. 13.8 Provide easy access to the data. 13.10 Test assumptions fur validity. 13.12 Use direct replications of evaluations to identify mistakes. 13.13 Replicate forecast evaluations to assess their reliability. 13.16 Compare forecasts generated by different methods. 13.17 Examine all important criteria. 13.18 Specify criteria for evaluating methods priorto analyzing data. 13.26 Use out-of-sample (ex ante) error measures. 13.27 Use ex post error measures to evaluate the effects of policy variables. 13.31 Base comparisons of methods on large samples of forecasts. Assessing uncertainty 14.3 Develop prediction intervals by using empirical estimates based on realistic representations of forecasting situations. 14.5 Ensure consistency over the forecast horizon. 14.9 Combine prediction intervals from alternative forecasting methods. 14.10 Use safety factors to adjust for overcontidence in the Pie. 14.11 Conduct experiments to evaluate forecasts. 14.13 incorpOrate the uncertainty associated with the prediction of the explanatory variables in the prediction intervals. 14.14 Ask for a judgmental likelihood that a forecast will fall within a predefined minimum-maximum interval (not by asking people to set upper and lower confidence levels). Presenting forecasts 15.1 Present forecasts and supporting data in a simple and understandable form. 15.2 Provide complete. simple. and clear explanations of methods. Table A.6: Principles apparently contravened in H6. Setting objectives 1.1 Describe decisions that might be affected by the forecasts. 1.2 Priorto forecasting. agree on actions to take assuming different possible forecasts. Structuring the problem 2.1 identify possible outcomes priorto making forecasts. 2.3 Decompose the problem into parts. Identifying data sources 3.2 Ensure that the data match the forecasting situation. 3.3 Avoid biased data sources. Collecting data 4.2 Ensure that information is reliable and that measurement error is low. 4.3 Ensure that the information is valid. et al.: Polar Bear Population Forecasts: A Public-Policy Forecasting Audit 394 Preparing data 5.3 Adjust intermittent series. 5.7 Damp seasonal factors for uncertainty. 5.8 Use graphical displays for data. implementing methods: General 7.6 Pool similar types of data. implementing methods: Quantitative models with explanatory variables 10.4 Use theory and domain expertise to estimate or limit the magnitude of relationships. 10.8 Apply the same principles to forecasts of explanatory variables. Evaluating methods 13.4 Describe conditions associated with the forecasting problem. 13.9 Provide full disclosure of methods. Assessing uncertainty 14.6 Describe reasons why the forecasts might be wrong. 14.7 When assessing Pis. list possible outcomes and assess their likelihoods. 14.8 Obtain good feedback about forecast accuracy and the reasons why errors occurred. Table A.7: Principles not rated because of lack of information in H6. Setting objectives 1.5 Obtain decision makers' agreement on methods. Structuring the problem 2.7 Decompose time series by level and trend. identifying data sources 3.1 Use theory to guide the search for information on explanatory variables. Collecting data 4.1 Use unbiased and systematic procedures to collect data. 4.5 Avoid the collection of irrelevant data. Preparing data 5.1 Clean the data. Selecting methods 6.4 Use quantitative methods ratherthan qualitative methods. 6.5 Use causal methods rather than naive methods it feasible. 6.9 Assess acceptability and understandabillty of methods to users. Evaluating methods 13.11 Test the client's understanding of the methods. 13.19 Assess face validity. Presenting forecasts 15.3 Describe your assumptions. Learning to improve forecasting procedures 16.2 Seek feedback about forecasts. 16.3 Establish a formal review process for forecasting methods. 16.4 Establish a formal review process to ensure that forecasts are used properly. Interfaces 38(5), pp. 332?405, @2008 INFORMS Table A.8: Principles properly applied or apparently properly applied in H6. Structuring the problem 2.2 Tailor the level of data aggregation (or segmentation) to the decisions. Collecting data 4.6 Obtain the most recent data. Selecting methods 6.3 Use structured rather than unstructured forecasting methods. implementing methods: Quantitative models with explanatory variables 10.1 Rely on theory and domain expertise to select causal (or explanatory) variables. 10.3 Fier on theory and domain expertise when specifying directions of relationships. 10.6 Prepare forecasts for at least two alternative environments. Assessing uncertainty 14.1 Estimate prediction intervals (Pls). 14.2 Use objective procedures to estimate explicit prediction intervals. Presenting forecasts 15.4 Present prediction intervals. 15.5 Present forecasts as scenarios. Acknowledgments We thank Don Esslemont, Milton Freeman, Paul Goodwin, Benny Peiser, Orrin I??key, Tom Stewart, Mitchell Taylor, and two anonymous reviewers for their comments on ear- lier drafts. Janice Dow and Kelly Jin provided editorial assistance. References Amstrup, S. C., G. W. Garner, G. M. Dumer. 1995. Polar bears in Alaska. E. T. Ln Rue, ed. Our Lining Resources: A Report to the Nation on the Abundance, Distributions, and Herrillr of tire 1.15 Animals. and Ecosystems. US Department of the interior? Narionai Biological Sciences, Washington, 351?353. Amstrup, S. C., B. G. Marcot, D. C. Douglas. 2007. Forecasting the rangewide status of polar bears at selected times in the 2lst century. Administrative Repert, USGS Alaska Science Center, Anchorage, AK. Amish-up, S. C., T. L. McDonald, G. M. Durner. 2004. Using satellite radiotelemet-ry data to delineate and manage wildlife popula- tions. Wildlife Soc. Bull. 32 661?679. Amstrup, S. C. T. L. McDonald, 1. Stirling. 2001. Polar bears in the Beaufort Sea: A 30-year mark?recapture case history. I. Agricul? lurnl Biol. Environ. Siniist. 6 221?234. Amstrup, S. C., l. Stirling, W. Lentfer. 1986. Past and present sta- tus of polar bears in Alaska. Wildlu?c Soc. Bull. 14 241?254. Angliss, R. P., R. B. Outlaw. 2007. Alaska marine mammal stock assessments, 2006. Retrieved June 4, 2008, J. S. 1978. Long?Rouge Forecasting: Prom Ball to Computer. New York. et 211.: Polar Bear Population Forecasts: A Public-Policy Forecasting Audit interfaces 38(5), pp. 382?405, @2003 INFORMS 395 J. S. 1980. The seer-Sucker theory: The value of experts in forecasting. Tech. Rev. BBUune?July) 16?24. S. 1985. Long-Range Forecasting: From Ball to Computer, 2nd ed. Wiley-Interscience, New York. J. S. 2001. Principles of Forecasting: A Handbook for Researchers and Practitioners. Kluwer Academic Publishers, Norwell, MA. J. S., F. Collopy. 1993. Causal forces: Structuring knowl- edge for time-series extrapolation. Forecasting 12 103?115. Bergen, 5., G. M. Burner, D. C. Douglas, S. C. Amsh?up. 200?. Predicting movements of female polar bears between summer sea ice {mag-mg habitats and terrestrial cleaning imbilals of Alaska in the list century: Proposed methodology and pilot assessment. Administrative Report, USES Alaska Science Cen- ter, Andiorage, AK. Brigham-Grette, J., D. M. Hopkins. 1995. Emergent marine record and paleoclimate of the last interglaciation along the North? west Alaskan coast. Quaternary Res. 43 159?173. Cerf, C., V. Navasky. 1998. The Experts Speak. Random House, New York. DeWeaver, E. 2007. Uncertainty in climate model projections of Arctic sea ice decline: An evaluation relevant to polar bears. Adminislrative Report. USGS Alaska Science Center, Anchorage, AK. Durner, G. M., D. C. Douglas, R. M. Nielson, S. C. Amstrup, T. L. McDonald. 200?. Predicting the future distribution of polar bear habitat in the polar basin from resource selection functions applied to let century general circulation model projections of sea ice. Administrative Report, USGS Alaska Science Center, Anchorage, AK. Dyck, M. G., S. Kornberg. 200?. Observations of a wild polar bear (Ursus umrr'timus} fishing Arctic charr (Salvati- uus and Fourl'lorn sculpin {Myurocephalus quadricomis). Polar Biol. 30 1625?0320. Dyck, M. G., W. Soon, R. K. Baydack, D. R. Legates, S. Baliunas, T. F. Ball, L. 0. Hancock. 2007. Polar bears of western Hudson Bay and climate change: Are warming spring air temperatures the "ultimate" survival control factor? Ecological Complexity 4 73434. Green, K. C., J. S. 2007. Global warming: Forecasts by scientism versus scientific forecasts. Energy Environ. 18 997?1021. Hamilton, T. D., l. Srigham-Grette. 1991. The last interglaciation in Alaska: Stratigraphy and paleoecology of potential sites. Qua- ternary tutti-nut. [0?12, 49?71. Holloway, 6., T. Sou. 2002. Has Arctic sea ice rapidly thiimed? I. Cliumte 15 1691?1701. Hunter, C. M., H. Caswell, M. C. Ruuge, S. C. Amati-up, E. V. Regehr, I. Stirling. 2007. Polar bears in the Southern Beaufort Sea ll: Demography and population growth in relation to sea ice conditions. Administrative Report, USES Alaska Science Center, Anchorage, AK. Malling, H., D. A. Riedel, Z. Gedalof. 2005. Trends in the draft and extent of seasonal pack ice, Canadian Beaufort Sea. Geophysical Res. Lett. 32(24) L24501, Nghiem, S. V., I. G. Rigor, D. K. Perovich, P. Clemente-Colon, J. W. Weatherly, G. Neumann. 2007. Rapid reduction of Arctic perennial sea ice. Geophysical Res. Lett. 34 L19504, doi:10.1029/ 2007GL031138. Neigaard-Pedersen, N., N. Mikkelsen, S. Lassen, Y. Kristofiersan, 13. Sheldon. 2007. Reduced Sea ice concentrations in Audit: Ocean during the last interglaciai period revealed by sedimth cores oft northern Greenland. Palroci'auagrapliy 22 PAIZIS, Obbard, M. T. L. McDonald, E. J. l-Iowe, E. V. Regchr, E. S. Richardson. 200?. Trends in abundance and survival for polar bears from Southern Hudson Bay, Canada. 1934?2005. Admin- istrative Report, USGS Alaska Science Center, Anchorage, AK. Ogi, lvl., J. M. Wallace. 2007. Summer minimum Arctic sea ice extent and the associated summer atmospheric circulation. Geophysi? cal Res. Lett. 34 1.123705, Pilkey, O. H., L. Pilkey?jarvis. 2007. Useless Arithmetic. Columbia University Press, New York. Prestrud, P., l. Stirling. 1994. The intemational polar bear agree- ment and the current status of polar bear conservation. Aquatic Mammals 20 113?124. Regehr, E. V., S. C. Amstrup, l. Stirling. 2006. Polar bear population status in the Southern Beaufort Sea. Retrieved November 17, 2006, ofr20061337.pdf. Rugehr, E. ?01, C. M. Hunter, H. Caswell, S. C. AmsIIup, I. Stirling. 200?. Polar bears in the Southern Beaufort Sea 1: Survival and breeding in relation to sea ice conditions, 2001?2006. Adminis- trative Report, USGS Alaska Science Center, Anchorage, AK. Richter-Mange, J., I. Overland, A. V. Romanovsky, R. l. Morison, 5. Nghiem, et al. 200?. State of the climate in ms: Arctic. Bull. Amer. Meteorological Soc. 33 562-52]. Rode, K. D., S. C. Amstrup, E. V. Regehr. 2007. Polar bears in the Southern Beaufort Sea Ill: Stature, mass, and cub recruit- ment in relationship to time and sea ice extent between 1982 and 2006. Administrative Report, USGS Alaska Science Center, Anchorage, AK. Schnaars, S. P. 1984. Situational factors affecting forecasting accu- racy. Marketing Res. 21 290?297. Schnaars, S. P., R. I. Bavuso. 1986. Extrapolation models on very short?term_forecasts. I. Bus. Res. 14 27?36. Soon. W., Baliunas, S. 3. idso, K. Y. Kondratyev, E. S. Posmentier. 2001. Modeling ciirna tic effects of anthropogenic carbon diox- ide emissions: Unlomwns and uncartaintits. Climate Res. 18 39?225. Stempniewicz, L. 2006. Polar bear predatory behavior toward moulting barnacle geese and nesting glaucous gulls on Spitsbergen. Arctic 59 247?251. Stirling. l. 2002. Polar bears and seals in the Eastern Beaufort Sea and Amundsan Gulf: A of population trends and ecological relationships over three decades. Arctic 55 (Suppl. 1) 59?76. Stirling, L, T. McDonald, E. S. Richardson, E. V. Regehr. 200?. Polar bear population status in the Northern Beau- fort Sea. Administrative Report, USGS Alaska Science Center, Anchorage, AK. Tetlock, P. E. 2005. Expert Political judgment: How Good Is It? How Can We Know? Princeton University Press, Princeton, NJ. Trenberth, K. 2007. Predictions of climate. Retrieved June 2, 2008, http://' blogs.nature.com/ climatefeedback/ 2007/ 06/ prediclions_of_climate.html. Zhang, J. 2007. Increasing Antarctic sea ice under warming atmo- spheric and oceanic conditions. I. Climate 20 2515?2529. ATTACHMENT ECOLOGICAL COMPLEXITY 5 (2008) 289?302 VieWpoint journal homepage: ECOLOGICAL available at g. .. Reply to response to et a1. (2007) on polar bears and climate change in western Hudson Bay by Stirling et al. (2008) M.G. W. Soon R.K. Baydaclz C, D.R. Legates d, S. Baliunas, T.F. Balle, L.O. Hancockf ?Environmental Technology Program, Nunavut Arctic College, Box 600, Iqaluit, Nunavut XOA 0H0, Canada hI-tlaward?Smithsonian Center for Astrophysics, Cambridge, MA 02138, USA CClayton H. Riddell Faculty of Environment, Earth, and Resources, University of Manitoba, Winnipeg, Manitoba R3T 2N2, Canada dOjj?ice of the State Climatologist, University of Delaware, Newark, DE 19716-2541, USA e205 27 Songhees Road, Victoria British Columbia, VSA 7M6, Canada H-S-503, 1818 Street, NW, Washington, DC 20433, USA ARTICLE INFO Article history: Received 2 April 2008 Received in revised form 12 May 2008 Accepted 13 May 2008 Published on line 11 July 2008 Keywords: Polar bear Climate change Hudson Bay Extinction Sea ice Ursus maritimus ABSTRACT We address the three main issues raised by Stirling et a1. [Stirling I., Derocher, A.E., Gou gh, W.A., Rode, K., in press. Response to et al. (2007) on polar bears and climate change in western Hudson Bay. Ecol. Complexity]: (1) evidence of the role of climate warming in affecting the western Hudson Bay polar bear population, (2) responses to suggested impor- tance of human?polar bear interactions, and (3) limitations on polar bear adaptation to projected climate change. We assert that our original paper did not provide any "alternative explanations [that] are largely unsupported by the data" or misrepresent the original claims by Stirling et al. [Stirling I., Lunn, N.J., Iacozza, 1., 1999. Long?term trends in the population ecology of polar bears in Western Hudson Bay in relation to climate change. Arctic 52, 294? 306], Derocher et al. [Derocher, A.E., Lunn, N.J., Stirling, I., 2004. Polar bears in a warming climate. Integr. Comp. Biol. 44, 163?176], and other peer-approved papers authored by Stirling and colleagues. In sharp contrast, we show that the conclusion of Stirling et a1. (Stirling, I., Derocher, A.E., Gough, W.A.. Rode, K., in press. Response to et al. (2007) on polar bears and climate change in western Hudson Bay. Ecol. Complexity] suggesting warming temperatures (and other related climatic changes) are the predominant determi? nant of polar bear population status, not only in western Hudson (WI-I) Bay but also for populations elsewhere in the Arctic is unsupportable by the current scienti?c evidence. The commentary by Stirling et al. [Stirling I., Derocher, A.E., Cough, W.A., Rode, K., in press. Response to et (2007) on polar bears and climate change in western Hudson Bay. Ecol. Complexity] is an example of uni-dimensional, or reductionist thinking, which is not useful when assessing effects of climate change on complex ecosystems. Polar bears of WH are exposad to a multitude of environmental perturbations including human inter- ference and factors unknown seal population size, possible competition with polar DOI of original article: 10.1016/j.ecocom.2008.01.004 Corresponding author. Tel.: +1 617 495 7488; fax: +1 617 495 7049. E-mail addresses: (MG. Dyck), wsoon?cfa.harvard.edu (W. Soon). see front matter. Published by Elsevier B.V. 290 ECOLOGICAL COMPLEXITY 5 (2008) 289?302 bears from other populations) such that isolation of any single variable as the certain root cause climate change in the form of warming spring air temperatures), without recognizing confounding interactions, is imprudent, unjusti?ed and of questionable scienti?c utility. et al. [Dyck, M.G., Soon, W., Baydack, R.K., Legates. D.R., Baliunas, 5., Ball, T.F., Hancock, L.0., 2007. Polar bears of Western Hudson Bay and climate change: Are warming spring air tem- peratures the "ultimate" survival control factor? Ecol. Complexity, 4, 73-84. com.2007.03.002] agree that some polar bear populations may be negatively impacted by future environmental changes; but an oversimpli?cation of the complex ecosystem interactions (of which humans are a part) may not be bene?cial in studying external effects on polar bears. Science evolves through questioning and proposing hypotheses that can be critically tested, in the absence of which, as Krebs and Borteaux [Krebs, (2.1., Berteaux, D., 2006. Problems and pitfalls in relating climate variability to population dynamics. Clim. Res. 32, 143-149] observe, ?we will be little more than storytellers." Published by Elsevier B.V. 1. Introduction Stirling et al. (2008) contains claims that are inconsistent with the underlying data and the papers they refer to; most of which were written by Stirling and colleagues themselves. For example. in their abstract, Stirling et al. (2008) argue that the "decline" in the VVH polar bear (Ursus maritimus) population has "accelerated over time". However, Fig. 1, adopted directly from Regehr et al. (2007), shows that the decline has been constant Our paper set out to answer the question posed in the title of our paper ?Polar bears of western Hudson Bay and climate change: Are warming spring air temperatures the 'ultimate' survival control factor?" - by examining the original hypoth- esis contained in Stirling et al. (1999) and the later extrapola- tions by Derocher et al. (2004). We concluded that it is neither correct nor useful to over-emphasize global warming (as caused by anthropogenic emissions of C02) as the predomi- nant explanatory variable for both climatic and polar bear population changes in the WH region. Extrapolating from the highly limited understanding of the WH regional polar bear population and climate to the situation occurring elsewhere in the Arctic is premature and speculative. Rosing~Asvid (2006) offered similar criticisms and alternative interpretations1 of data presented in Stirling et al. (1999). 1 For example, Rosing?Asvid (2006) commented that "Newborn ringed seal pups are the most reliable food source for polar bears during early spring, and the positive effects of a warm spring are therefore likely to be common in most areas, but the negative effects (a lack of fat ringed seal pups later in the season) might be less severe in areas where the bears are not forced on land during summer. Off east Greenland, polar bears can hunt throughout most of the summer, and the number of seals increases there during summer as both harp seals (Pagophilus groenlandicus 031x- belen, 1777)) and hooded seals (Cystophora cristata (Erxbelen, 1777)) from the west Atlantic Ocean seek out the dense drift ice that floWs down along the coast. The concentration of harp and hooded seals, as well as bearded seals (Erignathus barbatus (Erxbe- len, 1777)), therefore increases as the sea ice shrinks and their densities become higher during mild years with little ice. Polar bears therefore seem to have all the advantages of a trend towards a milder climate in this area, at least initially, until their predation has lowered the ringed seal population signi?cantly." 360-361) See more speci?c criticism of Stirling et al. (1999) by Rosing-Asvid (2006) in Section 2 below. We explained in our paper why the purported ?loss? of September sea ice in the Arctic, as projected by the computer climate models under the scenarios of future rise in atmo- spheric C02, is not relevant for the WH polar bear population. Stirling et al. (2008) continue to frame their hypothesis as "climate warming in western Hudson Bay is the major factor causing the sea ice to break-up at progressively earlier Furthermore, Stirling et al. (2008) have apparently modi?ed their original hypothesis that warming late spring temperatures are the predominant negative effect on polar bear populations by adding another contributing factor? unsustainable annual harvest by Inuit hunters.2 This new factor posited by Stirling et al. (2008) is fully consistent with our call for extreme caution in promoting a single factor as the cause of the apparent changes in WH polar bear population characteristics. We recognize the importance of the spring feeding period. A later freeze-up would also be a negative factor. Such an event would cause polar bear condition to deteriorate and possibly lead to the declining numbers of polar bears in the WH regional population; especially because the polar bears 2 Dr. Martina 'I?yrrell of the Scott Polar Research Institute at the University of Cambridge commented that ?Since the late 1960s polar bears in Nunavut/NWT have been hunted according to adaptive management practices, with quotas changing from year to year re?ecting changing scienti?c evidence regarding polar bear numbers. In Arviat, for example, between 1978 and 2006 the quota steadily rose from 15 to 20 and ?nally 22 bears per year. However, in 2007, the quota was reduced from 22 to 15 and in 2008 it is to be further reduced to 3 (This re?ects a Westem Hudson Bay-wide quota reduction from 56 in 2006 to 38 in 2007 to 8 in 2008.). I appreciate that the management of polar bears is not perfect, and that the 2005 quota increases were as much about politics as about evidence. I am also not in a position to comment on the sustain? ability of this hunt, but I simply want to make the point that it has been managed, and that hunting takes place under a strictly enforced management regime. Further, unlike some other spe- cies, hunting transgressions rarely if ever occur and the rules are adhered to across the board. .. .The point I?m making is that Stirling et al.?s reference to unsustainable harvest by Inuit hunters places blame on Inuit who have conducted most of their hunting under a strictly enforced management regime, while historically, up to the middle part of the 20th century, non-indigenous people were hunting bears in the absence of conservation rules and guidelines." (email communication 8th April 2008). ECOLOGICAL COMPLEXITY 5 (2008) 289?302 291 :Smoothed curve Bootslrapped 95% CL 3. .. 'il' 'a I I 1985 1990 1995 2000 Yr Fig. 1 - Estimates of the western Hudson Bay polar bear population from the domain-limited study by Regehr et al. (2007) [see our reply in Section 2.3 below for further information and discussion]. would likely have much less to eat during the ice-free periods from late spring through fall. However, we were unable to con?rm any substantive warming trends for fall temperatures at Churchill. Neither could Stirling et al. (1999) ?nd any clear evidence of delays in freeze-up dates3 around the WH. The lack of consistency makes a simple explanation impossible and precludes any direct causal connection between climate change and polar bears. Past correlation-based hypotheses about ecosystems have failed to survive subsequent empirical evidence (see Krebs and Berteaux, 2006; Berteaux et al., 2006; Botkin et al., 2007). 2. Evidence of the role of climate warming in effecting the western Hudson Bay polar bear population First and foremost, it is worth reminding from Rosing~Asvid (2006: 359-360) (not available at the time when et al. (2007) was prepared and submitted for publication in December 2005) that different and altemative interpretations about the relationship between climate and polar bear population and condition are possible: ?The catch statistics present?ed here indicate a strong increase in the number of polar bears and a reduction in the number of ringed seals during mild climatic periods and vice-versa during cold periods. If these trends reflect population dynamics, they contradict the general belief that polar bears suffer during mild periods and that both ringed seal and polar bear populations fluctuate positively with productivity in the sea. If numbers of both species re?ect productivity in the sea, one would expect a relatively close connection between the two, with ringed seals 3 See Fig. 3a of Gagnon and Gough (2005) where statistically signi?cant later freeze-up tendencies can only be shown for lim- ited spots around northem and northeastern Hudson Bay. In contrast to expectation, Gagnon and Gough (2005) found a ten- dency towards earlier freeze-up on the western side ofJames Bay. leading the trend. Mild periods have, however, been linked to strong polar bear predation on ringed seal pups, and in the following I will argue that the data from [western] Hudson Bay indicate that strong spring survival of polar bear cubs during some years coincides with early ice breakup. This again seems to contradict the poorer body condition and reduced natality of adult female bears in Hudson Bay documented during mild years. Yet there are factors that in theory can explain these contradictions and trigger different responses to climate change. Stirling et al. (1999) link lower polar bear reproduction and poorer body condition in fall in western Hudson Bay with increasingly earlier sea-ice breakup, but all major changes in body condition and birth rate took place from 1981 to 1985, when the population increased from ca. 500 to 1400 individuals (Stirling et al., 1999). During this period, both early fall freeze-ups and late spring breakups (NB: instead of late fall freeze?ups and early spring breakups] of the ice occurred, so the strong deterioration in various life-history traits during these years could not have been caused by shorter ice season, but is more likely density-dependent, and if these years are removed, the trends in the time series change signi?cantly. The trend towards increasingly earlier ice breakup began in 1993 (Fig. 3a in Stirling et al., 1999), and when this new trend occurred, cub survival (from spring to fall) increased from around 50% in 1988?1992 to more than 70% during the follo- wing 5 years with mild springs (Pig. 7 in Stirling et al., 1999)." 2.1. Increasing air temperatures in western Hudson Bay The main points on this issue from our original paper are (1) a large inter-annual variability in the temperature data from the western Hudson Bay exists and (2) trend analyses are highly sensitive to the chosen time period. Nevertheless, Stirling et al. ?5 (2008) discussion regarding increasing air temperatures in WH raises signi?cant questions. If polar bear populations are affected by an earlier spring ice melt in WH, why present annual mean temperatures and not temperatures from April and May?the key spring months according to Stirling et al. (1999)? In addition, the selection4 of March, June, and July bypasses the cumulativeiheat loading which is more physi- cally relevant to melting and accumulation of sea ice within Hudson Bay. Moreover, Stirling et al. (2008) suggest that the climatic impact on polar bears is nearly instantaneous and occurs in isolation from any additional factors and feedbacks. Such a suggestion is at odds with Rigor and Wallace (2004) who documented a delayed effect on Arctic ocean/sea ice of about 15 years from the atmospheric anomaly that occurred around 1989?1990. 4 In their original submission, Stirling et al. (2008) included the following key claim: ?They [Gagnon and Gough, 2005] also reported highly signi?cant warming trends over the same periods for the months Of March, June, and July at Churchill and Chester- ?eld Inlet, both of which are within the area occupied by the WH population" which has since been deleted but the gap remains? upon seeing our earlier review. 292 ECOLOGICAL COMPLEXITY 5 (2008) 289?302 ?Cldec I I I a .yt I ommhmo Pig. 2 - AVI-IRR clear-sky temperature trends [chart dated 12 July 2007, courtesy of Dr. Joey Comiso of It also is puzzling that Stirling et al. (2008) do not explain that the "temperatures" deduced from the AVHRR instru- ments onboard NOAA satellites are strictly a measure of "clear-sky" conditions. The most obvious evidence for caution in reading these data is from the ever-changing trend maps produced in various publications [for example, Comiso (2003, 2006), Overland and Wang (2005), Serreze and Francis (2006)]. Indeed Fig. 3b of Serreze and Francis (2006), as cited by Stirling et a1. (2008) to con?rm the 1.2 ?C/decade warming offshore to 0.4 ?C/decade warming along-the-coast in spring, makes the case for our concern (see also Fig. Sb of Overland and Wang (2005)). A large area of cooling air temperatures (of magnitudes of to ?1.4 ?Ct/decade for 1982?2005 in Serreze and Francis (2006) or roughly to for 1982?2003 data presented in Overland and Wang (2005)) are situated west and southwest from Hudson Bay, which contradicts any robust conclusion about a consistent pattern of surface air tempera- ture warming trends over the region. We now focus discussion around Fi g. 2, updated from trend maps of the AVHRR clear-sky temperature as presented in Overland and Wang (2005) and Serreze and Francis (2006). These maps were produced courtesy of Dr. Joey Comiso of NASA the expert responsible for the data (but certainly not our interpretation) presented in various papers noted by Stirling et a1. (2008). Trends for spring and summer months based on data from 1982 through 2006 are presented to help study the relationship between polar bears and sea ice. Fig. 2 reaf?rms our caution of relying too much on trend maps, deducing climatic information. and formulating scien? ti?c interpretation (either on climate or polar bears) from AVHRR. since the data interval is relatively short (covering a maximum of only 26 years). Patchy trends with both warming ECOLOGICAL COMPLEXITY 5 (2008) 289~302 293 and cooling areas. including relatively large warming trends over the Hudson Bay basin in March. are obvious while relatively weaker warming trends exist over Hudson Bay and cooling trends over nearby land areas. Next. Stirling et al. (2008) accuse us of ignoring the most rapidly warming area near the Davis Strait (based on the same AVHRR record of ?clear?sky" temperatures?see Fig. 2 for the ever-changing inconsistency in its trend amplitudes depend- ing on what months are being highlighted). We referred to the parallel result from Stirling and Parkinson (2006) that failed to discern any signi?cant earlier spring ice break-up trends. despite the very large spring warming trend around the Davis Strait region which exceeds ?4.1 ?C/decade" in March (see Fig. 2 here or Fig. 3b of Serreze and Francis. 2006. cited by Stirling et al.. 2008). Using limited data. Stirling and Parkinson (2006) suggested increases in populations of harp and hooded seals from the 19705 to the 19905.5 This is ?particularly relevant to the likely increase in the size of the polarbearpopulation between the late 19705 and early 19905" 270 of Stirling and Parkinson, 2005) although Taylor (2007). for examme. suggested that the increase in the Davis Strait area is strongly linked to the "recovery from historical over-hunting after quotas were introduced" than nutrition. Thus. we reiterate that caution is required when extrapolating the situation from one region to another. Stirling et al. (2008) also misunderstand our discussion of the Arctic Oscillation (A0) index and its impact on large-scale patterns of temperature change in Hudson Bay. They cite only Skinner at al. (1998) although the contributions of Ball (1995) and Catchpole (1995) cannot be ignored. particularly where large and rapid spring warrningwas observed in 1779. 1780. and 1782 (no data for 1781). The apparent dipolar see-saw pattern in air temperature trends resulting from the use of only two stations was only an illustrative discussion?the full pattern (Figs. 7 and 8 of Cohen and Barlow. 2005 and cited by et al., 2007) shows better the sense of a northeast-versus-southwest warming oscillation. Cooling anomalies were included only to caution about making broad statements of warming trends and base conclusions on regional results alone. Our relevant discussion on the A0 index and its impact on Hudson Bay shows why we believe the A0 index could potentially be a useful climatic index in tracing the "dynamic of trophic interactions under various settings of the arctic ecosystem". It also is rather curious for Stirling et al. (2008) to cite Serreze et al. (2007) while concomitantly insisting that "rising air temperatures" drive Arctic sea ice changes. Serreze et al. (2007:1534) clearly note "the recent cold-season 5 Dr. Martina Tyrrell commented that ?This time frame coin- cides with the Greenpeace campaign to boycott seal skin products. leading to a dramatic decline in the European market for seal skins and the eventual EC sea] skin ban in 1989. With regard to ringed seals. the annual sale of skins to the Hudson Bay Company in one east Baf?n community (Clyde River) declined from 2504 pelts in 1979?1980 to 532 pelts in 1984?85 (Wenzel. George. 1991. Animal rights. human rights: Ecology, economy and ideology in an Arctic community. Toronto: University of Toronto Press. p. 124). I don't know how signi?cant this is. or if the impact of a seal skin ban would be so quickly re?ected in an increase in polar bear numbers between the late 19703 and early 19905 according to Stirling and Parkinson (email communication April 8th. 2008). warming. . .is itself driven by the loss of ice (a positive feedback). . This positive feedback is exactly what we described in our paper. The complex interaction among atmospheric circulation, changes in atmospheric tempera- tures. oceanic heat transport. and advective exchanges among basins and sea ice is more realistic than the one-dimensional interpretation of Stirling et al. (2008) that warming air directly and solely leads to an earlier ice break-up. Furthermore. it is encouraging to ?nd the discussion by Serreze et al. (2007) of the potential role that the advection of warm Paci?c summer water through the Bering Strait plays in affecting sea ice in the central Arctic ocean. as ?rst documented by Shimada et al. (2006). With regards to the statistical analyses offered by Gough et al. (2004b). we simply disagree with Stirling et al. (2008). Our primary objective in introducing this was to document the analyses by Gough et al. (2004b) that covered sea ice records from 1971 to 2003. We are at a loss to understand how ?climatic noise' data trends and ?multi-decadal oscillations" can be distinguished with data records of only 33 years using statistical techniques and algorithms. however sophisticated. Our initial point that large and signi?cant variability exists as documented from instrumental. historical and proxy records of climatic variables on a multi-decadal timescale is still well? taken. Our citation of Saucier et al. (2004) was solely to illustrate that ?Detailed high-resolution modeling con- siders tides. river runoff and daily meteorological forcing. found tidal mixing to be critically important for ice-ocean circulation within. and hence the regional climate of the Hudson Bay basin. There was no intention to mislead anyone about the content of Saucier et al. (2004). Our original paper did not infer any uni-dimensional relationship among snowfall. sea-ice thickness. or the length of seasonal sea ice season. We reference Gough et al. (2004a) and state they ?recently identi?ed snow depth as the primary governing parameter for the inter-annual variability of winter sea-ice thickness in Hudson Bay because of its direct insulating effect on ice surfaces. By contrast. the concurrent winter or previous summer air temperatures yield only weak statistical correlations with ice thickness." We properly quoted their article and referenced it to illustrate the multi- dimensionality of factors that affect climate and polar bear populations in again. the hypothesis of Stirling et al. (1999) is an extreme oversirnpli?cation of a complex reality. 2.2. Timing of sea ice breakup and e?ects on polar bears in WH Stirling et al. (2008) react to footnote 4 in our original paper regarding statistical signi?canCe in the detection of an earlier spring ice break-up, but we believe that they took our note out of context. Throughout our paper. the consistent message was to establish a defensible hypothesis and examine physical links and connections rather than to accept blindly the convenient results of statistical arguments. We simply defer to our previous replies in Section 2.1 above on both the dif?culty and subjectivity in attempting to interpret sea ice and temperature effects around the Hudson Bay region from the references cited by Stirling et al. (2008). 294 ECOLOGICAL COMPLEXITY 5 (2008) 289?302 Readers may also be surprised to ?nd the puzzling result about sea and lake ice thickening around WH as described in Gagnon and Cough (2006) but not reported by Stirling et a1. (2008) even though Cough is a co-author of Stirling et a1. (2008). Gagnon and Gough (20062177) state, the ?Freeze-up (of ice- cover in the Hudson Bay Region, HBR) typically occurs in October and November, ice cover reaches its peak thickness from late March to May, and water bodies in the HBR are usually ice-free beginning in early August. signi?cant thickening of the ice cover over time was observed on the western side of the Hudson Bay, while a slight thinning lacking statistical signi?cance was observed on the eastern side. Increasing maximum ice thickness at a number of stations is correlated to earlier freeze?up due to negative temperature trends cooling trends) in autumn. These results are in contrast to the projections from general circulation models (GCMs), and to the reduction in sea-ice extent and thickness observed in other regions of the Arctic. This contradiction must be addressed in regional climate change impact assessments." We agree with Gagnon and Gough (2006). We are not convinced that the existing records for air temperatures, sea ice and polar bear numbers are of suf?cient quality and appropriate duration to establish robust trends and tendencies. We stand by our call for caution in over- interpreting and extrapolating these inadequate datasets. 2.3. Population trend in WH and density dependence To examine population trends, the entire area that the individuals of a population occupy must be surveyed. The area most commonly surveyed for the WH polar bear population ranged from about latitude 57?00? to and longitude 9225' to Derocher and Stirling, 1995a), with an extension to the east towards Cape Tatnam (Lunn et al., 1997). However, the established and recognized WH polar bear population boundary extends beyond Manitoba into Nunavut up to around Chester?eld Inlet Aars et al., 2006). Nunavut communities in recent years have raised the issue that more bears are seen in these areas (Stirling and Parkinson, 2006; Tyrrell, 2006). To obtain a better picture of the complete population estimate of WH polarbears and to con?rm whether and by how much the population has declined, the Nunavut portion of the WH population must be included in a complete population survey and further discussion should await the conclusion of such a study. Mitchell Taylor. a scientist with expertise in the estimate of polar bear population, commented and explained that (Taylor, 2007): ?The WH [Western Hudson Bay] mark-recapture study has Some problems. The capture work for the 1978- 1992 interval described in [Derocher and Stirling, 1995_a]Derocher and Stirling (1995) [1995a] did not cover the area from the Ontario border to the core area around Churchill in most years, and did not extend north of the Seal River. Manitoba. The estimate was only about 1000, but this was "expert corrected" up to 1200 by the Federal/ Provincial Polar Bear Technical committee to account for the areas missed. The estimate of 1200 was approximately the number needed to sustain the current quotas of the day, and the population was believed to be sustaining the harvest stationary or increasing). The M-R analysis (Model 2 Jolly-Seber) was repeated in 1994?1995 and this time the area to the Ontario border was covered. The new estimate increased to 1200, which the authors suggested served to validate the correction identi?ed for the ?rst study. The analysis was done independently on males and females older than cubs. Then the cubs were estimated based on the adult to cub ratio observed. For the time, this was a good analysis. However, simulation models indicated that the adult survival estimate from both ?rst and second analyses was too low for the population to sustain itself, even with no harvest. The current analysis pools the CWS [Canadian Wildlife Service] captures with the Manitoba deterrent captures, probably because the CWS "research" captures did not contain enough sub-adults and many of the Manitoba deterrent program captures in the vicinity of Churchill were sub-adults. The new analysis increased the estimates for the mid 19805 to 1200, reduced the mid-1990 estimate to something less than 1100, and we are down to 935 in 2004. They also provide age-speci?c survival and recruitment estimates, and we know the harvest, so the decline can be checked for consistency using a demographic simulation model. The simulation model suggests a more rapid rate of decline, indicating their survival rates could be under- estimated, which is typical of M-R studies with sampling problems (unexplained capture heterogeneity). If you increase their adult survival rates by 1.5% you recover the decline identi?ed by the M-R estimates. This could be important because one explanation for the decline is that some of the population has shifted its range north of the Seal River. and the recent study has mistaken emigration for mortality. Perhaps the population decline is not as great as their estimates suggest. Perhaps there had been no decline. We will do some capture work in the areas not covered by CWS in the fall of 2007 to check this out. Clearly the natality rate in the WH polar bears has declined, with a concurrent reduction in the sustainable harvest from this population. Both population numbers and vital rates determine the sustainable harvest and the time trajectory of populations. We are attempting to resolve this uncertainty with additional sampling.? (Taylor, 2007) The Hudson Bay sea ice is shared extensively during winter and spring by bears of WH, southern Hudson Bay (SH), and Foxe Basin (PB) (Aars et al., 2006). The limited aerial surveys of SH population showed an increasing trend, at least from 1963 to 1996 (Stirling et al., 2004). The PB population estimate is dated (Aars et al., 2006; Taylor and Lee, 1994) and both the amount of bears and the rate of increase is unknown since only one biomarker mark-recapture study was conducted. The latest population estimate was set at 2300 based on Inuit Knowledge. It is likely, if SH and PB populations were/are increasing, that food competition occurs on the ice. To better ECOLOGICAL COMPLEXITY 5 (2008) 289-302 295 assess this hypothesis, we must await a new population estimate of F3 bears that will also produce information on the rate of increase of this particular regional population. Although Regehr et al. (2007) was not published by the time our original manuscript was submitted in December 2005. it is certainly important and relevant. But three serious problems are raised: (1) on density dependence and ruling out this factor for the observed WI-I polar bear conditions. Stirling et al. (2008) appear to have ignored what we presented in our original paper: ?Given these long-term data on population estimates and responses, it is possible that density-dependent processes have been imprinted in the Observed records of polar bears at It is important, however, to recognize the great dif?culties in demonstrating den- sity dependence in the population studies Ray and Hastings, 1996; Mayor and Schaefer. 2005) among which is the sensitivity of the phenomenon on spatial scale covered by the population sampling techniques Taylor et al., 2001)." Although the issue is still far from settled, we note that Stirling and Derocher (1993:1243) argue that "This result is consistent with the hypothesis that density?dependent responses were being shown by the polar bear population in (Derocher and Stirling, 1992) and suggests the polar bear population might already be the maximum size that can be supported by the existing ringed seal population. If an additional 3 weeks spent hunting does not enable the bears to signi?cantly increase the amount of body fat stored6 then it is possible a trend toward earlier break-up and a shortening of the time spent hunting will be re?ected fairly quickly in the lower rates of reproduction and cub survive (2) on the latest WH polar bear population estimates by Regehr et a1. (2007). Stirling et al. (2008) quote the WH polar bear population estimates from Regehr et al. (2007) to be "from about 1200 in 1987 to 935 in 2004? while failing to indicate that the 95% con?dence intervals for the two estimates are between 6 We note that this direct observation by itself, notably with low numbers of polar bears studied, can be also simply interpreted as the lack of sensitivity of polar bear body weights to extended sea ice condition. The limited data with a small sample size certainly does not symmetrically extend to imply much lower body weights with less spring ice or earlier spring ice break-up, but we included this quote in full to avoid being accused of changing the meaning and context from the original paper. Hunting for food must surely involve not only age and skill of the predator but also the elements of circumstance, including the availability of prey. To insist that simply having more or less ice cover in spring will determine whether the polar bears are more or less well fed. to the extent of having more or less body fat stored is an arguable assumption that ignores complex ecological realities. 1020?1368 and 793?1076, respectively. Stirling et al. (2008) apparently ignored estimates from other studies such as Lunn et al. (1997), with Stirling as a co?author, that we cited in our original paper. Lunn et al. (1997) give "a 1995 WH polar bear population of 1233 with a 95% con?dence interval that ranges from 823 to 1643 bears, so the actual con?dence in the ?decline? of theWH polarbearpopulation in 2004, relative to the 1995 values, is dif?cult to con?rm." If Stirling et al. (2008) are prepared to defend their con?dence in the estimates by Regehr et al. (2007). they must explain how truly sensitive those estimates are depending on data bases and sampling errors and techniques. These are problems and questions that require scienti?c and objective resolutions rather than subjective selection of population surveys that may not be compatible. The relevant paper by Tyrrell (2006) shows another aspect in the current debate on polar bear numbers around WH that are dismissed by Stirling et al. (2008) but recognized by Stirling and Parkinson (2006)?the sugges- tion of an increased population of polar bears in the area, according to Inuit knowledge. We are not arguing for an absolute answer one way or another but note that at least the possibility of an increased polar bear population around the WH should be discussed. This is why we must question Stirling et al. (2008) on their unjusti?ed assumption of any serious polar bear population decline in WH, including the incorrect insis- tence that the "decline" of the WH polar bear population "accelerated", which was refuted by the actual results from Regehr et al. (2007). (3) on WH polar bear population ?is now well below historic levels". This exaggerated statement undermines the statistical nature of the population estimates and model results. such as those produced by Regehr et al. (2007). What is the basis for claiming the current (2004) WH regional polar bear population is now at a historic low level? Stirling et a1. (2008) failed to de?ne the term "historic" which we take to imply a longer term average level rather than any recent point estimate like the population estimate for 2004. Some direct quotes illustrate the contradictory claims made by Stirling. Derocher and Stirling (1995a) (as cited in Stirling et al., 2008) recognized that past polar bear populations in the WH region were signi?? cantly overharvested before the establishment of the hunting quota in 1968. This implies knowledge of polar bear popula- tions at a previous time: "Throughout the 19505. 50?100 polar bears were harvested annually within the study area [Churchill, (Stirling et al., 1977). Adult females and their cubs leaving the denning area were harvested during spring for hides and dog food. Closure of York Factory (a fur trading post at the south edge of the study area) in 1957 reduced the harvest. Unrecorded harvest by military personnel stationed at Churchill ended in 1964. Except the removal of problem bears, harvest in the Churchill area ended by the mid- 19605. Harvest continued north of the study area in the Northwest Territories, where quotas were introduced in 296 COMPLEXITY 5 (2008) 289?302 1968. We suggest the population was lower during the 19605 than after this period." 220) In his open exchange with Douglas Clark, the senior author and his collaborator. Nick Lunn, noted7 that: ?Priorto the establishment of quotas on polarbears in 1968, and the elimination of several sources of substantial but undocumented harvesting, it appears the population was signi?cantly over-harvested. Thus, although not well documented, it is likely true, as has been suggested by Derocher and Stirling (1995) [1995a] and more recently by Scott and Stirling (2002) that the population increased in size in the latter half of the 1960s and possibly Well into the mid-19705 or so." We conclude that it is dif?cult to ?nd support for Stirling et al.?s (2008) claim that the 2004 polar bear population of 935 (with a prescribed 95% statistical con?dence range of 794? 1076) is ?well below historic levels.? We refer the interested reader to Calvert et al. (1991) where a WH polar bear population estimate by AB. Derocher (also a co-author of Stirling et al., 2008) for 1985 was either 685 or 773 bears, depending on the index applied. 2.4. Timing of sea-ice breakup and trends in SH It is puzzling to see this renewed attempt by Stirling et (2008) to argue for a general (if not universal) extension of their spring temperature-earlier ice break-up and polar bear relation from WH using the limited data available from southern Hudson Bay as discussed in the research note by Obbard et al. (2006). We notice that Stirling et a1. (2008) added the previously ignored reference of Stirling and Parkinson (2006) in their ?nal version. But Stirling and Parkinson (2006) were unable to show or con?rm any statistically signi?cant earlier spring melt dates for the ice around SH. We are again surprised that this observed fact is not openly stated by Stirling et al. (2008). Here we offer no additional comments other than the caution already expressed in our paper, but a more complete quote from Obbard et al. (2006) may better explain the nature of the data and connection than the re-framing by Stirling et al. (2008): ?For the periods 1984?86 and 2000?2003, we examined inter- annual variability in BCI [Body Condition Index] related to timing of ice melt and to duration of ice cover in the previous winter (Table 3). There was a non-signi?cant negative correlation between BCI value and date (as Julian day) of break-up for both 1984?86 (r ?0.5164, 0.655) and 2000? 03 (r ?0.235, 0.765). Similarly, there was a non? signi?cant negative correlation between BCI and duration of ice cover in the previous winter for 1984?86 (r -0.403, 0.736) and 2000-03 (r ?0.354, 0.646). These results suggest that neither variation in the sea ice break-up date nor duration of ice coverin the previous winter fully explains the variations in BCI among years. This was so despite the 7 Available at fact there is strong evidence of a signi?cant trend towards both later freeze-up and earlier break?up (Gough et al., 2004b), Gagnon and Gough, 2005), and a signi?cant negative trend in body condition when comparing our data from the sampling periods 2 decades apart. These results suggest that other factors or combinations of factors (that likely include a later freeze-up and an earlier break-up) affect body condition in southern Hudson Bay polar bears. One such climatological factor maybe related to unusual spring rain events that occur during March or April when ringed seals are giving birth to pups in on-ice birthing lairs (Stirling and Smith. 2004). These authors documented a case of heavy spring rains that destroyed the roofs of many ringed seal birthing lairs, providing polar bears with easier access to newborn pups. So despite weather conditions that might contribute to an earlier melt of the sea ice (periods of warm daily airternperatures, springrains), polarbears might paradoxically have improved hunting success during these same conditions. Other factors such as depth of snow accumulation and roughness of the ice ?at, stable ice versus rough pressure ice) vary over time and also affect polar bear hunting success (Stirling and Smith, 2004; Ferguson et al., 2005). We note that the multi-variable discussion by Obbard et al. (2006) is, in fact, consistent with our call for a total avoidance in making one-dimensional predictions of how warming spring air temperatures, as Stirling et al. (2008) insist, to solely decide the fate of polar bears. Although Obbard et al. (2006) was not available at the time of the initial submission of our manuscript; nevertheless, their study found that body condition for all age and reproductive classes of polar bears has declined considerably since the mid-19805, although no reference to the SH population size and status was presented to assess possible food competition. However, they also suggest that variation in sea ice break-up or duration of ice cover in the previous winters do not fully explain the bear condition among years. We urge interested readers to consult and study Table 3 of Obbard et al. (2006) to fully verify this fact. Obbard et al. (2006) explain that other meteorological and climate-change factors depth of snow accumulation, roughness of sea ice) could also be important. Although parallels regarding polar bear body condition exist, it is dif?cult to compare results from WH and SH studies due to the different applications of the term ?condition? body mass versus Quetelet Index versus body condition index; Obbard et al., 2006; Derocher and Stirling, 1995b; Stirling et al., 1999). For future monitoring and ease of comparison among polar bear populations, we encourage investigators to use a standard approach in calculating body condition. 3. Responses to suggested importance of human?polar bear interactions 3.1. Research handling of polar bears Stirling et al. (2008) are incorrect in suggesting that et al. (2007) proposed that extensive handling of polar bears in WH ECOLOGICAL COMPLEXITY 5 (2008) 289?302 297 may be responsible for declines in body condition and reproduction. What et al. (2007) suggested was, as we pointed out numerous times previously. that "global warming may indeed have an effect on the polar beats of WH but it must be assessed in a more realistic framework that considers all the likely stress factors and their cumulative impacts? (Dyck et al., 2007: p. 74). In that context, we examined the extensive handling of WH polar bears. Stirling et al. (2008) are correct that the sampling was spread out over a wide area within the WH population boundary, but a survey encompassing all of the area was never conducted until recently (see discussions above in Section 2.3 and below). It is therefore erroneous of Stirling et al. (2008) to oversimplify the calculation of how many bears as proportion of the WH population were captured or handled. One cannot use the 1200 individuals as a base to proceed with such a calculation. Because the complete population area was never surveyed and assuming conserva? tively that approximately 100 bears are occupying the area north of the usual study areas, oneis left with 1100 individuals. Second, catching all individuals during an inventory is virtually impossible (Lancia et al., 1994). If one assumes that about 40% of a population is actually captured during a mark- recapture study, one would end up with about 440 bears that could be captured ideally. Therefore, probability of a bear being caught each year is now 42.5% (187/440), as compared to Stirling et al.'s (2008) 15.6% 187/1200). We also would like to refer the interested reader to Calvert et al. (1995) where, for example, recapture rates for WH bears greater than 1 year of age were found capture sample of 179 bears caught during 1990. This would indicate that bears are recaptured repeatedly over many years, but detailed informa- tion on that topic for WH bears is not readily available. Stirling et al. (2008) are correct in their elaborate recitation of the number of statistical analyses that were performed by Messier (2000) in his study on the effects of polar bear handling. What they fail to mention is that these 3237 bears, captured between 1989 and 1997, came from 7 different polar bear populations Northem Beaufort, Kane Basin, Baf?n Bay, Davis Strait, Viscount Melville Sound, Lancaster Sound, and Norwegian Bay) and data were lumped together in the subsequent analyses (NB: in contrast, 3300 bears were handled during the same time frame forWH alone; Table 3 in Messier (2000)). Also not explained by Stirling et al. (2008) was the de?nition of "long-term", used by Messier (2000). The author de?ned long?term as ?effects de?ned as changes due to handling during the ?rst year following tagging"; Messier (2000: p.18). Most importantly though, polar bear research activities in the Northwest Territories (or what is now Nunavut) occur in about 15-year rotational cycles where individuals of a polar bear population are marked and re- captured for up to 3 years and 'then left alone until the next inventory of the same population simply is not comparable to the annual research efforts to which WH polar bears are exposed. Although we recognize that polar bears are probably affected by short?term handling during these 15-year inventory cycles, true long-term (emphasis added) effects years; resulting from injuries, for example; (Cattet et al., 2006)) are poorly understood, and rarely investigated. Some of the studies that examine the effects of capture of polar bears do not provide a suf?ciently detailed description on how multiple captures of the same individuals over a long-term individual captured in year y, recaptured in year t, etc.) are incorporated into the analyses Derocher and Stirling. 1995b). We commend the authors of Rode et a1. (2007) for their detailed description and analyses of the handling effect on polar bears for the Southern Beaufort population and encourage them to examine data for in the same fashion. We assume that if an examination of these handling effects on all WH polar bears over the long-term are conducted now and if one would ?nd some effects, the results are most likely confounded and a cause and effect could not be established. A study that examined disturbance of pregnant polar bears (co-authored also by the senior author of Stirling et al., 2008) found that the movement of pregnant females (as examined in Ramsay and Stirling, 1986; Amstrup, 1993) "probably was as a direct consequence of being handled and not for other reasons? (Lunn et al., 2004: p. 354). Stirling et al. (2008) selectively neglect to mention that female cubs of handled autumn pregnant females had signi?cantly lighter weights for the same reason. The same study also concludes that ?it is not known what the effects of disturbance of the capture of pregnant females might be in late October or early November, closer to the time when cubs are born" (Lunn et al., 2004: p. 355). Stirling et al. (2008) insinuate that et al. (2007) proposed that females may suffer from handling by being displaced from feeding sites. We ?nd it astonishing that Stirling et al. (2008) seem to refute this thought, although the senior author also co-authored the original paper that suggested this. This was proposed by Ramsay and Stirling (1986) in their study on handling effects of polar bears from WH, which et al. (2007) cited and explained. Stirling et al. (2008) provide a description about lactation and how it is part of the female polar bear's life history. However, they appear to miss the point of et al. (2007) that emphasized that capture work is done either on animals during spring that are emerging from their dens family groups), or during the ice-free period while bears are distributed along the southwestern shore of Hudson Bay? times when bears are either stressed due to lactation (Amould, 1990) or are undergoing a fasting period while living off their stored fat reserves. However, since historical body masses are no longer attained by WH polar bears they have less body mass; Stirling et al., 1999), lactation and fasting, although a result of evolutionary adaptation, has likely become more energetically challenging (or stressful) for polar bears. Stirling and Derocher (1993) and Derocher et al. (2004) provide data for body mass losses, for example, for females as they come off the ice earlier and return later or body mass thresholds for reproduction, respectively, that indicate that these bears will be energetically challenged. 3.2. Tourism The section on tourism in et al. (2007) describes one additional factor to climatic effects that could possibly contribute to the negative energy balance of bears while on land. Stirling et al. (2008) claim that the description of tourism activities at Churchill, as described in et al. (2007), is 293 ECOLOGICAL COMPLEXITY 5 (2008) 289?302 "inaccurate" and that the conclusion is based on an ?incomplete review of available literature" on this topic. They go on by saying that the summary in et (2007) on page 3 (actually p. 75 of et al., 2007) implies that all bears congregate east of Churchill where tourism activities occur. If Stirling et al. (2008) would have read the previous sentence to their selection, it would have been clear that et al. (2007) mentioned that some bears are exposed to these tourism activities and not the complete pupulation. It is unclear how Stirling et al. (2008) derive the 40% of time that bears would spend around tourism activities. Stirling et al. (2008) go on by describing tourism activities at Gordon Point. which were mentioned in (2001) (as cited in and Baydack, 2004, 2006). The estimated number of tourists 6000) in the polar bear viewing area as described by et- al. (2007) are, in fact,-in line with what Lemelin and Wiersma (2007) reported the only difference between the two studies is the de?nition .of a "tourist" and a "visitation"). Stirling et al. (2008) fail to mention the true scope of Lemelin and Wiersma (2007) - the perceptions of polar bear tourists. In their paper, Lemelin and Wiersma (2007) describe that some polar bear viewing tourists were distracted by the vehicular impacts and "un-natural" behaviors of polar bears, and that "the negative aesthetics from tundra vehicle tracks and grey water discharge from the tundra vehicle lodges" in?uenced visitor experience and enjoyment (Lemelin and Wiersma, 2007: p. 49). Stirling et al. (2008) are correct in their claim that baiting has been illegal and that vehicles are only allowed on designated routes, etc. However, at times there is a difference between what "should" happen and what actually happens. During research activities, M. has personally observed tundra vehicles driving off trails, tundra vehicles approaching lying polar bears, and food rewards being given to polar bears. Such an uncontrolled problem can be con?rmed by other independent observers and scientists. The claim by Stirling et a1. (2008) that such behavior is restricted by enforcement may be true, but it still occurred occasionally during Dyck's (2001) study period and has continued to occur thereafter. Stirling et al. (2008) provide data regarding which possible proportion of the WH polar bear population would be exposed to tourism activities and for how long. We feel that these comments and value judgements can be interpreted as a justi?cation for tourism activities to occur, even if they may have negative effects on animals. To our knowledge, there is no published material that indicates that nature or wildlife tourism is adversely affecting, or must be adversely affecting. avertebrate population as a whole. However, the estimated 5% of polar bears that are in contact with tourists. according to Stirling et a1. (2008), could be used as a bench mark and safe- guard to protect wildlife spades from potential harmful tourism operations elsewhere. Stirling et al. (2008) fail to mention, hOWever, that concern about tourism operations at Churchill were expressed during the 12th Working Meeting of the Polar Bear Specialist Group Meeting (Calvert et al., 1998 as cited in et al., 2007). In their overview of the tourism operations at Churchill, Stirling et al. (2008) also fail to mention the helicopter viewing activities that occur during fall concurrently with the tundra vehicle activities (Dyck, 2001). Up to 36 helicopter over-?ights per day occurred during the 2000 polar bear viewing season, where helicopters passed over the viewing area on average every 30 min. These helicopter tours offer visitors a birds-eye perspective of the area and polar bears. Also taking spring viewing of polar bear dens into account, the 5% of bears being in contact with tourism. as estimated by Stirling et al. (2008), seems to be an underestimate. Stirling et al. (2008) discuss the polar bear viewing area size in relation to the area used by WH bears during summer and fall. Their argument seemingly provides another justi?cation for tourism activities because "only about 1.4% of the approximately 2200 km2 land area" is used by bears and ?therefore only a small proportion of polar bear habitat is affected by tourism activities and bears could easily avoid these areas if they chose; in fact more than 95% of the bears do avoid the area?; In part, it is intuitive that not every?in'diVid?ual of the WH polar bear population will be in the small viewing area and natural avoidance might be the case. However, it is also very likely that many bears avoid the area because of human disturbances that occur. Examples of such behavior are listed by Gill et al. (2001), as well as an elaboration of the concept of avoidance versus non-avoidance of disturbance and that species that do not avoid disturbance could be falsely assumed to not. be vulnerable. Even if bears do not avoid tourism activities and they appear "calm", signi?cant internal physiological changes can still occur Wilson et al., 1991). The Scienti?c work on the effects of wildlife tourism and human disturbance on wildlife species has grown during the past 15+ years and several studies demonstrated indirect and direct disturbance effects on populations and individuals Yannoloy et al., 1988; Harrington and Veitch, 1992; Phillips and Alldredge, 2000; Knight and Gutzwiller, 1995; Arno et al., 2006). When faced with a stimulus in the form of a human-caused disturbance, animals make decisions similar to predation avoidance (Frid and Dill, 2002). Bears that are faced with tourism have the option to avoid or tolerate the stimulus and need to make a decision based on costs (Gill et al., 2001). From an individual level, bears that tolerate the stimulus may incur higher costs when avoiding the stimulus than when stayingin the area. In other words, bears that tolerate tundra vehicular stimuli (and helicopter activity; see Dyck, 2001) may be in a different state body condition) as compared to bears that have excess energy to move to another location to avoid these activities. Polar bears are very individualistic and displace- ment can easily occur by any stimulus?and as a result a preferred resting area is abandoned. Currently there is no scienti?c data to support this hypothesis, but studies on other species have found similar results Arno et al., 2006) and we therefore encourage a study that would compare body mass and condition of bears in and outside the bear viewing area. In addition, Inuit livingin Nunavut and outside the usual study area for WH polar bears have commented in recent years that they see more bears on the land during the season when, in the past, few bears were seen, proposing that some WH polar bears have moved north, or the WH population is even increasing (Tyrrell, 2006). Stirling et al. (2008), in their brief elaboration on the study by and Baydack (2004), note that nothing about female polar bear vigilance behavior was mentioned in et al. (2007). and Baydack felt that the data collected from ECOLOGICAL COMPLEXITY 5 (2008) 289?302 299 females were too small, and any inference from that would be too speculative (as was pointed out in and Baydack, 2004: p. 348). We completely agree with Stirling et a1. (2008) that bears have the potential to habituate or adapt to human activity, particularly in circumstances of repeated and pre- dictable human activity (Aumiller and Matt, 1994; Herrera et al., 2005; Smith et al., 2005). During their study, and Baydack (2004, 2006) found that some tundra vehicles would approach and leave the immediate vicinity of viewed bears about every 10 min, creating an environment that would unlikely allow for easy habituation (NB: if vehicles remain stationary it would allow for better habituation, which was pointed out in and Baydack (2004)). It was therefore recommended "that the responsible wildlife management agency take a lead role in developing a visitor and tour operator system that makes tundra vehicle conduct more predictable and consistent for polar bears in the Churchill area? (Dyck and Baydack, 2006: p. 144), which Stirling et al. (2008) neglected to mention, as well in their comments about their "complete" review of the Churchill tourism industry. Although the study results by Rode et a1. (2006) are intriguing and very likely bene?cial to manage visitors in bear viewing areas, comparing visitors on foot in a group of 6 visitors/group on average to 15-18 multi-ton motorized tundra vehicles carrying ~40 passengers each per day through the viewing area in Churchill seems somewhat unwarranted. and Baydack (2004) used a novel, non-invasive tool in the form of vigilance to examine how resting polar bears respond to a human disturbance and how management agencies could examine possible effects of tourism activities on resting polar bears. Some studies on wildlife disturbance documented various physiological and behavioral responses (see and Baydack, 2004 for some detailed references). Another study examined polar bear physiology, in particular, their heart rate and metabolism (Ziritsland et al., 1977). These studies found that heart rate increases when an animal is presented with a stimulus and that the heart rate in polar bears in particular increases from a resting to a lying and to a standing position. If heart rate is indicative of metabolic activity ((Dritsland et al., 1977), then it follows that increases in heart rate would mean increases in metabolic activity. Since and Baydack (2004, 2006) did not perform surgical procedures to implant data?logging systems in the polar bear viewing area at Churchill to measure heart rates and metabolic activity of polar bears, they used deductive reason- ing to come to their conclusions given the published facts on polar bear physiology and animal responses to human disturbances. We are not as con?dent as Stirling et a1. (2008) that the current body of literature is as conclusive on the physiological effects of tourism on large carnivores. 3.3. Polar bear alert program We agree with Stirling et al. (2008) that the Polar Bear Alert Program (PBAP) is important for the safety of Churchill residents (and bears). It is however surprising to see that Stirling et al. (2008) were rather uncritical towards the handling deaths or control kills by the PBAP of?cers. Stirling et al. (2008) indicate that there was an increase in the handling from 48 bears per year between 1969 and 2000 (which et al., 2007 used for their analyses) up to 135 bears per year between 2001 and 2004. This indicates more bears are coming in contact with humans and are being handled, either because more bears move to the surroundings of Churchill or increased efforts by the PBAP, or possibly both. It is likely that several social, rather than strict ecological hungrier bears noted below), factors could have been responsible for the increased handlings but it is surprising to ?nd Stirling and Parkinson (2006) 266] suggest that ?the more likely reason for increasing numbers of polar bears coming into coastal settlements in WH (and hence increased handling by PBAP) is that they are There were observations in the late 19905 that nearly 50% of the bears handled were captured at the Churchill dump?the PBAP of?cers were using the dump as a giant interceptor bait site so that by trapping the bears there less bears would be coming into town. 4. Limitations on polar bear adaptation to projected climate change 4.1. Hunting of species other than ringed seals As discussed in et al. (2007), although some climate models predict a complete disappearance of sea ice over the central Arctic ocean for the month of September, while in contrast, the whole of Hudson Bay is always ice-free during this time, regardless of forcing by anthropogenic greenhouse gases (see for example Figs. 8 and 9 in Johannessen et al., 2004). Moreover, sea ice for Hudson Bay was never predicted to completely disappear for late winter or early spring March). Therefore, polar bears of WH still have sea ice left to hunt seals, even in the highly unlikely event the extreme conditions of climatic warming come true. Contrary to what Stirling et a1. (2008) indicated, however, et al. (2007) proposed that bears may resort more to eating berries and vegetation to overcome some food de?cits when not enough seals were caught or from a shorter hunting season), but did not suggest bears would depend entirely upon vegetation for food. It has been estimated by Clark (1997) that the highest productivity site of berries in the WH Lowlands was about 0.48 kg/ha, which is one to two orders of magnitude lower than for other northern areas where bears are known to feed on berries. There are other food sources like migratory birds and their eggs and caribou (Rangifer app.) that would be necessary to include in a more comprehensive study of potential adaptation strategies. Polar bears are intelligent animals that learn from obser- ving conspeci?cs (Stirling, 1974). This is perhaps how they learned to catch seals during the ice?free season (Fumell and Oolooyuk, 1980) and to capture ?sh by diving (Dyck and Romberg, 2007). It is very likely that berries and vegetation alone will not be suf?cient to "bridge" energetically any extended ice-free season. However, as long as polar bears have a hunting platform throughout the year to hunt seals, a combination of these together with the occasional seal and ?shes may provide as much energy as is lost per day, or possibly more (Dyck, unpublished data), although our view is not shared by all authorities on polar bears since it would 300 ECOLOGICAL COMPLEXITY 5 (2008) 289?302 probably require more data and analyses. Given this potential for learning and acquiring new skills, we believe that individuals of a population can therefore gain suf?cient energy throughout the ice-free period to at least maintain body mass. 4.2. Dependence on signi?cant amounts of terrestrial vegetation We also agree with Stirling et al. (2008) that terrestrial feeding of polar bears occurred at levels that were negligible. However, the cited studies used polar bear samples collected between 1986 and 1991 (Ramsay and Hobson, 1991; Hobson and Stirling, 1997), which was more or less at the onset when decreases in body masses were detected. It would be useful to re?examine this topic with samples from the early 1980 and early 2000s (see Fig. 3 in Stirling and Parkinson, 2005). One could then test whether terrestrial feedingincreased over the entire time frame where decreases in body masses were detected. It should also be pointed out that Stirling and (eritsland (1995) (as cited in Stirling et al., 2008) calculated that polar bears need 43 ringed seals per year to survive. We ?nd this a good approach, but somewhat speculative. The study was based on energetic research done on tread mills in a Churchill facility and therefore not based directly on any ecological ?eld additive fashion. could produce the results that are currently observed in WH polar bears. We emphasize that it would be dif?cult to single-out climate change as the only factor see Krebs and Berteaux, 2006). Contrary to the arguments of Stirling et al. (2008), et al. (2007) did not provide "alternative explanations? for the observed changes. Rather, et al. (2007) provided brief summaries of possible factors human?bear interactions in the form of research, the polar bear alert program, and tourism; food availability and competition) that could, together with the effects of climate change, result in what has been observed in the ecology of polar bears of WH?namely decreased reproduction, decreased body condition of adult male and female polar bears accompanied by dependent young, and a decreased proportion of independent yearling captured during the open water season in summer and autumn (Stirling et al., 1999). Global warming (or climate change) was not the focus of our paper, but rather et al. (2007) questioned whether air temperature at Hudson Bay was actually the ultimate factor causing earlier ice break-up, as was suggested by Stirling et a1. (1999). Acknowledgements We thank Kesten Green, Mitchell Taylor and Martina Tyrrell much more ef?cient tham?ferimportanteoat?bu?onof their expertise and knowledge in the lab studies suggest (Taylor, 2007). We fully agree with Stirling et a1. (2008) that polar bears need meat and protein, aside from berries and vegetation, to maintain body size and population densities. Nunavut, where most of the world?s polar bears occur (Aars et al., 2006), is also rich in other marine mammals that are hunted. Many communities harvest Beluga whale (Delphinapterus leucas), narwhal (Monodon monooeros), and walrus (Odobenus rosmarus). For example, Nunavut hunters harvested approximately 8000 narwhals and 7500 Beluga whales (excluding wounded and lost whales) between 1977 through 2000 (Dyck, 2005). Not all of that meat and fat is uu'lized by the Inuit, and bears can gain easy access to these carcasses during the ice-free period along shorelines. It is not rare to ?nd eight to ten bears of different sex and age classes feeding on a dead walrus (M. Dyck, personal observation from Davis Strait 2006), consuming fat and protein that aid in maintaining their body mass. 4.3. Evolution of ?a true hibernation state? We accept the criticism by Stirling et al. (2008) canceming our previous statement on extended hibernation as a means for polar bear to adapt to climate change but we wish to note that no comprehensive study has yet been conducted on potential adaptations of polar bears to either gradual or abrupt climate and environmental changes. 5. Conclusions The paper by et al. (2007) provides a more realistic or holistic less reductionistic) view of possible interacting factors humans and the environment) that, in an preparing this reply. REFERENCES Inns. 1., Lun, N.J., Derocher. A.E. (Eds), 2006. Polar bears: Proceedings of the 14th Working Meeting of the Polar Bear Specialist Group, IUCN, Gland, Switzerland and Cambridge, UK. Arno, L., Lopez, P., Martin, J., 2006. Nature-based tourism as a form of predation risk affects body condition and health state of Podarcis muralis lizards. Biol. Conserv. 131, 402-409. Amstrup, S.C., 1993. Human disturbances of denning polar bears in Alaska. Arctic 46, 246?250. Amould, 1990. The energetics of lactation in polar bears (Ursus maritimus Phipps). Thesis. University of Saskatchewan, Saskatoon, 80 pp. Aumiller, L.D., Matt, C.A., 1994. Management of McNeil River State Game Sanctuary for viewing of brown bears. Int Conf. Bear Res. Manage. 9, 51?61. Ball, 1995. Historical and instrumental evidence of climate: western Hudson Bay, Canada, 1714?1850. In: Bradley, KS, Jones, P.D. (Eds), Climate Since A. D. 1500. Routledge, pp. 40?73. Berteaux, D., Humphries, M.M., Krebs, C.J., Lima, M., McAdam, A.G., Pettorelli, N., R?ale, D., Saitoh, T., Tkadlec, E., Weladji, R.B., Stenseth. N.Chr., 2006. Constraints to projecting the effects of climate change on mammals. Clim. Res. 32, 151-158. Botkin, D.B., Saxe, H., Arauja, M.B., Betts, IL, Bradshaw, R.H.W., Cedhagen, T., Chesson, P., Dawson, T.R., Etterson, LR, Faith, D.B., Perrier, 5., Guisan, 1L, Hanson, A.S., Hilbert, D.W., Loehle, C., Margles, C., New, M., Sobel, M.J., Stockwell, D.R.B., 2007. Forecasting the effects of global warming on biodiversity. Bioscience 57, 227?236. ECOLOGICAL COMPLEXITY 5 (2008) 289?302 301 Calvert, W., Taylor, M., Stirling, I., Atkinson, A., Ramsay, M.A., Lunn, N.J., Obbard, M., Elliott, C., Lamontagne, G., Schaefer, 1., 1998. Research on polar bears in Canada 1993?1996. In: Derocher, A.E., Gamer, G.W., Wiig, (El. (Eds), Polar Bears: Proceedings of the 12th Working Meeting of the Polar Bear Specialist Group, IUCN, Gland, Switzerland and Cambridge, UK, pp. 69?91. Calvert, W., Stirling, 1., Taylor. M., Lee, L.J., Kolenosky, G.B., Kearney, S., Crete, M., Smith, 3., Luttich, S., 1991. Research on polar bears in Canada 1985-87. In: Amstrup, E.C., Wiig. Polar Bears: Proceedings of the 10th Working Meeting of the Polar Bear Specialist Group, IUCN, Gland, Switzerland, pp. 11?24. Calvert, W., Stirling, L, Taylor, M., Ramsay, M.A., Kolenosky, G.B., Crete, M., Kearney, 5., Luttich, S., 1995. Research on polar bears in Canada 1988?92. In: Wiig, (25., Born, E.W., Gamer, G.W. (Eds), Polar Bears: Proceedings of the 11th Working Meeting of the Polar Bear Specialist Group, IUCN, Gland, Switzerland and Cambridge, UK, pp. 33?59. Catchpole, AJ.W., 1995. Hudson?s Bay Company ships' log-books as sources of sea ice data, 1751?1870. In: Bradley, P.D. (Eds), Climate Since A. D. 1500. Routledge, pp. 17?39. Carter, M.R.L., Bourque, A., Elkin, Powley, K.D., D.B., Caulkett, N.A., 2006. Evaluation of the potential for injury with remote drug-delivery systems. Wildl. Soc. Bull. 34, 741-749. 7* Clark, D., 1997. Terrestrial habitat selection by polar bears (Ursus maritimus Phipps) in the Western Hudson Bay Lowlands. Thesis. Univ. 0f Alberta, Edmonton, AB. Cohen, 1., Barlow, M., 2005. The NAO, the A0 and global warming: how closely related? 1. Clim. 18, 4498?4513. Comiso, J.C., 2003. Warming trends in the Arctic from clear sky satellite observations. 1. Clim. 16, 3498?3510. Comiso, J.C., 2006. Arctic warming?signals from satellite observations. Weather 61, 70?76. Derocher, A.E., Lunn, Stirling, 1., 2004. Polar bears in a climate. Integr. Comp. Biol. 44, 163-176. Derocher, A.E., Stirling, L, 1992. The population dynamics of polar bears in western Hudson Bay. In: McCullough, D.R., Barrett, R.H. (Eds), Wildlife 2001: Populations. Elsevier Science Publishers, pp. 1150?1159. Derocher, A.E., Stirling, I., 19953. Estimation of polar bear population size and survival in western Hudson Bay. J. Wildl. Manage. 59, 215?221. Derocher, A.E., Stirling, L, 1995b. Temporal variation in reproduction and body mass of polar bears in western Hudson Bay. Can. J. Zool. 73, 1657?1665. Dyck, M.G., 2001. Effects of tundra vehicle activity on polar bears (Ursus maritimus) at Churchill, Manitoba, Canada. MNRM Thesis. University of Manitoba, Winnipeg. Dyck. M.G., 2006. Characteristics of polar bears killed in defense of life and property in Nunavut. Canada, 1970?2000. Ursus 17, 52?62. Dyck, M.G., Romberg, S., 2007. Observations of a wild polar bear (Ursus maritimus) successfully ?shing Arctic chart (Saluelinus alpinus) and I-?ourhom sculpin (Myoxocephalus quadricornis). Polar Biol, Dyck, M.G., Baydack, R.K., 2006. Human activities associated with polar bear viewing near Churchill, Manitoba, Canada. Hum. Dimensions Wildl. 11, 143?145. Dyck. M.G., Baydack, R.K., 2004. Vigilance behaviour of polar bears (Ursus maritimus) in the context of wildlife-viewing activities at Churchill, Manitoba, Canada. Biol. Conserv. 116, 343?350. Dyck, M.G., Soon, W., Baydack, R.K., Legates, D.R., Baliunas, 5., Ball, Hancock, L.O., 2007. Polar bears of western Hudson Bay and climate change: are warming spring air temperatures the ?ultimate? survival control factor? Ecol. Complexity 4, 7384, Ferguson, Stirling, I., McLoughlin, P.. 2005. Climate change and ringed seal (Phoca hispida) recruitment in Western Hudson Bay. Mar. Mammal Sci. 21, 121?135. Frid, A., Dill, L., 2002. Human-caused disturbance stimuli as a form of predation risk. Cons. Ecol. 6, 11. Fumell, D.I., Oolooyuk, D., 1980. Polar bear predation on ringed seals in ice-free water. Can. Field-Nat. 94, 88?89. Gagnon, A.S., Gough, W.A., 2005. Trends in the dates of freeze-up and breakup over Hudson Bay, Canada. Arctic 58, 370?382. Gagnon, A.S., Gough, W.A., 2006. East-west asymmetry in long- terrn trends of landfast ice thickness in the Hudson Bay region, Canada. Clim. Res. 32, 177?186. Gill, Norris, K., Sutherland, W., 2001. Why behavioural responses may not re?ect the population consequences of human disturbance. Biol. Conserv. 97, 265?268. Gough, WA, Gagnon, A.S., Lau, H.P., 2004a. Interannual variability of Hudson Bay ice thickness. Polar Geogr. 28, 222?238. Cough, W.A., Comwell, A.R., Tsuji, LJ.S., 2004b. Trends in seasonal sea ice duration in Southwestern Hudson Bay. Arctic 57, 298?304. Harrington, F.H., Veitch, AM, 1992. Calving success of woodland caribou exposed to low-level jet ?ghter over?ights. Arctic 45, 213?218. Herrero, 5., Smith, T., DeBruyn, T.D., Gunter. K., Matt, C.A., 2005. From the ?eld: brown bear habituation and people-safety, risks, and bene?ts. Wildl. Soc. Bull. 33, 1?12. Hobson, K.A., Stirling, I., 1997. Low variation in blood 13C among Hudson Bay polar bears: implications for metabolism and tracing terrestrial foraging. Mar. Mammal Sci. 13, 359?367. Johannessen, O.M., Miles, M.W.. Kuzmina, S.L., Semenov, V.A., Alekseev, G.V., Nagumyi, A.P., Zakhorov. Bobylev, L.P., Petterson, L.H., Hasselman, K., Cattle, P., 2004. Arctic climate change: observed and modeled temperature and sea-variability. Tellus 56A, 328?341 (plus Conigendum in Tellus 56A, 559?560). Knight, R.L, Gutzwiller, K.J., 1995. Wildlife and Recreationists: Coexistence through Management and Research. Island Press, Washington, DC. Krebs, C.J., Berteaux, D., 2006. Problems and pitfalls in relating climate variability to population dynamics. Clim. Res. 32, 143-149. Lancia, R.A., Nichols, J.D., Pollock, K.H., 1994. Estimating the number of animals in wildlife populations. In: Bookhout, T.A. Research and Management Techniques for Wildlife and Habitats. Fifth edition. The Wildlife Society, Bethesda, Maryland, pp. 215?253. Lernelin, R.H., Wiersma, E.C., 2007. Perceptions of polar bear tourists: a qualitative analysis. Hum. Dimensions Wildlife 12, 45?52. Lunn, NJ., Stirling, 1., Andriashek, D., Richardson, E., 2004. Selection of maternity dens by female polar bears in western Hudson Bay, Canada and the effects of human disturbance. Polar Biol. 27, 350?356. Lunn, Stirling, I., Andriashek, D., Kolenosky, 6.3., 1997. Re- estimating the size of the polar bear population in western Hudson Bay. Arctic 50, 234?240. Mayor, S.J., Schaefer, J.A., 2005. Many faces of population density. Oecologia 145, 276?281. Messier, P., 2000. Effects of capturing, tagging, and radio- collaring polar bears for research and management purposes in Nunavut and NWT. Department of Biology, University of Saskatchewan, Saskatoon, SK, pp. 64. Obbard, M.E., Cattet, M.R.I.., Maddy, T., Walton, L.R., Potter, D., Inglis, I, Chenier, C., 2006. Temporal trends in the body condition of southern Hudson Bay polar bears. Climate Change Research Information Note Available from http:// sit.mnr.gov.on.ca 3, 1?8. 302 ECOLOGICAL COMPLEXITY 5 (2008) 289?302 (eritsland, N.A., Stallman, Jonkel, CJ., 1977. Polar bears: heart activity during rest and exercise. Comp. Biochem. Physiol. 57 (1A), 139?141. Overland, I.E., Wang, M., 2005. The Arctic climate paradox: the recent decreases of the Arctic Oscillation. Geophys. Res. Lett. 32 20046L021752. Phillips, Alldredge, A.W., 2000. Reproductive success of elk following disturbance by humans during calving season. J. Wildl. Manage. 64, 521?530. Ramsay, M.A., Hobson, K.A., 1991. Polar bears make little use of terrestrial food webs: evidence from stable-carbon isotope analysis. Oecologia 86, 598?600. Ramsay, M.A., Stirling, I., 1986. Long-term effects of drugging and handling free-ranging polar bears. 1. Wild]. Manage. 50, 619?626. Ray, C.. Hastings, A., 1996. Density dependence: are we searching at the wrong spatial scale? J. Anim. Ecol. 65, 556?566. Regehr, E.V., Lunn, NJ., Amstrup, S.C., Stirling, L, 2007. Effects of earlier sea iCe breakup on survival and population size of polar bears in western Hudson Bay. J. Wild]. Manage. 71, 2673?2683. Rigor, LG., Wallace, I.M.. 2004. Variations in the age of Arctic sea-ice and summer sea-ice extent. Geophys. Res. Lett. 31 20046L019492. Rode, K.D., Amstrup, S.C., Regehr, E.V., 2007. Polar bears in the Southern Beaufort Sea stature, mass, and cub recruitment in relationship to time and sea ice extent between 1982 and 2006. USGS Administrative Report. Alaska Science Center, Anchorage, AK. Rode, K.D., Farley, S.D., Robbins, C.T., 2006. Behavioral responses of brown bears mediate nutritional effects of experimentally introduced tourism. Biol. Conserv. 133, 70-30. Rosing?Asvid, A., 2006. The in?uence of climate variability on polar bear (Ursus maritimus) and ringed seals (Pusa hispida) population dynamics. Can. J. Zool. 84, 357?364. Saucier, F.J., Senneville, S., Prinsenberg, S., Roy, F., Smith, G., Gachon, P., Caya, D., Laprise, R., 2004. Modeling the sea ice- ocean seasonal cycle in Hudson Bay, Foxe Basin and Hudson Strait. Canada. Clim. Dyn. 23, 303?326. Scott, P.A., Stirling, L, 2002. Chronology of terrestrial den use by polar bears in western Hudson Bay as indicated by tree growth anomalies. Arctic 55, 151?166. Serreze, M.C., Francis, LA, 2006. The Arctic on the fast track of change. Weather 61, 65-69. Serreze, M.C., Holland, M.M., Stroeve, 1., 2007. Perspectives on the Arctic's shrinking sea?ice cover. Science 315, 1533?1536. Shimada, K., Kamoshida, T.. Itoh, M., Nishino, S., Carmack, E., McLaughlin, F., Zimmermann. S., Proshutinsky, A., 2006. Paci?c Ocean in?ow: in?uence on catastrophic reduction of sea ice cover in the Arctic ocean. Geophys. Res. Lett. 33 2OOSGL025624. Skinner, W.R., Jefferies, R.L.. Carleton, T.J., Rockwell, RR, Abraham, 1998. Prediction of reproductive success and failure in lesser snow geese based on early season climatic variables. Glob. Change Biol. 4, 3?16. Smith. T.S., Herrera, S., DeBruyn, T.D., 2005. Alaskan brown bears: habituation and humans. Ursus 16, 1?10. Stirling, I., 1974. Midsummer observations on the behaviour of wild polar bears (Ursus maritimus). Can. J. Zool. 52, 1191? 1198. Stirling, 1., Derocher, A.E., 1993. Possible impacts of climatic warming on polar bears. Arctic 46, 240?245. Stirling, 1., Derocher, A.E., Gough, W.A., Rode, K., 2008. Response to et al. (2007) on polar bears and climate change in western Hudson Bay. Ecol. Complexity 5, 193?201. Stirling, 1., Jonkel, C., Smith, P., Robertson, R., Cross, D., 1977. The ecology of the polar bear (Ursus maritimus) along the western shore of Hudson Bay. Can. Wild]. Serv. Occas. Pap. No. 33. Stirling, I., Lunn, N.J., Iacozza, 1., Elliott, Obbard, M., 2004. Polar bear distribution and abundance on the southwestern Hudson Bay coast during open water season, in relation to population trends and annual ice patterns. Arctic 57, 15-26. Stirling, 1., Lunn, NJ., Iacozza, 1., 1999. Long-term trends in the populatiOn ecology of polar bears in western Hudson Bay in relation to climate change. Arctic 52, 294?306. Stirling, I., @?tsland, N.A., 1995. Relationships between estimates of ringed seal (Phoca hispida) and polar bear (Ursus maritimus) populations in the Canadian Arctic. Can. J. Fish. Aquat. Sci. 52, 2594?2612. Stirling, 1., Parkinson, C.L., 2006. Possible effects of climate warming on selected populations of polar bears (Ursus maritimus) in the Canadian Arctic. Arctic 59, 261?275. Stirling, 1., Smith, T.G., 2004. Implications of warm temperatures and an unusual rain event for the survival of ringed seals on the coast of southeastern Baf?n Island. Arctic 57, 59?67. Taylor, M., 2007. Signed open communication to W.S. and MD. on August 8, 2007 (available from Willie Soon wsoon?cfaharvardedu upon request). Taylor, M.K., Akeeagok, S., Andriashek, D., Barbour, W., Born, E.W., Calvert, W., Cluff, H.D., Ferguson, 8., Laake, 1., Rosing- Asvid, A., Stirling, Messier, R, 2001. Delineating Canadian and Greenland polar bear (Ursus maritimus) populations by cluster analysis of movements. Can. J. Zool. 79, 690?709. Taylor, M.. Lee, L.J., 1994. Tetracycline as a biomarker for polar bears. Wild]. Soc. Bull. 22, 83?89. Tyrrell, M., 2006. More bears, less bears: inuit and scienti?c perceptions of polar bear populations on the west coast of Hudson Bay. Etudes/Inuit/Studies 30, 191?208. Wilson, R.P., Culik, B., Dan?eld, R., Adelung, D., 1991. People in Antarctica?how much do Ad?lie Penguins Pygoscelis adeliae care? Polar Biol. 11, 363?370. Yarrnoloy, C., Bayer, M., Geist, V., 1988. Behaviour responses and reproduction of mule deer, Odocoileus hemionus, does following experimental harassment with an all-terrain vehicle. Can. Field-Nat. 102, 425?429. ATTACHMENT 2356 JOURNAL OF CLIMATE Centennial Variations of the Global Monsoon Precipitation in the Last Millennium: Results from ECHO-G Model EDUARDO *Smre Key Laboratory of Lake Science and Environment. Nanjing Institute of Geography and Limnotogy. Academy afSciences, Nanjing. China (TPEU. Ut?ean University of China. Qingdao. China i'iarvtn'ri-Smithsonian Center for Astrophysics. Cambridge. tirinssuchnsetts Institute for Carma! Rtl'i'?dl?li'il. GEES Return-cit Center. {intuition-in, {St-nanny (Manuscript received 29 November 2007, in ?nal form 3 September 2008) AB STRA CT The authors investigate how the global monsoon (GM) precipitation responds to the external and an- thropogenic forcing in the last millennium by analyzing a pair of control and forced millennium simulations with the ECI-IAM and the global Hamburg Ocean Primitive Equation (ECHO-G) coupled ocean?atmosphere model. The forced run, which includes the solar, volcanic, and greenhouse gas forcing, captures the major modes of precipitation climatology comparably well when contrasted with those captured by the NCEP reanalysis. The strength of the modeled GM precipitation in the forced run exhibits a signi?cant quasi- bicentennial oscillation. Over the past 1000 yr, the simulated GM precipitation was weak during the Little Ice Age (1450?1850) with the three weakest periods occurring around 1460. 1685. and 1800, which fell in. respectively, the Spb?rer Minimum. Maunder Minimum, and Dalton Minimum periods of solar activity. Conversely, strong GM was simulated during the model Medieval Warm Period (ca. 1030?1240). Before the industrial period, the natural variations in the total amount of effective solar radiative forcing reinforce the thermal contrasts both between the ocean and continent and between the Northern and Southern Hemispheres resulting in the millennium-scale variation and the quasi-bicentennial oscillation in the GM index. The prominent upward trend in the GM precipitation occurring in the last century and the notable strengthening of the global monsoon in the last 30 yr (1961-90) appear unprecedented and are due possibly in part to the increase of atmospheric carbon dioxide concentration, though the authors? simulations of the effects from recent warming may be overestimated without considering the negatiVe feedbacks from aero- sols. The simulated change of GM in the last 30 yr has a spatial pattern that differs from that during the Medieval Warm Period. suggesting that global wanning that arises from the increases of greenhouse gases and the input solar forcing may have different effects on the characteristics of GM precipitation. It is further noted that GM strength has good relational coherence with the temperature difference between the Northern and Southern Hemispheres, and that on centennial time scales the GM strength responds more directly to the effective solar forcing than the concurrent forced response in global-mean surface temperature. VOLUME 22 JIAN Liu,* BIN QINGHUA XUEYUAN WILLIE AND Department of Meteorology. and intt-nirnirnmi Paci?c Research Center. Unite-min- of Hawaii at Manoa, Honolulu. Hawaii 1. Introduction Monsoon climate varies on several characteristic time scales in addition to ?uctuations of random origin. In the last two decades, signi?cant progress has been made in the study of monsoon variability on intraseasonal, Corresponding author address: Dr. J. Liu, State Key Laboratory of Lake Science and Environment. Nanjing Institute of Geography and Limnology, Chinese Academy of Sciences. 73 East Beijing Road, Nanjing 210008, China. E-mail: jianliu@niglas.ac.cn DO I: 10.117512008JCL12353J 2009 American Meteorological Society interannual, interdecadal, and orbital time scales of tens of thousands of years. A review of recent progresses in the understanding of Asian monsoon variability has been summarized in Wang (2006). However, the centennial- scale variability of the global monsoon system has been considerably less studied. On time scale of centuries, the internal feedback processes that control the interannual- to interdecadal-scale variations become less important; in the meantime, the persistent, external forcing from centennial-scale variations in solar radiation, when viewed from the perspective of local and regional spatial domains. 1 MAY 2009 may be considerably stronger and more effective than the effect of internal feedback. Both the nature and cause of the centennial-scale variability are largely not known and understudied. In contrast to numerous studies on the global-mean temperature, we focus on global monsoon precipitation in this study. Why are we particularly concerned with the global monsoon precipitation? The monsoon rain- fall provides water resources to about two-thirds of the world?s population, so any improved knowledge on its variation will be of great societal importance. Monsoon precipitation is also of scienti?c importance as it is the key variable of the global water cycle and it provides a critical heat source for driving atmospheric circulation- What is the global monsoon? Previous studies have mostly focused on the monsoon changes in speci?c re? gions because of considerable regionality of the mon- soon variations. Yet regional monsoons are coordinated by the same annual cycle of the solar forcing and their variations are interrelated- Trenberth et al. (2000) ar- gued that the conservation of atmospheric mass, mois? ture, and energy spread over global domain with ex- changes in the lower boundary. They have de?ned the global monsoon system as a global-scale overturning circulation that varies annually. Wang (1994) ?rst at- tempted to delineate the global monsoon regime using outgoing longwave radiation data as the proxy for deep LIU ET AL. 2357 Hamburg Ocean Primitive Equation model (Rodgers et 2004; Min et al. 2005a,b; Wagner et al. 2005), and Meteorological Research Institute Coupled General Circulation Model, version 2.2 (MRI CGCM2.2) model (Kitoh 2006) for control simulations; and the Climate and Biosphere Model model (Bauer et al. 2003), ECHO-G model (Zorita et al. 2003; Gonzalez?Rouco et al. 2003; Zorita et al. 2005; Gouirand et al. 2007a,b), ocean?atmosphere?sea ice model of intermediate complexity de Bilt?coupled large- scale ice?ocean model (Goosse et al. 2005). Model for the Assessment of Greenhouse-Gas Induced Climate Change (MAGICC) model (Osborn et 2006), Isopycnal Model (OPYCS: Stendel et al. 2006), third climate con?guration of the Met Of?ce Uni?ed Model Tett et al. 2007), and National Center for Atmospheric Research Cli- mate System Model (NCAR Mann et al. 2005b; Ammann et a1. 2007) for forced simulations of the past millennium or past half millennium. Based on these model results, the internal variability of temperature, precipitation, mean sea level pressure, and major modes of climate variation such as the North Atlantic Oscillation (NAO), ENSO, and Indian mon? soon have been discussed (Min et al. 2005a,b; Zorita et al. 2003; Kitoh 2006). Variabilities on decadal, multidecadal, and centennial time scales under convection and and anthropogenic torc1ng have also been ex- further demarcated a global monsoon (GM) precipita- tion domain based on characteristics of monsoon pre- cipitation and examined the trends of the GM rainfall -over [and using four sets of rain gauge precipitation datasets compiled for the period 1948?2003. But how the GM changes on multidecadal to millennial time scales and what mechanisms are responsible for them remain poorly studied. One of the major roadblocks for studying GM vari- ability on centennial or millennial scale is the lack of direct observations, especially on the global scale. To make progress, we consider numerical simulations with atmosphere?ocean coupled climate models for the last millennium. These simulations may provide useful sur- rogate datasets for analyzing and understanding any changes of GM precipitation on multidecadal to cen- tennial time scales. These kinds ?of simulations have been constructed using a wide variety of climate models with different levels of complexity, including the second climate con?guration of the Met Of?ce Uni?ed Model Johns et al. 1997), Geophysical Fluid Dy- namics Laboratory (GFDL) model (Manabe and Stonffer 1993; Stouffer et al. 2000), Commonwealth Sci- enti?c and Industrial Research Organisation (CSIRO) model (Vimont et al. 2002), ECHAM and the global amined in terms of surface and subsurface temperatures (Stouffer et al. 2000; Beltrami et al. 2006; Gonzalez? Rouco et al. 2003; Zorita et al. 2005; Wagner et al. 2005; Osborn et al. 2006). Both model?model comparison (Goosse et al. 2005; Osborn et al. 2006) and model?data comparison (Liu et al. 2005; Beltrami et al. 2006) among convenient indexes like SST, ENSO, NAO, and Arctic Oscillation Vimont et a1. 2002; Rodgers et al. 2004; Goosse et al. 2005; Gouirand et al. 2007a,b) have been performed in order to assess the reality and ro- bustness of models? simulations. Roles of centennial- scale variations in solar activity (Wagner et al. 2001; Fleitmann et al. 2003; Weber et al. 2004; Wiles et al. 2004; Holzkamper et al. 2004; Delmonte et al. 2005; Lim et al. 2005; Wang et al. 2005; Haltia?Hovi et al. 2007) and changes between volcanic pulse forcing (Crowley 2000; Goosse and Renssen 2004; Mann et al. 2005a) have also been recently investigated and argued through a combination of paleo-proxy data analyses and climate modelings. But how does the GM rainfall respond to natural and anthropogenic forcing in the last millennium? How do the spatial patterns of GM precipitation anomalies differ between the warm and cold periods in the millennium? How do the changes of GM rainfall relate to the changes 2358 of global temperature and interhemispheric temperature difference? These open questions will be explored here. To study these questions, we ?rst explore the perfor- mance of climate model (ECHO-G) in simulating annual mean and annual cycle of the precipitation in the global tropics and subtropics (section next we de?ne the global monsoon indexes (GMI) for the variability of global monsoon precipitation and its centennial-to-millennial- scale variability (section the simulation clearly yielded the mock-up versions of the Medieval Warm Period (MWP), Little Ice Age (LIA), and Present Warm Period durations in the model will be de?ned later. We will then compare the spatial struc? ture of GM precipitation during the MWP, LIA, and (section 4), and ?nally we discuss the forcing-response relationships between GMI and radiative forcings from changes in solar activity, volcanic eruptions, as well as atmospheric and CH4 concentration (section 5). Section 6 summarizes the main results of this paper- 2. Model and its validation a. Model and simulation The ECHO-G climate model (Legutke and Voss 1999) consists of the spectral atmospheric model ECHAM4 (Roeckner et al. 1996) and the Hamburg Ocean Prim- itive Equation global (HOPE-G) model (Wolff et al. 1997), both developed at the Max Planck Institute for Meteorology in Hamburg. The ECHAM4 is based on primitive equations with a mixed p?o coordinate system. The model con?guration used for these simulations has 19 vertical levels in the atmosphere and 20 levels in the ocean, and horizontal resolutions are approximately 3.75? (atmosphere) and 28? (ocean) in both latitudes and longitudes. The ocean model HOPE-G has a grid re?nement in the tropical regions, where the meridional grid point separation reaches 05". To enable the cou- pled model to sustain a simulated climate near to the real present-day climate with minimal drift, both heat and freshwater ?uxes between the atmosphere and ocean are modi?ed by adding a constant (in time) ?eld of adjustment with net-zero spatial average (Roeckner et al. 1996; Wolff et al. 1997). Two millennial integrations with the ECHO-G model will be analyzed here- One is the 1000-yr control simu? lation (CTL), which was generated using ?xed external (annually cycling) forcing set 'to the present?day values (Zorita et al. 2003). This CTL experiment can simulate annual-to-decadal climate oscillations through the in- ternal dynamics of the coupled climate system (Min et al. 2005a,b). The second simulation, named ERIK (covering the period 1000?1990), is forced by three ex- ternal forcing factors (Gonzalez-Rouco et al. 2003; von JOURNAL OF CLIMATE VOLUME 22 Storch et al. 2004; Zorita et al. 2005): solar variability (Crowley 2000; von Storch et 2004), greenhouse gas concentrations in the atmosphere including C03 and CH4 (Blunier et al. 1995; Etheridge et al. 1996). and the effective radiative effects from stratospheric volcanic aerosols (Crowley 2000) for the period AD 1000?1990. The volcanic forcing is parameterized in this simulation as a simple reduction of the annual-mean solar constant. starting in the year with a volcanic eruption and usually lasting a couple of years, according to the reconstruc- tions of volcanic aerosol forcing (Crowley 2000). This second experiment includes the major natural and anthro- pogenic forcings in the past millennium, but a number of other potentially important forcings anthropogenic tropospheric sulfate aerosols, the effect of land-use and vegetation changes, and some other greenhouse gases, such as halocarbons and ozone) were excluded in the ERIK experiment. We keep in mind that the neglected anthropogenic factors may have a signi?cant impact on the climate in the twentieth century. For example, sul- fate aerosols exert a negative temperature forcing. Ne- glect of the tropospheric sulfate aerosols in the ERIK simulation, therefore, excluded their cooling in?uences on temperature, leading to excessive warming in the last 30?50 yr of the twentieth century. The initial conditions of the ERIK simulation were taken from year 100 of the control run. Those initial conditions are, however, representative of present-day rather than preindustrial climate and the experimental design therefore included a 30?yr adjustment period during which the control run forcing was linearly re- duced until it matched the forcing imposed around AD 1000, followed by a 50-yr period with ?xed forcing to allow the model?s climate to readjust to the modified forcing. The ERIK simulation then proceeded from the conditions at AD 1000 to AD 1990. Note that uncer- tainties remained unresolved with the speci?cation of initial conditions, and that the uncertainty in the initial conditions, in turn, might cause initial climate drift that could potentially in?uence the relationship between applied forcing and simulated response. This issue has been recognized by the authors of the simulation and has been addressed in the comment published in the Journal of Climate (Fig. 1 of Zorita et al. 2007). 1). Validation of the model annual-mean. precipitation and annual cycle of precipitation To assess the performance of ECHO-G in modeling precipitation climatology, we examine the annual?mean precipitation and the leading EOF mode of the mean annual cycle. The leading EOF mode accounts for about 70% of the total annual variance and its spa- tial pattern can be faithfully represented by the une? I MAY 2009 September (JJ AS) minus December?March (DJF M) precipitation (Wang and Ding 2008). This solstice mode captures the major portion of the global monsoon; for sim? plicity, it will be referred to as the global monsoon mode- Figure 1 compares the spatial patterns of the annual mean (Fig. 1a) and global monsoon (Fig. 1b) derived from the Climate Prediction Center (CPC) Merged Analysis of Precipitation Xie and Arkin 1997), ERIK run, and National Centers for Envi- ronmental Prediction (N Reanalysis (Kanamitsu et al. 2002). The CMAP data are considered here as the observed ?ground truth.? Pattern correlation coef- ?cients (PCC) and root-mean?square errors (RMSE) are used to gauge the model performance. The FCC of annual-mean precipitation between ERIK and CMAP (0.87) is lower than that between NCEP-2 Re- analysis and CMAP (0.89). The pattern correlation of the global monsoon mode between ERIK and CMAP (0.81) is also lower than that between Reanalysis and CMAP (0.85). However, the RMSE of the annual-mean precipitation and global monsoon mode between ECHO-G ERIK and CMAP are 1.09 mm day" and 1.88 mm day?, which are both smaller than those between NCEP-2 Reanalysis and CMAP data (1.16 and 2.02 mm day"). These results suggest that the simulated precipitation climatology in ERIK run is comparable to those assimilated data in Re- analysis. This overall agreement adds con?dence to our subsequent analysis of the centennial-scale precipitation variability using the outputs generated by the ERIK run. While the ERIK run simulates the rainfall climatole ogy realistically, biases do exist. It can be seen from Fig. 1a that the annual-mean precipitation has signi?- cant errors in the Asian monsoon region, subtropical South Paci?c convergence zone, and South Atlantic convergence zone. The monsoonal mode has signi?cant errors in the East Asian subtropical monsoon and the Mexican?North American monsoon regions. However, the overall model results on global scale are realistic and adequate for our study of the long-term modulations of the global monsoon system by externally imposed nat- ural and anthropogenic forcings. 3. Temporal variation of the global monsoon precipitation (1. De?ning the GM domain and indexes The GM is the dominant mode of the annual cycle of the global tropical circulation (Wang and Ding 2008). To analyze the Spatiotemporal variation of GM, the global monsoon precipitation domain and the global monsoon precipitation strength are de?ned here fol- lowing Wang and Ding (2008). Monsoonal climate is not LIU ET 2359 only characterized by annual reversal of surface winds but also by a contrasting wet summers and dry winters (Webster 1987). The global monsoon precipitation do- main is de?ned by the region in which the annual range (AR) of precipitation exceeds 2 mm day?l and the local summer precipitation exceeds 55% of annual rainfall. Here AR is de?ned as MJJAS precipitation minus precipitation in the Northern Hemisphere (NH) and minus MJJAS precipitation in the Southern Hemisphere (SH). Figure 2 shows the global monsoon domain de?ned by CMAP data, which consists of 6 major monsoon regions: the Northern African (N1), Southern African (SI). Asian (N2), Australian (32), North American (N3), and South American (S4) monsoon. Note that all these ma- jor regional monsoons involve continent?ocean con- trast. There is a minor region in the central South Paci?c (83), Which does not involve land?ocean thermal con? trast and is a ?pure? oceanic monsoon-like region. In the following analysis, one will ?nd that the behavior in 53 is different from those in the other major continent? ocean monsoon regions. Therefore, we exclude S3 re- gion in the aggregate measure of the intensities for the global monsoon and Southern Hemisphere monsoon. The monsoon strength can be represented by the annual range of the total monsoon precipitation. Since the annual range is largely controlled by the local sum- mer precipitation, an alternative measure of the mon- soon strength is simply the local snmmer monsoon rain- fall. Thus, a NH monsoon index (NI-IMI) is de?ned as the HA rainfall falling in the observed NH monsoon domain including both land and ocean; a SH monsoon index (SHMI) is de?ned as the DJF rainfall falling in the ob- served SH monsoon domain, which includes southern African, Anstralia, and South America but not the cen- tral South Paci?c. Here the DJF is the season following the NH JJA. As such, the two hemispheric monsoon indexes measure the strength of the aggregated NH and SH monsoon domains, respectively. To quantify the strength of the global monsoon, we de?ne the global monsoon index as the sum of the two hemispheric in- dexes; that is, GMI NHMI SHMI. The GMI is a measure of the GM strength in terms of the precipitation within the global monsoon rainy do- main. Under climate change, the annual cycle of the tropical circulation is expected to change. The GM strength can faithfully re?ect this change as it represents the dominant mode of the tropical circulation. b. Centennial and millennial variations of the global monsoon indexes Figure 3 shows the time series of the 7- and 31~yr running means of the monsoon indexes (NHMI, SHML 2360 (0) Annual mean CMAP eon g_ JOURNAL OF CLIMATE VOLUME 22 Global monsoon mode CMAP ERIK 15w mm/day i?hE i?u tabs rnrn/day a FIG. 1. Comparison of climatology of global precipitation (mm day?): annual mean and global monsoon mode (JJAS minus DJFM: leading EOF mode of annual cycle) for (top) CMAP. (middle) ECHO-G ERIK run, and (bottom) NCEP-Z Reanalysis. CMAP and NCEP-2 Reanalysis climatological data were derived for the period 1979?2004. ERIK 25?yr climatology was derived for the period AD 1965?90. The numbers shown in the upper-left corners and the lower-left comers indicate pattern correlation coef?cients and RMSES with the CMAP data. respectively. GMI) for both the CTL (Fig. 3a) and ERIK (Fig. 3b) runs. The control run provides an opportunity to ex- amine the internal variability due to unforced feedback processes in the coupled system. In the control run, there is no trend in NHMI, SHMI, and GMI. Further- more, the NHMI and SHMI are not related, as evi? denced by the correlation coef?cients between them valued at about zero for both the 7? and 31-y1' running- mean series. This result suggests a lack of coherency of the monsoon intensity between the two hemispheres without external forcing- From the observed data for the last 56 yr, it was found that the correlation coe?i? cients between NHMI and SHMI are also very low us? ing either the raw data or the 7-yr running-mean data (Wang and Ding 2006). But the data records are simply too short to con?rm or reject any particular hypothesis concerning relation between monsoon indexes of the two hemispheres. In the ERIK run (Fig. 3b) on the other hand. the simulated, forced responses on multidecadal scales and longer are quite different. Signi?cant centennial varia- tions can be seen. These variations correspond to the evolution of the global-mean temperature. where a model MWP, LIA, and PWP can be recognized (Zorita I MAY 2009 Lil) ET 2361 Global monsoon precipitation domain 60mm I ?r - 5?30 stir 120E 1&0 120i! saw a FIG. 2. The global monsoon precipitation domain de?ned by the region in which the AR of precipitation exceeds 2mm day?I and the local summer precipitation exceeds 55% of annual rainfall by using CMAP data. at al. 2005). According to the model simulation, strong global monsoon signal is observed around 1030?1240, which is de?ned here as the model Medieval Warm Period with three distinguished peaks around 1050, 1140, and 1200 (Fig. 3b). On the other hand, weak global monsoons are observed during the model Little Ice Age period from 1450 to 1850. It is of particular interest to ?nd that during the Little Ice Age the GM strength exhibits three minima, which occur around 1460, 1685, and 1800 (Fig. 3b). These rainfall minima fell in the Sporer Minimum (1420?1570), the Maunder Minimum (1645?1715), and the Dalton Minimum (1790? 1820) periods of low sunspot activity, and in the two latter cases increased volcanic activity as well (Soon and Yaskell 2004; Haltia?Hovi et a1. 2007). This suggests a connection with the centennial-scale modulation of the solar andfor volcanic radiative forcings. In sharp con-' trast, such historically timed GM minima were not found in the unforced runs. The strengthening of mod- eled GM in the twentieth century and especially during the 1960?90 interval seems unprecedented, which corre- sponds to the sharp multidecadal increase in solar forcing during the ?rst half of the twentieth century and the large increase in atmospheric C02 and CH4 concentration since around 1800 (see further discussion in section 5 below). Additionally, the NHMI and SHMI are signi?? cantly correlated on multidecadal to centennial time scales, the correlation coef?cients between them are 0.71 for 31-yr running-mean series. Centennial variations are also signi?cant in the ERIK run. Power spectrum analyses (Wei 2007) of the 31-yr running-mean series of the hemispheric-scale monsoon indexes (GMI, NHMI, and SHMI) for the ERIK run are shown in Fig. 4. It can be seen that a 192-yr peak is sig- ni?cant above 95% con?dence level (by red noise test) for the global and SH monsoon indexes. The NHMI (Fig. 4b) has a less signi?cant l92?yr peak probably because of larger random noises on this bicentennial scale in the Northern Hemisphere. Other signi?cant periods in the spectrum are marked around 107 yr (mainly be- cause of NHMI) and 74 yr (mainly because of SHMI). We have also calculated the power spectra for all three monsoon indexes using both the raw annual data and the 7-yr running-mean data with quantitatively different spectral peaks and different statistical signi?cances of the peaks, cautioning on the arti?cial effects of the real spectra convoluted with the spectra of the smoothing ?lter. However, we believe that the bicentennial and centennial scales in the modulated GMI, NHMI, and SHMI indexes are qualitatively robust and may be plausibly connected to the prescribed forcing and re- sponse of the global monsoon system studied here. On the global scale, there have been no integrated observations that can be used for either checking or 2362 CTL run NHMI JOURNAL OF CLIMATE VOLUME 22 ERIK run NHMI SHMI SHMI GMI 041 02- -04- -os- Year 0 100 260 300 400 500 500 700 300 900 1000 1100 1200 1300 1i00 1500 1000 l?00 1000 1000 Year (AD) FIG. Time series of the 7-yr running-mean monsoon indexes: CT (free coupled) run and ERIK (forced) run for (top) NHMI, (middle) SHMI, and (bottom) GMI. The thick solid lines represent the 31-yr running means, which highlight centennial variations. confronting the model results. However, the 200-yr os- cillation has been noted empirically at many individual sites Zhong et a1. 2007; van Beynen et al. 2007; Mangini et al. 2007; Allen et al. 2007; Vonmoos et al. 2006; Wang et al. 2005; Lim et al. 2005; Holzkamper et al. 2004; Dehnonte et al. 2005; and Damon 2003; Agnihotri et al. 2002; Wagner et a1. 2001; Chambers and Blackford 2001; Hong et 2001, 2000; Ram and Stolz 1999; Yu and Ito 1999; Leventer et al. 1996). The period of 74 yr is very signi?cant (60~80 yr) in the South Asian and East Asian monsoon region for both temperature and precipitation (Zhu and Wang 2002; Goswami 2004; Ding et al. 2007). The ECHO-G results might provide useful clues for further assembling empirical proxy data on the global scale and for in- terpretation of underlying physical mechanisms to the oscillation. 4. Variations in the spatial structure of the 30-yr climatology of the global monsoon The Spatial structure of annual?mean precipitation anomaly and the global monsoon precipitation anom- aly with reference to the corresponding long-term means (1000?1990) during the three periods of MWP, LIA. and PWP are shown in Fig- 5. The annual-mean pre~ cipitation anomalies during the MWP, LIA, and PWP depict how the mean precipitations during these three epochs deviate from the long?term mean precipitation 1 MAY 2009 ET AL. 2363 GMI in ERIK run tillNormalized Fourier power spectrum 99% con?dence -- 95% con?dence 90% con?dence 960.0 240.0 137.1 96.0 73.0 60.0 50.5 43.6 30.4 331.3 31.0 Periodicity (yr) NHMI in ERIK run 150 . i 1-10- 110? 120- 110- 99% con?dence .- 95% con?dence 90% con?dence 501 Normalized Fourier poorer spectrum 960.0 24001311 95.0 73.3 60.0 50.5 43.6 30.4 34.3 31.0 Periodicity (yr) SHMI in ERIK run 150 . i 110- 130- i 120? 110- -- 99% con?dence 1901 1 95% con?dence as 90% con?dence Normalized Fourier poWer spectrum 960.0 241001311 90.0 73.0 60.0 50.5 43.3 33.4 34.3 31.0 Periodicity (yr) FIG. 4. Spectrum analyses on the 31-yr running mean of the monsoon indexes for ERIK simulation: (3) GMI, NHMI. and SHMI. (1000?1990). Note that the global monsoon precipitation anomaly is de?ned by the A precipitation anomaly in the NH and DJF precipitation anomaly in the SH, which depicts the strength of the local summer monsoon pre- cipitation or roughly the annual range of the monsoon precipitation. The MWP and LIA periods are plotted around the maximum of GMI intensity at 1200 and the minimum of GMI at 1685, respectively (see Fig. 3). Figure 5a compares the annual-mean precipitation distribution for the three periods. The annual-mean precipitation beMeen and is 3.12 mm, 3.09, and 3.12 mm day" for MWP, LIA, and PWP, respec- tively (Table 1), which indicates the increase of the total rainfall between and during the warm periods of MWP and PWP. The annual-mean precipitation in all the 6 continental monsoon domains is 3.78, 3.75, and 3.80 mm day?l (Table 1) for MWP, LIA and PWP, re- spectively, which means that the annual-mean precipi- tation in the global monsoon domain decreased during the relatively cold LIA period compared to the MWP and PWP periods. Although with a slight difference in the timing for LIA and MWP and probably beyond the spatial resolution of our model outputs, the recent high- quality reconstruction of monsoon rainfalls for south- west China (between the moisture transport pathway of Dongge Cave and Heshang Cave; broadly represented by our N2 monsoon region) by Hu et a1. (2007) sug- gested a relatively wetter and dryer conditions for the MWP and LIA, respectively. Figure 5b compares the local summer precipitation changes connected to the changes in the local mon- soon strength. During the MWP and PWP periods. monsoon strengthens nearly globally in each of the con- tinental regional monsoons. This is especially so for the present-day monsoon climatology. On the other hand, during the LIA, there is a general decrease in each 2364 (0) Annual mean precipitation anomaly . 40$- BUS 608 a sin trio 12in: This! 0 JOURNAL OF CLIMATE VOLUME 22 Global monsoon precipitation anomaly a sin 12's]: 160 tabs min a WI I I i -i.5 ?o.e -a.3 a 0.3 0.5 0.9 1. 1.5 turn/day FIG. 5. Comparison of precipitation patterns for the three 30-yr epochs: (top) MWP (1185?1214). (middle) LIA (1685?1714), and (bottom) PWP (1961?90). The annual-mean precipitation anomaly and the global monsoon precipitation anomaly with reference to the corresponding long-term means (1000?1990). The global monsoon precipitation is de?ned by the HA precipitation in the NH and DJ precipitation in the SH. The enclosed red lines outline the monsoon domains. of the monsoon regions except the oceanic S3. The precipitation anomaly of global monsoon during the MWP, LIA, and PWP is 0.014, ?0.021, and 0.029 mm day-1, respectively. This fact suggests that forced re- sponses of the regional monsoons have a cohesive pat- tern and they are coordinated by the superposed changes in the external forcing. Thus, the global mon- soon index otfers the measure of a global-scale trend common to the regional monsoons except oceanic monsoon region SB. We next note that, although the increase in GMI is similar between and PWP, the spatial patterns have some differences (Fig. 5b). The present-day cli- mate features largest increase of the annual-mean pre- cipitation over the equatorial western Paci?c. while during MWP rainfall over the equatorial western Paci?c warm pool decreases signi?cantly. Changes in the man? soon strength in the Mexican monsoon also differ sig- ni?cantly for the simulated monsoon climatologies for the MWP and PWP. Table 1 shows that the oceanic monsoon S3 behaves differently from all other 6 continental monsoon regions. In the continental monsoon regions, LIA precipitation is less than MWP and PWP period, but the oceanic I MAY 2009 TABLE 1. The annual-mean precipitation (mm day?) of re gional and global monsoon regions (exclude S3) and belt during the MWP (1185?1214), LIA (1685?1714), and PWP (1961?90). The notation N1represents the northern African, Asian, North American, Southern African, Australian. Central South Paci?c. and South American monsoon regions, respectively. Region MWP LIA PWP N1 3.187 3.181 3.285 N2 3.734 3.642 3.790 N3 5.108 5.047 5.056 31 3.463 3.334 3.408 52 3.599 3.405 3.473 53 6.088 6.137 6.087 S4 3.885 3.832 3.985 Global monsoon (without S3) 3.782 3.747 3.798 belt 3.122 3.092 3.117 monsoon region (83) is just opposite. For this reason, we had excluded S3 from our integrated monsoon indexes, namely the SHMI and GMI. 5. Attribution and mechanisms Figure 6 compares the time series of the direct solar radiative forcing, indirect radiative forcing from volca- nic eruptions, effective radiation forcing (the sum of the solar and volcanic forcing), and atmospheric C02 concentration, along with the global monsoon index, global-mean temperature, and interhemispheric tem- perature difference (NH minus SH). All time series were smoothed with a 31?yr running-mean ?lter in order to better highlight centennial-to-millennial variations. The correlation coef?cients between GMI and the sus? pected relational or causal factors are shown in the lower-right corners of Fig. 6 and presented in greater details in Table 2. a. Variations of the forcing factors The amplitude of variations of the solar irradiance at centennial and longer time scales is still being debated (Krivova et a1. 2007; Solanki and Krivova 2006; Bard and Frank 2006). The amplitude of these variations is usually scaled numerically by the difference between present values and the Late Maunder Minimum. In this simulation this difference (1960?90 mean minus 1680? 1710 mean) is Thus, in the simulation the solar radiation reaching the t0p of the atmosphere shows signi?cant variation on millennium time scale with a maximum around 1100?1250 and the present day to- gether with a minimum around 1450 (Fig. 6a). The latest value in the simulation (1990) is about 1 rn?2 higher LIU ET AL. 2365 1358- Solar constant 1366 - 1364' 1362- 0.72 Volcanic effect 0.37 Effective solar radiation a i concentration (ppm) Global monsoon index(GM ) (mm/day) 287.4 - Global mean temperature (K on Fin?e NH minus SH temperature 00 IE tuna 110a 1211:: 13in 14hr: 15110 who who who who Your (AD) FIG. 6. Smoothed time seriesvof the solar radiative forcing (W volcanic effect (W effective solar radiation (W concentration (ppm), GMI (mm day?), global-mean temperature (K), and interhemispheric tempera- ture difference (K). All time series are 31?yr waning means from AD 1000 to 1990. The numbers shown in the lower?right comers indicate correlation coef?cients of with the four external forcing factors and two temperature indexes, respectively. The interhemiSpheric temperature difference is de?ned by the NH averaged temperature minus the SH averaged temperature. 2366 JOURNAL OF CLIMATE VOLUME 22 TABLE 2. Correlation coef?cients between GMI and other variables in ERIK run. Those values that are statistically signi?cant at the 95% con?dence level-are indicated in bold numbers. The method of signi?cant test follows Chen (1982). NH-SH Solar Volcanic Effective solar Global-mean Correlation coef?cient constant activity radiation C02 CH4 temperature temperature Original data 0.328 0.245 0.335 0.294 0.251 0.382 0.427 7?yr running mean 0.579 0.384 0.638 0.518 0.438 0.724 0.659 31-yr running mean 0.724 0.374 0.777 0.618 0.500 0.862 0.834 than the MWP period. There are centennial ?uctuations superposed on the millennium variation. Spectral anal- ysis shown in Fig. 7a con?rms that the variance is con- centrated on centennial and bicentennial time scales and with prominent peaks occurring on 192 and 120 yr. Similar to our calculations of the power spectra for the monsoonal indexes in Fig. 4, we have also found quantitatively different results for both the peaks and their statistical signi?cances for all the indexes in Fig. 7 using both the raw annual-mean and the 7-yr smoothed series. We believe that these quantitative differences will not strongly affect the key conclusion in the nar- row context of our study of the bicentennial and cen- tennial scales of forcing and response of the global monsoon system. The episode of volcanic forcing changes from year to year; its 31-yr running mean shows primarily a variation on bicentennial time scales (Fig. 6b). The spectral analysis con?rms peaks on 192 and 107 yr, respectively, which are both signi?cant at the 95% con?dence leVel (Fig. 7b). What gives rise to the periodicity in the ef? fective volcanic forcing is very curious because volcanic eruptions have been .thought to be more or less chaotic and unpredictable with no regularity in time. The effective solar forcing reaching the top of the atmosphere, which is the sum of the solar forcing at the top of the atmosphere and the radiative equivalent of volcanic activity, shows both a long-term variation (presumably on a millennial time scale owing mainly to the variation of the solar radiation) and quasi-bicentennial (192 yr) and quasi?centennial (107 yr) variations (pri- marily due to the superposed variations of both the volcanic and solar forcings; see Figs. 60 and 7c). These Spectral peaks are signi?cant at the 99% con?dence level by red noise test, but we rather place emphasis on physical mechanisms than statistics. The atmospheric C02 concentration in the prein- dustrial period is ?at (around 285 ppm) except a rela- tively low period between 1600 and 1800 at about 275?280 ppm; see Fig. 6d). The smoothed C02 series has increased near?expenentially?since 1850-4975 (around 330 ppm). Atmospheric CH4 is also important, for it is responsible for about 25% of the increase anthropo- genic radiative forcing between preindustrial period and the present (Solomon et a1. 2007). b. Response of the global monsoon precipitation to external forcing How does the global monsoon precipitation respond to the changes in the aforementioned forcings? Fig. 6c shows that the GMI tends to vary in phase with the effective radiative shortwave forcing (solar forcing plus volcanic forcing), especially on the millennium time scales. The correlation coef?cient between GMI and the effective radiation is 0.78 for the 31-yr smoothed series, which is better than the correlation with the solar forcing (0.72) and much better than the correlation with the volcanic forcing (0.37; see Fig. 6 and Table 2). The better correlation between the GMI and solar forcing comes from their millennium variations. The spectrum of the GMI has pronounced peaks on 192, 107, and 74 yr (Fig. 7d), which corresponds well to the sig- ni?cant spectral peaks from effective radiative forcing at around 192, 107, and 80 yr. The variation of the GM precipitation may thus be linked to three factors. First, its millennium variation (peaks in MWP and present and dips in LIA interval) can be well explained by changes in the direct solar ir- radiance. The three GMI minima during the LIA con- cur with the three minima in shortwave forcing, which further supports the impact of the effective solar forcing on the global monsoon precipitation. Second, a com- parison of Figs. 7a-c with Fig. 7d suggests that the quasi- bicentennial (192 yr) oscillations in the GM precipita- tion appear to be primarily induced by the solar forcing with ampli?cation by the volcanic forcing. The quasi- centennial (107 yr) oscillation may be related primarily to volcanic forcing with ampli?cation by the solar forcing. Third, while the direct solar irradiance in the last two decades of the simulation is higher than that of MWP by about 1 m'l, the effective solar irradiance in the late twentieth century is 0.52 m?2 lower than that during MWP because of the increase in effective volcanic?forcing?rn?tlre? late twentieth century (see Fig. 6). On the other hand, the GM rainfall rate in the PWP is 0.016 mm day? higher than that during the MAY 2009 Solar constant :50 MM 130- 120i 99% con?dence -- 95% con?dence - - 90% con?dence Normalized Fourier power spectrum . 0 0 {In 130 3 900.0 240.0137.1 96.0 73.3 60.0 50.5 43.5 33.4 34.3 31.0 Periodicity (yr) Volcanic effect :50 i4o99% con?dence .. 95% con?dence 90% con?dence Normalized Fourier power spectrum 950.0 240.0131?! 95.0 73.3 60.0 50.5 43.5 33.4 34.3 31.0 Periodicity (yr) Effective solar radiation 159. 140' i20~ I) iwi 99% con?dence 3 100- -- 95% con?dence 3 so- con?dence 560.0 24031311 96.0 73.5 50.0 50.5 43.6 33.4- 3453 31.0 Periodicity (yr) LIU ET AL. 2367 Global monsoon index 150 Ht] 1.10- 99% f'd 110- con: ence 3 mn- -- 95% con?dence 3 90- 90% con?dence D. El960.0 240.0 137.1 96.0 73.8 50.0 50.5 43.6 33.4 30.3 31.0 Periodicity (yr) (9) Global mean temperature 99% con?dence 95% con?dence 90% con?dence Normalized Fourier powor spectrum I I 960.0 24031311 95.0 73.3 60.0 50.5 43.6 35.4 34.3 31.0 Periodicity (yr) NH minus SH temperature 150 140 - 120 110- um '90- aa- 70- so- 50- 40? 30? an] 99% con?dence -- 95% con?dence 90% con?dence Normalized Fourier power spectrum 900.0 240.0 137.1 90.0 73.0 60.0 50.5 43.5 Periodicity (yr) in}! 38.4 34.3 31.0 FIG. 7. Spectra of the 31-yr running-mean external forcing factors and atmospheric responses. (3) Solar constant, (13) volcanic effect, effective solar radiation, global-mean temperature, NH minus SH temperature. MWP (Table 1), which indicates that the solar and volcanic forcing can explain the simulated GM rainfall from 1000 to 1950, but fails to account for most of the observed increase of GM precipitation in 1961?90, es- pecially after 1975. This fact may suggest that the rapid increase of atmospheric C02 and CH4 might have a positive contribution to the recent increase in the GM precipitation. 2368 c. The mechanism by which effective radiative forcing modulates GM rainfall In the preindustrial period, changes in the total amount of effective shortwave radiative forcing can re- inforce the thermal contrast between the continent and ocean, thereby resulting in the centennial- to millennial- scale variations in the global monsoon strength. Land has much smaller heat capacity than the ocean. When effective radiative ?ux increases during the local sum? mer, the magnitude of land warming is much stronger than that in the adjacent ocean, thus the thermal contrast between continent and ocean gets reinforced (Table 3). This thermal contrast further enhances the pressure differences between land monsoon regions and the sur- rounding oceans (Table 3) and thus strengthens the monsoon circulation in the presence of Coriolis force and associated rainfall. As such, each regional compo? nent of the global monsoon system will intensify from the increased radiative heating and thus the composite global monsoon will be strengthened as well. Quantitatively, the change in GM rainfall rate be? tween MWP and LIA is about 0.035 mm day? which is about 0.93% change in precipitation strength (Table 1), while the change of solar irradiance between and LIA is about 2.71 m?2 (Table 3), which is about 0.2% of the solar constant. Given a 0.2% increase in the ex? ternal forcing, the increase in GM rainfall is 4?5 times larger. Such ampli?ed response is similar to a near- resonant response of a dynamical system to an external forcing. What processes have catalyzed and ampli?ed the response? We argue that the effective radiative forcing-induced land?ocean thermal contrast causes an initial increase in monsoon precipitation. This initial increase is further reinforced by the increase in moisture supply because the warming induced by the effective radiative heating tends to increase atmospheric mois? ture content. The increase of moisture supply can in- duce a positive feedback between the latent heat release (in precipitation) and monsoon ?ow convergence, thus further amplify the latent heating release, which may ultimately amplify the atmospheric circulation response. Therefore, the humidity feedback is a key ampli?er linking solar irradiance and monsoon. A vigorous in- crease in rainfall is expected in response to a moderate change in radiative heating. Relationship between changes in. global temperature and monsoon rainfall It can be observed from the results in Fig. 6 and Table 2 that the change of the global monsoon strength tends to roughly vary in accord to the changes in the global- mean surface temperature. The correlation coef?cient JOURNAL OF CLIMATE VOLUME 22 TABLE 3. The anomalies of effective solar radiation (AS) and summer mean (JJA for NH and DJF for SH) differences of sea level pressure and temperature between land and sea for MWP, LIA, and PWP intervals in reference to 1000?1990. The difference between the land and ocean was computed based on the averaged quantity over all land grids minus that over all oceanic grids. Increment MWP LIA PWP as (w 1.576 [.138 1.052 Mam?flan?) (hPa) 0.273 ?0.150 0.309 (K) 0.080 ?0.137 0.217 (hPa) 0.457 -0.307 0.779 AT.qu (K) 0.016 -0.078 0.147 between the 31-yr running?mean GM and global?mean temperature is 0.86. However, the global monsoon rainfall follows effective radiative forcing more closely than the global-mean temperature does. This point is particularly well ful?lled during the LIA (Figs. 6c,e,f). Overall, the correlation coef?cient between the 31-yr running-mean GMI and effective solar radiation is 0.78, while that between global?mean temperature and ef- fective solar radiation is 0.69. The global monsoon is essentially driven by the in- terhemispheric temperature difference and differential heating between the NH and SH. To understand the changes in the global monsoon strength or GMI, one should naturally seek for the root cause from the inter- hemispheric contrast in the temperature and precipita- tion. It is found that the GMI varies coherently with the interhemispheric temperature difference. The variation of the interhemispheric temperature difference has 192- and 107-yr periodicities; both are signi?cant above 99% con?dence level (Fig. 7f). This result means that the quasi-bicentennial and quasi-centennial GM changes coherently with the interhemispheric temperature dif- ference. The interhemispheric temperature difference is related to the solar activity change and the land?ocean distribution- Poleward of earth's surface is covered largely by land and the proportion of land there is 47%. In contrast, poleward of land proportion is much smaller (only In response to increased radiative forcing, the NH warms up more than the SH because of the smaller heat capacity of land. As such, the NH minus SH temperature should vary in concert with the effective radiative forcing, and is thus correlated closely with the GM precipitation changes. 6. Conclusions In this paper, we study how the global monsoon (GM) precipitation responds to the external forcing in the last millennium by analyzing a pair of control and forced millennium simulations. The forced ERIK run has been 1 MAY 2009 shown to capture precipitation climatology realistically (Fig. 1) despite the incomplete accounting of all forcing factors. More speci?cally, two variables are used to gauge the model?s performance in simulation of the precipitation. One is the global-mean precipitation and the other is the leading mode of the annual cycle of precipitation. The leading mode of the annual cycle is characterized by a solstitial monsoonal mode whose strength can be described by a global monsoon precip- itation index. We demonstrate that the ERIK run cap? tures the two modes of climatology comparably well when compared with those captured by the NCEP re- analysis. This adds con?dence to the analysis of the change in the annual cycle in the model?s millennium simulation. The 31-yr running averaged global monsoon index in the forced run reveals both the variability on the mil- lennial time scale and signi?cant quasi-bicentennial (192 yr) and quasi-centennial (107' yr) variability. Over the past millennium, the simulated global monsoons are observed to be strong in the model Medieval Warm Period (ca. 1030?1240), while the simulated global mon- soon intensity gets weaker during the model Little Ice Age (ca. 1450?1850). During the LIA interval there are three GMI minima occurring around 1460, 1685, and 1800, which correspond, respectively, to the Sporer Minimum (1420-1570), Maunder Minimum (1645?1715), and Dalton Minimum (1790?1820) periods of solar activity minima and increased volcanic activity. The prominent increase of the global monsoon strength in the last century and the remarkably strengthening of the global monsoon in the last 30 yr of simulation ending in 1990 seem large; the latter may signify a possible impact from the rapid increase in atmospheric green- house gases. Before the industrial period, the changes in the sum of the direct solar radiative forcing and volcanic forcing (effective radiative forcing) can explain the natural global monsoon precipitation variations well. Simulated changes of the GM strength?in the last century have a spatial pattern that differs from that during the WP, suggesting the different effects of global warming on monsoon precipitation patterns contributed by both the increases of atmospheric greenhouse gases and the in- coming solar radiation. On a centennial time scale, the change of the global monsoon strength follows the El?feTIive radial it? forcing better than the changes of the global-mean surface temperature. Physically, the GMI has a good correla- tion with interhemispheric temperature difference as elaborated in the previous section. We leave two main areas of research for future at- tention in order to bring forth a more complete char? LIU ET AL. 2369 acterization and understanding of global monsoon and its variation on centennial to millennial time scales. First, a more complete set of relevant forcing factors and physical processes must be included in the model simulation. Then a more direct and meaningful com- parison of the model simulated outputs with all the available regional monsoon proxies can be performed. Acknowledgments. Jian Lin and Bin Wang acknowl- edge the ?nancial supports from the Innovation Project of Chinese Academy of Sciences (Grant 315),the National Basic Research Program of China (Grant 2004CB720208), and the National Natural Sci- ence Foundation of China (Grant 40672210). Bin Wang and Qinghua Ding acknowledge the support received from the National Science Foundation (NSF) climate dynamics program (ATMO6-29531) and NOAAJOGP. REFERENCES Agnihotri, R., K. Dutta, R. Bhushan, and B. L. K. Somayajulu, 2002: Evidence for solar forcing on the Indian monsoon during the last millennium. Earth Planet. Sci. Lem, 198, 521? 527. Allen, J. R. M., A. J. Long, C. Ottley, D. G. Pearson. and B. Huntley, 2007: Holocene climate variability in northernmost Europe. Quai. Sci. Rev., 26, 1432-1453. Ammann, C. M., F. cos, D. S. Schimel,B. L. Otto-Bliesner, and R. A. Tomas, 2007: Solar in?uenca on climate during the past millennium: Results from transient simulations with the NCAR Climate System Model. Proc. Natl. Acad. Sci. USA, 104, 3713?3718. Bard, and M. Frank, 2006: Climate change and solar variability: What?s new under the sun? Earth Planet. Sci. Lent, 248, 1?14. Bauer, E., M. Claussen, V. Brovkin. and A. Huenerbein, 2003: Assessing climate forcings of the earth system for the past millennium. Geophys. Res. Lem, 30, l276, doi:10.1029/ Beltrami, H., J. F. Gonzalez-Renee, and M- B. Stevens. 2006: Subsurface temperatures during the last millennium: Model and observation. Geophys. Res. Lett, 33, L09705, doi:10.1029/ 20060L026050. 'B?inier, Chappellaz, J. Schwander. B. Stauffer, and D. Raynaud, 1995: Variations in atmospheric methane concen- tration during the Holocene epoch. Nature, 3'74, 46?49. Chambers, F. M., and J. J. Blackford, 2001: Mid? and late?Holo- cene climatic changes: A test of periodicity and solar forcing in proxy-climate data from blanket peat bogs. J. Quai. Sci, 16, 329-338. Chen, W. Y., 1982: Fluctuations in Northern HemisPhere 700 mb heiglil ?eld associated with the Mon. Wea. REM, 110, 808?832. Crowley, T. J., 2000: Causes of climate change over the past 1000 years. Science, 289, 270?277. Delmonte, B., J. R. Petit, G. Krinner, V. Maggi, J. Jouzel, and R. Udisti, 2005: Ice core evidence for secular variability and ZOO-year dipolar oscillations in atmospheric circulation over east Antarctica during the Holocene. Climate Dyn., 24, 641-654. 2370 Ding, Y., Z. Wang, and Y. Sun, 2007: Inter-decadal variation of the summer precipitation in East China and its association with decreasing Asian summer monsoon. Part 1: Observed evi- dences. Int. J. Climatol, 28, 1139?1161. Etheridge, D. M., L. P. Steele, R. L. Langenfelds, R. J. Francey, Bamola, and V. I. Morgan, 1996: Natural and anthro? pogenic changes in atmospheric over the last 1000 years from air in Antarctic ice and J. Geophys. Res, 10], 4115?4128. Fleitmann, D., S. J. Burns, M. Mudelsee, U. Neff, J. Kramers. A. Mangini, and A. Matter, 2003: Holocene forcing of the lndian monsoon recorded in a stalagmite from southern Oman. Sci- ence, 300, 1737?l739. Gonzalez?Rouco, F., H- von Storch, and E. Zorita, 2003: Deep soil temperature as proxy for surface air-temperature in a coupled model simulation of the last thousand years. Geophys. Res. Lat, 30, 2116, Goosse, H., and H. Renssen, 2004: Exciting natural modes of variability by solar and volcanic forcing: Idealized and real- istic experiments. Climate Dyn, 23, 153?163. . A. Timmermann, and R. S. Bradley, 2005: Internal and forced climate variability during the last millennium: A model-data comparison using ensemble simulations. Qaal. Sci. Rev.. 24, 134541360. Goswami. B. N., 2004: Interdecadal change in potential predict- ability oi the Indian summer monsoon. Geophys. Res. Lem, 31, L16208, doi:10.l02912004GL020337. Gouirand. I., A. Moberg, and E. Zorita, 2007a: Climate variability in Scandinavia for the past millennium simulated by an atmosphere-ocean general circulation model. Tellus, 59A, 30?49. V. Moron, and E. Zorita, 2007b: Teleconnections between ENSO and North Atlantic in an ECHO-G simulation of the 1000?1990 period. Geophys. Res. Lett, 34, L06705, doi:10.1029/ 20060L028852- Haltia-Hovi, T. Saarinen, and M. Kukkonen, 2007: A 2000-year record of solar forcing on varved lake sediment in eastern Finland. Qtlm'. Sci. Rev., 26, 678?689. Holzkiimper, S., A. Mangini, C. Spiitl', and M. Mudelsee, 2004: Timing and progression of the Last Interglacial derived from a big alpine stalagmite. Geophys. Res. Lett, L07021, doi:10. 1029120030101 91 12. Hong, Y. T., H. B. Jiang, T. S. Liu, X. G. Qin, L. P. Zhou, J. BeerLeng, 2000: Response of climate to solar forcing recorded in a 6000-year delta 0-18 time-series of Chinese peat cellulose. Holocene, 10, and Coauthors, 2001: A 6000?year record of changes in drought and precipitation in northeastern China based on a delta C-13 time series from peat cellulose. Earth Planet. Sci. Lem, 185, 111419. Hu, C., G. M. Henderson, J. Huang, S. Xie, Y. Sun, and K. R. Johnson, 2007: Quanti?cation of Holocene Asian monsoon rainfall from spatially separated cave records. Earth Planet. Sci. Lctt., 266, 221?232. Johns, T. C., R. E. Carnell, J. F. Crossley, J. M. Gregory, J. F. B. Mitchell, C. A. Senior, S. F. B. Tett, and R. A. Wood, 1997: The second Hadley Centre coupled ocean?atmosphere GCM: Model description, spinup and validation. Climate Dyn, 13, 103?134. Kanamitsu, M., W. Ehisuzaki, J. Woollen, S. K. Yang, J. J. Hnilo, M. Fiorino, and G. L. Potter, 2002: AMIP-II Reanalysis (R-2). Bull. Amer. Meteor. Soc, 83, l631-1643. Kitoh, A., 2006: Variability of Indian monsoon-ENSO relationship in a lOOD-year MRI-CGCM22 simulation. Nat. Hazards, 42, 261?272. JOURNAL OF CLIMATE VOLUME 22 Krivova. N. A., L. Balmaceda, and S. K. Solanki. 2007: Recon- struction of solar total irradiance since 1700 from the surface magnetic ?ux. Astraa. Astrophys. 467, 335?346. Legutke, S., and R. Voss, 1999: The Hamburg Atmosphere-Ocean Coupled Circulation Model ECHO-G. German Climate Com- puter Center (DKRZ) Tech. Rep. 18, 62 pp. Leventer, A., E. W. Domack, S. E. Ishman, S. Brachfeld, C. E. McClennen. and P. Manley, 1996: Productivity cycles of 200- 300 years in the Antarctic Peninsula region: Understanding linkages among the sun, atmosphere, oceans, sea ice, and biota. Geol. Soc. Amer. Bali, 1118, 1626?1644. Lim, J., E. Matsumoto, and H. Kitagawa, 2005: Eolian quartz ?ux variations in Cheju Island, Korea, during the last 6500 yr and a possible Sun-monsoon linkage. Qttat. Res, 64, 12?20. Liu, 1., H. Storch, X. Chen, E. Zorita, J. Y. Zheng, and S. Wang. 2005: Simulated and reconstructed winter temperature in the eastern China during the last millennium. Chin. Sci. Bali, 50, 2872?2877. Manabe, 3., and R. Stouffer, 1993: Century-scale effects of in- creased atmospheric CO: on the ocean?atmosphere system. Nature, 364. 215?218. Mangini, A., P. Verdes, C. Spot], D. Scholz, N. Vollweiler, and B. Kromer, 2007: Persistent in?uence of the North Atlantic liydrography on central European winter temperature during the last 9000 years. Geophys. Res. Lett, 34, L02704, doi:10. Mann, M. E., M. A. Cane, S. E. Zebiak, and A. Clement, 2005a: Volcanic and solar forcing of the tropical Paci?c over the past 1000 years. J. Climate, 18, 447?456. S. Rutherford, E. Wahl, and C. Ammann, 2005b: Testing the ?delity of methods used in proxy-based reconstructions of past climate. J. Climate, 18, 4097?4107. Mi S. Legutke, A. Hense, and W.-T. Kwon. 2005a: Internal variability in a 1000-yr control simulation with the coupled climate model Near-surface temperature, preci- pitation and mean sea level pressure. Telltar, 57A, 605?621. and 2005b: Internal variability in a 1000?yr control simulation with the coupled climate model El Ni?o Southern Oscillation and North Atlantic Oscillation. Telhts. 57A, 622?640. Osborn, T. S. C. B. Raper, and K- R. Briffa, 2006: Simulated climate change during the last 1,000 years: Comparing the ECHO-G general circulation model with the MAGICC sim? ple climate model. Climate Dyn, 27, 185?197- A. N.. and P. E. Damon, 2003: Persistence of the 88-year solar cycle over the last ~12,000 years: Evidence from cosmogenic isotopes. J. Geophys. Res, 108, 1003, doi:10.1029l20021A009390. Ram, M., and M. R. Stolz, 1999: Possible solar in?uences on the dust pro?le of the GISPZ ice core from central Greenland. Geophys. Res. Lett, 26,1043?1046. Rodgers, K. B., P. Friederichs, and M. Latif, 2004: Tropical Paci?c decadal variability and its relation to decadal modulations of ENSO. J. Climate, 17, 3761?3774. Roeckner, E., and Coauthors, 1996: The atmospheric general cir- culation model ECHAM4: Model description and simulation of present-day climate. Max Planck Institute for Meteorology Tech. Rep. 218, 90 pp. Solanki, S. K., and N. A. Krivova, 2006: Solar variability of possible relevance for planetary climates. Space Sci. Rev., 125, 37. Solomon, 5., and Coauthors, 2007: Climate Change 2007: The Physical Science Basis. Cambridge University Press, 996 pp. MI 2009 Soon. W. and S. H. Yaskell, 2004: The Meander Minimum and the Variable Sun-Earth Connection. World Scienti?c. 296 PP. Stendel. M., l. A. Mogensen, and H. Christensen, 2006: In?uence of various forcings on global climate in historical times using a coupled atmosphere?ocean general circulation model. Cli- mate Dyn., 26, 1?15. Stouffer. R. J., G. Hegerl. and S. Tell, 2000: A comparison of surface air temperature variability in three 1000-yr coupled ocean?aUnOSphere model integrations. J. Clinrate, 13, 513?537. Tett, S. F. B., and Coautltors. 2007: The impact of natural and anthropogenic forcings on climate and hydrology since 1550. Climate Dyn., 28, 3?34. Trenberth, K. E., D. P. Stepaniak, and .I. M. Caron, 2000: The global monsoon as seen through the divergent atmospheric circulation. J. Climate. 13, 3969-3993. van Beynen, P. E., Y. Asmerom. V. Polyak, L. Soto. and J. 3. Polk, 2007: Variable intensity of teleconnections during the late Holocene in subtropical North America from an isotopic study of speleothem from Florida. Res. Lett, 34, doi:10.1029l2007GL031046. Vimont, D. J., D. S. Battisti, and A. C. Hirst, 2002: Paci?c inter- annual and interdecadal equatorial variability in a simulation of the CSIRO coupled general circulation model. J. Climate, 15, 160?178. Vonmoos, M., J. Beer, and R. Muscheler, 2006: Large variations in Holocene solar activity: Constraints from lOBe in the Green? land Ice Core Project ice coreJ. Geophys. Res. 111, A10105, doi:10.l029i2005.lA011500. von Storch, H., E. Zorita, J. M. Jones, Y. Dimitriev, F. Gonzalez- Rouco. and S. F. B. Tett. 2004: Reconstructing past climate from noisy data. Science, 306, 679?682. Wagner, 6., and Coauthors, 2001: Presence of the solar de Vries cycle (~205 years) during the last ice age. Geophys. Res. Lett, 28, 303406. Wagner, 8., S. Legutke, and E. Zorita, 2005: European winter temperature variability in a long coupled model simulation: The contribution of ocean dynamics. Climate Dyn., 25, 37-50. Wang, B.. 1994: Climatic regimes of tropical convection and rainfall. J. Climate, 7, 1109?1118. 2006: The Asian Monsoon. Springer, 78? pp. and Q. Ding, 2006: Changes in global monsoon precipitation over the past 56 years. Geophys. Res. Lett., 33, 1, L02S347. ET AL. 2371 and 2008: Global monsoon: Dominant mode of annual variations in the tropics. Dyn. Atmos. Oceans, 44, 165?183. Wang, Y. ., and Coauthors, 2005: The Holocene Asian monsoon: Links to solar changes and.North Atlantic climate. Science, 308, 854?857. Weber, S. L., T. Crowley, and G. van der Schrier, 2004: Solar irradiance forcing of centennial climate variability during the Holocene. Climate Dyn., 22, 539-553. Webster, P. J., 1987: The variable and interactive monsoon. Monsoons, S. Fein and P. L. Stephens, Eds., Wiley, 269?330. Wei, F., 2007: Modern Techniques on Climatic Statistics, Diag- nostics and Forecast (in Chinese). China Meteorological Press, 296 pp. Wiles, G. C., R. D. D?Arrigo, R. Villalba, P. E. Calkin, and D. J. Barclay, 2004: Century-scale solar variability and Alaskan temperature change over the past millennium. Geapiiys. Res. Lett, 31, L15203, Wolff, J. 0., E. Maier-Reimer, and S. Legutke, 1997: The Ham- burg Ocean Primitive Equation Model. German Climate Com- puter Center (DKRZ) Tech. Rep. 13, 98 pp. Xie, P. P., and P. A. Arkin, 1997: Global precipitation: A [7-year analysis based on gauge observations, satellite esti- mates, and numerical model outputs. Bull. Amer. Meteor: Soc, '78, 2539?2558. Yu. Z. G, and E. lto, 1999: Possible solar forcing of century-scale drought frequency in the northern Great Plains. Geology. 27, 263?266. Zhong, W., J. B. Xue, Q. Shu, and L. G. Wang. 2007: Climatic change during the last 4000 years in the southern Tarim Basin, Xinjiang, northwest China. J. Quat. Sci, 22, 659?665. Zhu, J., and S. Wang, 2002: 80 yr oscillation of summer rainfall over North China and East Asian Summer Monsoon. Geo- phys. Res. Lett, 29, 1672, Zorita, E., J. F. Gonzalez-Rouco, and S. Legutke, 2003: Testing the Mann et al. ([998) approach to paleoclimate reconstructions in the context of a 1000?yr control simulation with the ECHO-G coupled climate model. J. Climate, 16, 1378?1390. H. von Storch, J. P. Montavez, and F. Valero. 2005: Natural and anthropogenic modes of surface temperature variations in the last thousand years. Geophys. Res. Lent, 32. L08707. doi:10.1029l2004GL02] 563. and 2007: Comments on "Testing the ?delity of methods used in proxy-based reconstructions of past climate." J. Climate, 20, 3693-3698. ATTACHMENT Libraries Durham ?r of 5mm in 1-1-2009 University of ScholarlyCommons Validity of Climate Change Forecasting for Public Policy Decision Making Kesten C. Green University of South Australia University of Willie Soon Harvard-Smithsonian Center forAstraphysics . .(2009). Forecasting. .2501). .826-832. :1 RL: . . noww 2009.05.01] . . . I160 . Intma?onaljoumal of Validity of Climate Change Forecasting for Public Policy Decision Making Kesten C. Green Business and Economic Forecasting, Monash University, Vic 3800, Australia. Contact: PO Box 10800, Wellington 6143, New Zealand. kesten@kestencgreen.com; +64 4 976 3245; +64 4 976 3250 J. Scott The Wharton School, University of 747 Huntsman, Philadelphia, PA 19104 +1 610 622 6480 Willie Soon Harvard-Smithsonian Center for Astrophysics, Cambridge MA 02138 wsoon@cfa.harvard.edu; +1 617 495 7488 February 24, 2009 ABSTRACT Policymakers need to know whether prediction is possible and if so whether any proposed forecasting method will provide forecasts that are substantively more accurate than those from the relevant benchmark method. Inspection of global temperature data suggests that it is subject to irregular variations on all relevant time scales and that variations during the late 19005 were not unusual. In such a situation, a ?no change? extrapolation is an appropriate benchmark forecasting method. We used the U.K. Met Of?ce Hadley Centre?s annual average thermometer data from 1850 through 2007 to examine the performance of the benchmark method. The accuracy of forecasts ?'om the benchmark is such that even perfect forecasts would be unlikely to help policymakers. For example, mean absolute errors for 20- and 50-year horizons were and We nevertheless demonstrate the use of benchmarking with the example of the Intergovernmental Panel on Climate Change?s 1992 linear projection of long?term warming at a rate of libertarian?sample of errors from ex ante projectionsat?OaOSBG-per-year for 1992 through 2008 was practically indistinguishable from the benchmark errors. Validation for long-term forecasting, however, requires a much longer horizon. Again using the IPCC warming rate for our demonstration, we projected the rate successively over a period analogous to that envisaged in their scenario of exponential C02 growth?the years 1851 to 1975. The errors from the projections were more than seven times greater than the errors from the benchmark method. Relative errors were larger for longer forecast horizons. Our validation exercise illustrates the importance of determining whether it is possible to obtain forecasts that are more useful than those from a simme benchmark before making expensive policy decisions. Key words: climate model, ex ante forecasts, out-of-sample errors, predictability, public policy, relative absolute errors, unconditional forecasts. Introduction We examine procedures that should be used to evaluate forecasts of global mean temperatures over the policy-relevant long term. A necessary condition for using forecasts to inform public policy decisions is evidence that the proposed forecasting procedure can provide ex ante forecasts that are substantively more accurate than those from a simple benchmark model. By ex ante forecasts, we mean forecasts for periods that were not taken into account when the forecasting model was developed. Benchmark errors provide a standard by which to determine whether alternative scienti?cally- based forecasting methods can provide useful forecasts. When benchmark errors are large, it is possible that alternative methods would provide useful forecasts. When benchmark errors are small, it is less likely that other methods would provide improvements in accuracy that would be useful to decision makers. An benchmark model Exhibit 1 displays Antarctic temperature data from the ice-core record for the 800,000 years up to 1950. The temperatures are relative to the average for the last one?thousand-years of the record (950 to 1950 AD), in degrees Celsius. The data show large irregular variations and no obvious trend. For such data the no?change forecasting model is an appropriate benchmark. INSERT EXHIBIT 1 ABOUT HERE 800,000-year Record of Antarctic Temperature Change 800 600 400 200 Time (thousands of years before 1950) Antarctic temperature anomaly relative to 950-1950 AD mean 1 The ability of a model to ?t time series data hears little relationship to its ability to forecast; a ?nding" that has often puzzled researchers 2001, pp. 460-462). Performance of the benchmark model We used the Hadley ?best estimate" annual average temperature differences from 1850 to 2007 ?'om the UK. Met Of?ce Hadley Centre2 to examine the benchmark errors for global mean temperatures (Exhibit 23) over policy-relevant forecasting horizons. IN EXHIBIT 2 Hadley annual temperature data for 1850 to 2008 Global surface temperature deviation from 1961-1990 average ?Celcius -0J -02 ?03 -0A Tail -o1850 1860 1870 1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 Year Errors from the benchmark model We used each year?s mean global temperature as a forecast of each subsequent year in the future and calculated the errors relative to the measurements for those years. For example, the year 1850 temperature measurement from Hadley was our forecast of the average temperature for each year from 1851 through 1950. We calculated the differences between this benchmark forecast and the Hadley measurement for each year of this IOU-year forecast horizon. In this way we obtained from the Hadley data 157 error estimates for one?year?ahead forecasts, 156 for two?year?ahead forecastserror estimates for lOO-year?ahead forecasts; a total of 10,750 forecasts across all horizons Exhibit 3 shows that mean absolute errors from our benchmark model increased ?'om less than for one-year?ahead forecasts to less than for 100-year?ahead forecasts. Maximum absolute errors increased from more than for one-year?ahead forecasts to less than for lOO-year?ahead forecasts. 2 Obtained from on 9 October, 2008. 3 Exhibit 2 has been updated to include the 2003 figure. Overwhelmingly, errors were no-more?than as shown in Exhibit 4. For horizons less than 65?years, fewer than one-in?eight of our ex-ante forecasts were more than different from the Hadley measurement. All forecasts for horizons up to 80 years and more than 95% of forecasts for horizons from 81 to 100-years?ahead were within of the Hadley ?gure. The overall maximum error from all 10,750 forecasts for all horizons was (from an 87-year? ahead forecast for 1998). IN SEET EXHIBIT 3 Mean and maximum benchmark forecast absolute errors from Hadley temperature data, by forecast horizon 1.1 - ll 1-0 +MaXImum 0_9 Mean 0100 Forecast horizon: Years in the future INSERT EXHIBIT 4 Performance of Intergovernmental Panel on Climate Change projections Since the benchmark model performs so well it is hard to determine what additional bene?ts public policymakers would get from a better forecasting model. Governments did however, via the United Nations, establish the IPCC to search for a better model. The IPCC projections provide an opportunity to illustrate the use of the benchmark. Our intent in this paper is not to assess what might be the true state of the world; rather it is to illustrate proper validation by testing the IPCC projections against the benchmark model. We used the 1992 projection, which was an update of their 1990 projection, for our demonstration. The 1992 projection was for a linear increase of per year (IPCC 1990 p. xi, IPCC 1992 p.17). The IPCC 1992 projections were based on the judgments of the report?s authors and the process they used was not speci?ed in such a way that it would be replicable. We nevertheless used the IPCC projection because it has had a major in?uence on policymakers, coming out as it did in time for the Rio Earth Summit, which produced inter alia Agenda 2] and the United Nations Framework Convention on Climate Change. According to the United Nations webpage on the Summit 4, ?The Earth Summit in?uenced all subsequent UN conferences. . To test any forecasting method, it is necessary to exclude data that were used to develop the model; that is, the testing must be done using out-of-sample data. The most obvious out-of? sample data are the observations that occurred after the forecast was made. By using the 1992 projection, we were able to conduct a longer ex ante forecasting test than if we had used projections from later IPCC reports. Evaluation method We followed the pIOCedure that we had used for our benchmark model and calculated absolute errors as the unsigned difference between the IPCC 1992 projection and the Hadley ?gure for the same year. We then compared these IPCC projection errors with forecast errors ?'om the benchmark model using the cumulative relative absolute error or and Collopy 1992). The is the sum across all forecast horizons of the errors (ignoring signs) from the method being evaluated divided by the equivalent sum of benchmark errors. For example, a of 1.0 would indicate that the evaluated-method errors and benchmark errors came to the saine total while a ?gure of 0.8 would indicate that the sum of evaluated?method errors was 20% lower than the um of benchmark errors. We are concerned about forecasting accuracy by forecast horizon and so calculated error scores for each horizon, and then averaged across the horizons. Thus, the we report are the cumulated sum of the mean absolute errors across horizons divided by the equivalent sum of benchmark errors. 4 Forecasts ?'om 1992 through 2008 using 1992 IPCC projected warming rate We created an IPCC projection series from 1992 to 2008 by starting with the 1991 Hadley ?gure and adding per year. It was also possible to test the IPCC projected warming rate against the University of Alabama at Huntsville?s (UAH) data on global near surface temperature measured from satellites using microwave sounding units. These data are available from 1979. To do that, we created another projection series by starting with the 1991 UAH ?gure. Benchmark forecasts for the two series were based on the 1991 Hadley and UAH temperatures, respectively, for all years. This process, by including estimates for 2008 ?om both sources, gave us two small samples of 17 years of out-cf-sample forecasts. When tested against Hadley measures, IPCC errors were essentially the same as those from our benchmark forecasts 0.98); they were nearly twice as large 1.82) when tested against the UAH satellite measures. We also employed successive forecasting by using each year of the Hadley data from 1991 to 2007 in turn as the base from which to forecast from one to 17 years ahead. We obtained a total of 136 forecasts from each of the 1992 IPCC projected warming rate and the benchmark model over horizons from one to 17 years. We found that averaged across all 17 forecast horizons, the 1992 IPCC projected warming rate errors for the period 1992 to 2008 were 16% smaller than forecast errors from our benchmark as the was 0.84. We repeated the successive forecasting test using UAH data. The 1992 IPCC projected warming rate errors for the period 1992 to 2008 were 5% smaller than forecast errors from our benchmark 0.95). Assessed against the UAH data, the average of the mean errors for all 17 horizons was 0.215 for rolling forecasts from the benchmark model and for the IPCC projected warming rate. The IPCC projections thus provided an error reduction of for this small sample of short-horizon forecasts. The difference of is too small to be of any practical interest. The concern of policymakers is with long-term climate forecasting, and the ex ante analysis we have described was limited to a small sample of short-horizon projections. To address these limitations, we calculated rolling projections ?'om 1851 to illustrate a proper validation procedure. Forecasts from 1851 through 1975 using 1992 IPCC projected warming rate Dangerous manmade global warming became an issue of public concern alter NASA scientist James Hansen testi?ed on the subject to the 1.1.3. Congress on June 23, 1988 (ELM-IEEQMJ after a Iii-year period 00E 1975 over which global temperature estimates were up m?Ol?c than they were down. The IPCC (2007) authors explained however, that ?Global atmospheric concentrations of carbon dioxide, methane and nitrous oxide have increased markedly as a result of human activities since 1750? 2). There have even been claims that human activity has been causing global warming for at least 5,000 years (Bergguist 200 8). It is not unreasonable, then, to suppose for the purposes of our validation illustration that scientists in 1850 had noticed that the increasing industrialization of the world was resulting in exponential growth in ?greenhouse gases? and to project that this would lead to global warming of per year. We used the Hadley data from the beginning of the series in 1850 through to 1975 to illustrate the testing procedure. The period is not strictly out-of-sample, however, in that the IPCC authors knew in retrospect that there had been a broadly upward trend in the Hadley temperature series. From 1850 to 1974 there were 66 years in which the temperature increased from the previous year and 59 in which it declined. There is some positive trend so the benchmark is disadvantaged for the period under consideration. As shown in Exhibit 1, the temperature variations shown by the longer temperature series suggest that there is no assurance that the irregular trend observed in retrospect will continue in the future. We ?rst created a single forecast series by adding the 1992 IPCC projected warming rate of to the previous year?s ?gure, starting with the 1850 Hadley ?gure, and repeating the process for each year through to 1975. Our benchmark forecast was equal to the 1850 Hadley ?gure for all years. This process provided forecast data for each of the 125 years. The warming- rate projection errors totaled more than ten times the benchmark errors 10.1). We then successively used each year ?om 1850 to 1974 as the base from which to forecast from one up to 100 years ahead using the 1992 IPCC projected warming rate and the benchmark model. This yielded a total of 7,550 forecasts covering the period 1851 to 1975. Across all horizons, the projection errors for the period were more than seven times greater than errors ?om our benchmark 7.67). The relative errors increased rapidly with the horizon. For example, for horizons one through ten the was 1.45, while for horizons 41 through 50 it was 6.77 and for horizons 91 through 100 it was 12.6. Discussion We have illustrated how to validate a forecast. There are other reasonable validation tests for global mean temperatures. For example, one reviewer argued that the relevant forecasts for climate change are for decades or longer periods. For decadal forecasts, the appropriate benchmark forecast is that the decades ahead will be the same as the decade just gone. The mean absolute error of a rolling one?decade-ahead benchmark forecast, calculated using the entire Hadley series from 1850 to 2007, was The Mean Absolute Error (MAE) for ?ve decades ahead was and for 10 decades ahead was 0.345 The decadal benchmark errors are smaller than the annual errors. Validation tests should properly be conducted on forecasts from evidence-based forecasting procedures. The models should be clearly speci?ed, fully?disclosed, and replicable. The conditions under which the forecasts apply should be described. Speculation is not suf?cient for forecasting. The belief that ?things have changed" and the future cannot be judged by the past is common, but invalid. The 1980 bet between Julian Simon and Paul Ehrlich on the 1990 price of resources was a high-pro?le example. Ehrlich espoused the Malthusian view that the human population?s demands had, or soon would, outstrip the resources of the Earth. Simon?s position was that real resource prices had fallen over human history and that there were good reasons Why this was so; the fundamental reason being ingenuity. It was therefore a mistake, Simon maintained, to extrapolate recent price increases. Ehrlich dictated the terms of the bet: a ten-year period and the ?ve commodity metals copper, chromium, nickel, tin, and tungsten. The metals were selected with the help of energy and resource experts John Harte and John P. Holdren. All ?ve commodities fell in price over the ten?year period, and Simon won the bet (Tierney 1990). To base public policy decisions on forecasts of global mean temperature one would have to show that changes are forecastable over policy?relevant horizons. and that a valid evidence-based forecasting procedure would provide usefully more accurate forecasts than those from the ?no change? benchmark model. We did not address the issue of forecasting the net bene?t or cost of any climate change that might be predicted. Here again one would need to establish a benchmark forecast, presumably a model assuming that changes in either direction would have no net e?ects. Researchers who have examined this issue are not in agreement on what is the optimum temperature. Finally, success in forecasting climate change and the effects of climate change must then be followed by valid forecasts of the effects of alternative policies. And, again, one would need benchmark forecasts; presumably based on an assumption of taking no action, as that is typically the least costly. The problem is a complex one. A failure at any of one of the three stages of forecasting? temperature change, impacts of changes, and impacts of alternative policies?would imply that climate change policies have no scienti?c basis. Conclusions Global mean temperatures were found to he remarkably stable over policy-relevant horizons. The benchmark forecast is that the global mean temperature for each year for the rest of this century will be within of the 2008 ?gure. There is little room for improving the accuracy of forecasts from our benchmark model. In fact, it is questionable whether practical bene?ts could be gained by obtaining perfect forecasts. While the Hadley temperature data shown in Exhibit 2 shows an upwards drift over the last century or so, the longer series in Exhibit 1 shows that such trends can occur naturally over long periods before reversing. Moreover there is some concern that the upward trend observed over the last century and half might be at least in part an artifact of measurement errors rather than a genuine global warming (McKitrick and Michaels 2007). Even if one puts these reservations aside, our analysis shows that errors from the benchmark forecasts would have been so small that they would not have been of concern to decision makers who relied on them. Acknowledgements We thank the nine people who reviewed the paper for us at different stages of its development and the two anonymous reviewers for their many helpful comments and suggestions. We also thank Michael Guth for his use?JI suggestions on the writing. REFERENCES 15. (2001), Evaluating forecasting models,? Principles of Forecasting. Kluwer Academic Publishers: Boston- J. S., Collopy, F. (1992). Error measures for generalizing about forecasting methods: Empirical comparisons. International Journal of Forecasting, 8, 69?80. Bergquist, L. (2008). Humans started causing global warming 5,000 years ago, UW study says. Journal Sentinel, posted 17 December, Green, K.C., LS. (2007). Wail?nlinu: I1 Energy Environment, 18, 997-1022. IPCC (1990). Climate Change: The IPCC Scienti?c Assessment. Edited by J.T. Houghton, GJ. Jenkins, and JJ. Ephraums. Cambridge University Press: Cambridge, United Kingdom. IPCC (1992). Climate Change 1992: The Supplementary Report to the IPCC Scienti?c Assessment. Edited by .T. Houghton, B.A. Callander, and SK. Varney. Cambridge University Press: Cambridge, United Kingdom. IPCC (2007). Summary for Policymakers, in Climate Change 2007: The Physical Science Basis. Contribution of Working Group 1 to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change [Solomon, 8., D. Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt, M.Tignor and H.L. Miller Cambridge University Press, Cambridge, U.K. and New York, NY, USA. McKibben, W. (2007). Warning on warming. New York Review of Books, 54, 15 March. McKitrick, R., Michaels, P. J. (2007). _IJLIanlih*inu llu: mum-m garnering; and inhuimmgmeitirs Uli??l'iilii?fi elabal climate Lima Journal queophysical Research, 112, Tierney, J. (1990). Betting the planet. New York Times, December 2. ATTACHMENT .vt Astronomy Astrophysics manuscript no. 1 1304 December 30, 2013 2013 Multiple and changing cycles of active stars II. Results K. OlahI Z. T. Granzer2 K.G. Strassmeier2 AJF. Lanza5 S. It'trvinen2'6'7 H. Korhonen3 S.L. Baliunas?, W. Soon4, S. Messina5 and G. Cutispoto5 Konkoly Observatory of the Hungarian Academy of Sciences, Budapest, Hungary; e-mail: olah@konkoly.hu Astrophysical Institute Potsdam (AIP), An der Stemwarte 16, Potsdam, Germany ESO, Strasse 2, 85748 Garching bei Munchen, Germany INAF - Osservatorio Aslro?sico di Catania, via S. So?a 78, 95123 Catania, Italy Thorla Observatory, University of Turku, 21500 Piikkio, Finland 1 2 3 Harvard?Smithsonian Center for Astrophysics, Cambridge, MA 02138, USA 6 7 Astronomy Division, PO. Box 3000, 90014 University of Oulu, Finland Received accepted ABSTRACT Aims. We study the time variations of the cycles of 20 active stars based on decades-long photometric or spectroscopic observations. Methods. A method of time-frequency analysis, as discussed in a companion paper, is applied to the data. Results. Fifteen stars de?nitely show multiple cycles; the records of the rest are too short to verify a timescale for a second cycle. The cycles typically show systematic changes. For three stars, we found two cycles in each of them that are not harmonics, and which vary in parallel, indicating that a common physical mechanism arising from a dynamo construct. The positive relation between the rotational and cycle periods is con?rmed for the inhomogeneous set of active stars. Conclusions. Stellar activity cycles are generally multiple and variable. Key words. stars: activity stars: atmospheres - stars: late-type starspots 1. Introduction In the middle of the last century it was realised that certain ob- served features of late-type stars could be explained by magnetic phenomena (Kron 1947), similar to those that had been detected on the Sun. The cyclic pattern of solar activity had already been known for more than 120 years (Schwahe 1843) by the time systematic research of magnetic aetivity of late?type stars com- menced. In 1966 long?term monitoring of the relative ?uxes of cores of the Call lines thought to indicate the strength and cov- erage of surface magnetism through the enhancement of chro? mospheric [lax in the line cores of solar type stars began (of. Wilson ?963), to search for stellar cycles analogous to the solar case and that monitoring program continued for more than three decades. Eleven years after systematic monitoring had begun (Mlson 1978) published landmark results on the ?rst decadal survey of chromospheric variability and stellar magnetic cycles. While ?Wilson was early in his observational work, a contem? porary comment on the possibility?ofrhe existence of stellar cy? cles appeared in an editorial note of Detre (1971) to a paper pre- senting photometric observations of BY Dra by RF. Chugainov, as follows: The continuous observation of Chugainov's stars would be extremely important because diagrams like that on the opposite page may reveal the existence of cycles similar to the solar cycle in these stars. Studying the late type eclipsing binary XY Geyer (1978) suggested an activity cycle to explain the outside-eclipse variability of the system. Dumey Stenflo (1972) presented a ?rst theoretical forecast, namely, with increasing rotation rate, Send o?print requests to: K. Olah cycle periods should be decreasing. Baliunas et al. (1996b) sum- marized knowledge on the dynamo interpretation of stellar ac- tivity cycles prior to 1996. As databases relevant to decadal stellar activity continued to lengthen and contain more stars, study began on the variability of stellar activity cycles. Multiple cycles, analogous to the known solar multicyclic variability, were recovered by Baliunas et al. (1996b) from the records of the Ca 11 index. Using these Ca in- dex data supplemented with photometric results of active dwarfs and giants Saar Brandenburg (1999) studied time variability of stellar cycles and multiple cycles on the evolutionaly timescale, draw the distribution of Pmle in the function of the Rossby number and compared the results with theoretical models. At the beginning of the current century, several photomet- ric records became long enough to study photospheric magnetic cycles on a relatively large sample of active stars. Using conven- tional methods Olah et a1. (2000, 2002) derived activity cycles and multicyelesforemostiobjects in the present investigation, and Messina Guinan (2002) for six young solar analogues. Radick et al. (1993) investigated the relation between the Ca index and contemporaneous photometric measurements on 35 Sun-like stars. Lockwood et al. (2007) studied the magnetic cy- cle patterns seen in the photospheres and chromospheres of 32 stars primarily on or near the lower main?sequence.1n both those last two papers it was found that on a decadal scale, younger stars decrease in brightness when their chromosperic activity in- creases, while the older, less active stars increase in brightness when chromOSpheric activity increases, as is the case for the Sun. The existence and relation of photospheric and chromospheric cycles are well established. 2 Olah et al.: Changing cycles results On the basis of the variability of the 11-yr solar cycle, which ?uctuates between approximately 9 and 14 years, and the as- sumption that stellar activity shares similar properties to solar activity, one may presume that stellar cycles should also show multidecadal variability. The ?rst attempt to follow changing stellar activity cycles was made by Frick et al. (1997), who de- veloped and applied a modi?ed wavelet technique suitable for data with gaps, and found a variable cycle of one of the targets in the present investigation, HD 100180, from its Ca II index record. Subsequent work of Frick et al. (2004) and Baliunas et al. (2006) used double-wavelet analysis of the records of stars in the Wilson sample to study interdecadal activity variations. Recent efforts have focused on methods to predict solar ac- tivity based on different, earlier observations and on dynamo theory. For critical reviews of those methods see Cameron Schiissler (2007), and Bushby Tobias (2007). The knowledge of the cycle pattern of as many active stars as possible may yield improved insight on the solar activity seen in the context of stel- lar activity. To that end, we have developed a method suited for study of multi?decadal variability from gapped datasets. The method is presented in detail by Kollath Olah (2008, here- after Paper I), where it is applied to solar data and used to re? cover a complicated pattern of multi-scale variability of the Sun in the last few hundred years. In the current paper we apply the method to a sample of active stars with photometric or spectro? scopic records that extend over several decades. 2. Observations 2.1. The sample The most important criterion for the selection of stars for anal- ysis was a long record with few interruptions. In most of the cases gaps in the records are seasonal interruptions, and We re? quire that the gap does not exceed two seasons. Limited gaps are crucial because the method of analysis requires equidistant data, and therefore interpolation during gaps in the data (i between observing seasons) is necessary. The stellar sample consists of 21 objects, of which the Sun is studied in Paper 1. Most of the stars have photomet- ric records, and V833 Tau has a century-long record if aug- mented with photographic photometry; six stars have Ca II index records. The photometrically observed active stars are listed in Table 1 together with their rotational (or orbital) pe- riods, spectral types, vsini, and inclination. The last column lists the extent of the records, in years. Single stars are AB Dor, LQ Hya, V410 Tau, and FK Com (about its possible binarity see Kjurkchieva Marchev 2005), and the rest (V833 Tau, EI Eri, V711 Tau, UZ Lib, UX Ari, HU Vir, IL Hya, XX Tri, HK Lac and 1M Peg) are close binaries. In the case of binaries Table 1 gives the orbital period and for single stars the rotational period. In addition we study six objects from the Wilson sample (Wilson, 1978): HD 131156B, HD 100180, HD 201091, HD 201092 and HD 95735. In our analysis, those targets are considered to be effectively single stars, though HD 131156A with HD 131156B and HD 201091 with HD 201092 form wide, visual binaries. Their rotational pe- riods, spectral types and length of the Ca index records are listed in Table 2. 2.2. Mum-decade! photometry The photometric records contain all published material for each star in the sample, in V?colour. Those stars are well studied, Table 1. The stellar sample. I. Photometric observations. star rot. per.? sp. type vsini i time-base (days) (km (years) AB Dor 0.515 KOV 91 60 18 LQ Hya 1.601 K2V 27 65 25 V833 Tau 1.788 KSV 6.3 20 20 photographic photometric data 109 V410 Tau 1.872 K4 74 70 34 E1 Eri 1.947 (351V 51 46 28 FK Com 2.400 G4l]1 155 60 28 V711 Tau 2.838 41 40 30 U2 Lib 4.768 K0111 67 50 17 UX Ari 6.437 39 60 23 HU Vir 10.388 25 65 17 11.. Hya 12.905 26 5 55 20 XX Tri 23.969 K0111 21 60 21 HK Lac 24.428 K0111 20 65 50 IM Peg 24.649 K2111 26.5 70 29 The stellar parameters are from Strassmeier (2002), and the references therein. in case of binary the orbital period is given Table 2. The stellar sample. 11. Ca index measurements. star rot. per. sp. type time-base (days) (years) HD 131156A 6 B00 A 6.25 GSV 35 HD131156B =?BooB 11.1 35 HD 100180 38 Leo 14.3 GOV 34 HD 201092 61 35 5 KW 35 HD 201091 61 A 36.1 K75 35 HD 95735 G1 411 54.7 WV 34 with numerous papers describing the records. Recent summaries for the stars are as follows: AB Dor: Jarvinen et al. (2005), LQ Hya: Berdyugina et al. (2002) and K?vari et al. (2004), V833 Tau: Olah et al. (2001), V410 Tau: Strassmeier et al. (1997b), EI Eri: Strassmeier et al. (1997b), FK Com: Olah et a1. (2006), Tau: Lanza et al. (2006) and Strassmeier et al. (1997b), UZ Lib: Olah et al. (2002), UX Ari: Aarum?Ulvas Henry (2003), HU Vir: Strassmeier et al. (1997b), IL Hya: Strassmeier et al. (1997b), XX Tri: Strassmeier et al. (1997b), Lac: Olah et al. (1997), IM Peg: Ribarik et al. (2003). For several stars the records were updated through 2007?2008 using new photometry from the Vienna APT (Strassmeier et al. 1997a). For further details, see the references of the cited papers. Periods of sparsely-sampled data leading to large gaps, some- times seen in the onset of records, were disregarded. 2.3. Ca II index measurements The Ca 11 measurements were collected at the Mount Wilson Observatory, with two sets of equipment on two different tele- scopes, from 1966 to 1978 and after, for details see Vaughan et a1. (1978). The relative Ca emission is the ?ux ratio of two 0.1 nm passbands centred on the cores of the and lines and two 2 nm passband in the nearby continuum. 01a. et a1.: Changing cycles results 3 0.6 0 .4 - 0.2. a unth sum 0.020 0.025 emu A(noise)lA(signal) Fig. 1. False Alarm Probability that a structure appears in STFT when Gaussian noise, with different amplitudes, is added to the record of LQ Hya. Five di?erent standard deviations were used and the results are plotted from left to right for a-rl 0.01 0.02, 0.03, 004,005 mags. Details are in the text. 3. Method We apply a time-frequency analysis called short?term Fourier transform to the photometric and spectroscopic records to study time variability of the activity cycles that occur in the sample of twenty stars. As stated earlier, the analysis requires equidistant data, and therefore we had to interpolate through gaps among observations. To interpolate, we used a smoothing spline. Before calculating the spline interpolation, we removed the rotational signal and the next four strongest components in the periodogram near the rotation period; the removal of the next four strongest signals addresses removal rotation signals that are non-sinusoidal in shape or other activity changes on similar time scales from the record. Because rotational modulation is sub- ject to change owing to differential rotation, removing the re- tational signal (and the additional signals of similar timescale) was done separately for subsets in the record, typically around 200 days long, which were then averaged for the spline inter- polation. We argue that such a procedure is necessary, because unevenly sampled rotational modulation may alter the average seasonal light level, and consequently cause a false signal in the analysis for multidecadal variability. This step is especially im- portant for stars with high amplitude rotational modulation of large amplitude, arising from, high axial inclination to our line of sight. The FWHM of the Gaussian used in the time series is usually ~180 days. Note, however, that in the actual calcu- lation we use ?ltering in the Fourier space instead of temporal ?ltering. The method is explained in detail in Paper I, where tests of the eifects of seasonal gaps, rotational modulation and observational errors are also found. The effect of active region growth and decay, as discussed by Messina Guinan (2003) has little in?uence on our results because we removed the rotational frequencies plus four other signals of similar timescale, thereby signi?cantly reducing the amplitudes of the modulation due to rotation, in most cases. Moreover, cycles below about 1 .5-2 years are not considered sig- ni?cant, as was discussed in Paper I, and below. To check the effect of added Gaussian noise on false fea- tures in the time-frequency diagram, we performed a Monte- Carlo simulation. Here we present the results of that test on the record of LQ Hya as an example; data of other stars in the sam- ple are of similar quality. First Gaussian noise was added to the original observational record. We performed the same prepro- cessin (averaging and spline-smoothing interpolation) as for the standard processing of the actual records. The of the dif- ference between the noisy and the original smoothed data was calculated, and the statistics of the maximum amplitude (largest peak or ridge) was estimated from 10000 cases. Fig. 1 shows the False Alarm Probability e.g. the probability that a struc- ture with amplitude larger than 2. appears in owing to noise (A(noise), and A(obs) is the maximum am- plitude in is the inverse of the signal-to-noise ratio). The test was performed with different standard deviations of the added Gaussian noise. On the ?gure from left to right the curves belonging to 0.01, 0.02, 0.03, 0.04, 0.05 mag are displayed. In case of Gaussian noise ~0.01-0.02 mag which is poorer precision than typical photometric precision - the proba- bility of false signals with amplitudes over 0.01th of the highest signal is less than 10% and fast decreases toward larger astro- physical amplitudes. 4. Results of the time series analysis from photometric data From analysis of the records we found changing and multiple cycles for most of the stars. Below we present a short descrip? tion of the cycle pattern for each object. The records and results are plotted in Fig.2. In the time?frequency plots darker colours mean higher am- plitudes. The signals are modi?ed by different ampli?cation fac- tors that are employed to make the smaller amplitude signals better visible (see Paper I for more). The ampli?cation factors are given electronically in Table 3. It is not straightforward to estimate the signi?cance of the signals - moreover, no indepen- dent measure can be given. We consider a cycle or cycles signif- icant if at least two of these three features are true: the signal runs throughout the period of the observations, it has high amplitude or when two or more components are changing parallel. Cycles shorter than about 1.5-2 years are considered to be insigni?cant: we showed in Paper I that from datasets having yearly gaps such cycles cannot be recovered safely. On the other hand, a long-term cycle is not signi?cant if its length is commen- surable to the time span of the observations, when the data do not cover approximately two full cycles. AB Dor. The record of AB Dor is not well sampled, the number of observations per year is low and gaps longer than two years occur twice. The only clear signal is a cycle ~3.3-yr. The other, weaker signal of ~2-yr appears in the beginning of the record, and is not considered reliable, and possibly owes to undersampling. Variability of high amplitude is reported on the time scale of ~20-yr (Innis et a1. 2008), but the record is not long enough to verify it from our analysis. LQ Hya. This star is a fast rotating, single star, one of the most interesting targets. The light variation is well-sampled by the measurements and all gaps are less than one year long. Previously, cycles between 11.4-11.6 years, 6.8?6.5 years and 2.8-3 .2 years were determined by Olah et a1. (2000, 2002) from observations spanning a shorter period, and more recently, cy- cles of and its harmonic 69:03 and were found from long term photometry by Kt?ivari et a1. (2004). The present analysis recovers two short cycles ?25 and 3.6-yr, the latter being stronger in the middle of the dataset. Another cycle ~7-yr is present, and that period increases continuosly to 12.4 years with the same amplitude, while a small amplitude yr signal remains. V833 Tim. The short (20-yr) photometric record of this bi- nary was supplemented by archive photographic data (Hartmann 4 Olah et Changing cycles results AB Dondns dhl?] LO "Jill-ll Phil-)slml days [gl-_ Ji_ 1i - I 5 11? on? ?tum-Th? '?umm 3 name urn?1 lt' . an ill?)! will}: Tillie-Minna ?ue-frequency dim-[hum {told}! I ([ddl fldd] I ,n sea-J - i are. LN ??uo no nuns? In ~11! emt? 2.1- 11: 5?1- am 5.Jamil ?um? we ?ll? till-tum ll "Esme 5 Lu?: 5? V833 Tl?l'l I. I?ll: 1110 m3 '1"an II. (IBM I347 day! I. i (I .34? 5-1.?at: sun! 'n r?mm??m TmT??an :5 IiIll?)! ?in?ll [5111']: (SITE): The-frequency (51TH: an Item - 1.11 . 1n nuI7 :? am:- can. "an Ill will 7 am I- mun" l? - l' Imp?stuns Twig im? Hum-22mm cum IllJul Evil i . '1 ?mm. It.? 53!; ii ?n iini?? ?Jame one In search unites man; 11m (MIT): '11an (Sl?Fl?) 'l?f-listom- ?unitamn Hon me I Fig. 2. Results of the STE-T analysis. In the upper panel for each segment for each star} the observational data (grey,hlue in the electronic version). the data prewhitened with the mean rota the electronic version) and the corresponding spline interpol of the data subsets are given in the top line of each segment. colours denote higher amplitudesThe amplitudes are also mo LQ Hya. V410 Tau. Middle: V333 Tau photometric data, V71 Tau. UZ Lib. Details are in the teat. ations et a1. 1981), thereby extending the dataset to 109 years. We analysed separately the well sampled short and more sparse, long records; the long one includes also the photometric data. The short photometric record shows two modulations, with the weaker ~2.2?yr long and the stronger Both are tional periods and its ?ve harmonics for each data subset (black, red in {solid grey (blue) and black (red) lines) are plotted. The The lower panels show the time~frequency distributions. where darker di?ed by the ampli?caIIOn factors (see on-line Table 3). Top: AB Dot, V833 Tau photographic+photometric data. El Eri. Bottom: FK Com. present throughout the record, which agrees with our previ? ous results (Olah et al. (2000, 2002). The rotational modula? tion has a small amplitude (few hundredths of magnitude) re- sulting from the low inclination of the star (z This low amplitude of rotation modulation is an important fact when we Olah et a1 .: Changing cycles results 5 UX Arle?s trot.) - 6113711139 d1: I80 leL): [0388 ch}! Ill: 20!! Punt.) 12.90511?: Ill: "0 i54__n1 I A II Eo?tlrg?gi it] .ggnii?g?gg??tn??. lawm?ssm mu ?ne 5: Ina? 5 ?irun? gWinn?'? on inlan? ?is'aL-ah? um um: um anon] The-mm Time-frequuy distribution Time-Mum distribution 1r.u i Hull out}. :31 em: LI: um.- nmm- mm. In nuns;- I "tun l??Tnansr? -- 1.will?l .I?ibn_ _lnil?lln ?ll 41 :Immn 51mm ill-lung 'm?Ji?m she?b? Hillel- uncum u?mcIHi'Ll ?Trilng?ll P(ml)= 13.969 dl? dl: IE) [.noerlne lelJuu-Am days 320 IM Pepsi days I11?411.? i 'l grif?nin.? I Il1-.- .. 1 man my sienna ?ms?mm mnu nuqu ?um I I 4" In zu'rnn Stu-nu ulna nu Ila-r n. lime in] Tlrne?l'requmcy distribution Time-Ira;qu distribution IT): Tler ?in HM rn Nd?l 1 um: ?I.u up": --, undul- In] sum 7 Lu umli? mm in mm 11'] I Immi? 1mm -7 I .Wl I- Iii/t! I . l? _h "Its-?r urnn? 4mm ?Inn: "Lorri-m Fig.2. (cont.) Results of the STFT analysis. Top: UX Ari, HU Vir, 1L Hya. Bottom: XX Tri, 1-K Lac, Peg. Details are in the text. study the century long dataset from the archive photographic measurements, even though only 1-2 points/year are available in Hartmann et a1. (1981). However, those data represent the stel? lar brightness, within the usual error of the photographic mea- surements of ~0.1 mag, because the rotational modulation itself has a lower amplitude than this value. Thus, a cycle amplitude above ~0.l mag is well documented by the photographic record, and is considered real. A long-term modulation is present dur? ing the 109?yr period of observations, with a large amplitude of ~0.9 mag. A long cycle of ~27-30-yr is also recovered. V410 Tau. The only Tauri-type star in the sample, V410 Tau shows rotational modulation with a large amplitude (often mag). However, its mean brightness has re- mained within ~0.2 mag during the 34?yr length of the record. The analysis reveals a small-amplitude cycle of ~6.5?6.8?yr, which abruptly changes to 5 .2-yr near ID 2449500 and afterward slowly decreases in period. The long-term trend seems to follow this decrease. Our result supports that of Stelzer et ai. (2003), who found ~5.4?yr cycle in yearly mean magnitudes after 1990 (~2448500). While Steltzer et at. remarked that the cycle had been out of phase in the earlier observations, we suggest a dif- ferent, longer, and slowly variable cycle length in the beginning of the record. EI Eri. In the recently available record covering 28 years, two short cycles of and are present. They smoothly and in parallel increase and decrease; the longer one has a high amplitude between JD 2446000-50000. A long cycle of ~l4?yr with variable amplitude is also seen. TWO cy? cles of ~2.4 and ~16.2-yr had earlier been found by Olah et a1. (2000). Subsequently, using a longer record, Olah et a1. (2002) con?rmed the ~2.4?yr cycle and revised the longer One to ~122- yr. The ?rst determination of the long cycle was 11:1:1 years by Strassmeier et al. (1997b); all determinations of the long cy- cle point to an approximate decadal cycle. Rotational periods of El Eri in different seasons has been investigated in detail by Washuettl et a1. (2008), and no correlation is found between the seasonal periods (or multiperiods owing to di??erential rotation of 2-3 active regions at different latitudes for most seasons) and the cycle pattern. FK Com. One dominant cycle is present in the record; it changes smoothly betWeen 4.5 and 6.1-yr. This is consistent with the earlier result by Olah et a1. (2006), who found quasiperiods of about 5.2 and 5.8?yr in the longitudes of starspots, based on a record of 18 years, between JD 2446800-53200 (1987?2004). V711 Tau. The 30-yr record has the best observational cov- erage arnon our sample. The observations themselves show two waves, of ~18 and 9-yr, and they are clearly de?ned in the time? frequency diagram. Apart from those two periods, a cycle of mini?yr is present throughout most of the record. A short, weak cycle of ~33?yr is also apparent from JD 2445000 onwards. Several papers give estimates of the cycles of this well-obserVed system, and all are in accordance with the present results: Henry et a1. (1995) suggests a cycle of ~5.5i0.3?yr and some evidence for a long one of ~16:I:l-yr from photometry, Vogt et a1. (1999) 6 Olah et Changing cycles results reported short cycles of ~3.0:t:0.2 and for the po- lar spot area and for the low latitude spots from Doppler im- ages. Lanza et a1. (2006) using a longer record of photometry found a cycle of ~3-5-yr with variable amplitude and a longer cycle, 1 Berdyugintt 8: Henry (2007) using an inver- sion technique on photometric data between 1975?2006 found ~5.3:I:li.l and 15-16-yr cycles. Taking into account all these in- dependent results both from photometry and spectroscopy, the cycle pattern of V711 Tau is well documented. UZ Lib. The cycle length slowly varies from 4.3 'to 3.1-yr. Earlier Olah et a1. (2002) had found a cycle length of 4.8-yr. The long-term signal near lS?yr is comparable to the length of the dataset itself, and thus cannot be established as a cycle. UX Ari. Its record contains a signi?cant cycle of ~4.6-yr that deacreases to ~3.6?yr by the end of the record. A long-term signal seems to decrease in parallel with the shorter cycle, from ~25 to ~15 years: however, that longer period timescale is com- parable to the length of the record, and is thus not considered sig- ni?cant. On the other hand, the fact that the long cycle changes in parallel with the short one strengthens its reliability. We note that Aarum-Ulvas St. Henry (2003) report an activity cycle of 25-yr on UX Ari. HU Vir. The dominant cycle length of this star is ~5.5-6-yr, in agreement with the values found by Olah et al. (2000, 2002). The shorter-period signals are considered to be insigni?cant. IL Hya. A long gap (of approximately 4 years) occurs in the beginning of the record . which makes the early part of the record unacceptable for our analysis. Hence we considerably shortened the record for study here; compared to the previous studies of Olah et al. (2000, 2002, 2007), we are unable the study of the 13- yr cycle found earlier. However, we con?rm the previous result of Olah et al. (2007) on the changing nature of the short cycle (from 3.5 to 4.3-yr in that paper). The present result shows that a ~4.4-yr cycle persists throughout the 20-yr record, and a short cycle, increasing from ~27 to ~3 yr also appears. XX 'Ii'i. Apart from a long trend, only a weak ~3.8-yr cycle is present with signi?cant amplitude only in the second half of the record. HK Lac. The ?rst cycle of this star were estimated by Olah et a1. (2000) as ~6.8 and ~13-yr. Fr'dhlich et a1. (2006) extended the dataset with measurements from Sonneberg Sky- patrol plates, and not only con?rmed the previous results but also found a third cycle with a length of 9.65-yr. The origin of the third cycle lies in the changing cycle of HR Lac, as was ?rst communicated by 01511 (2007). The very beginning of the record was omitted from the present analysis becasue of scarce sampling. The Sonneberg results start well before the photomet? ric measurements, thereby lenghtening the record to 50 years; the photographic measurements partly overlap the photometry and ?ll some gaps. The present results, in agreement with all previous ones, shows a cycle varying slowly between ~5 .4 and ~5 .9?yr, together with the cycle increasing from ~10.0 to ~133- yr. IM Peg. One cycle of ~9.0-yr is present, in accordance with the ~10.1-yr cycle found by Olah et a1. (2002). A long-term sig- nal of 18.9 years is not considered signi?cant, because it is ten years shorter than the record. 5. Results of the time series analysis from Ca II index data Studying stellar activity cycles was one of the main aims of the Wilson project Wilson 1978) and from the Ca 11 index Table 4. Rotational periods from Ca index records. Star rotation periods in days (1) (2) (3) (4) (5) HD 131156A 6.25 6 6.31 8.2 HD131156B 11.1 11 11.94 HD 100180 14.3 14 38 HD 201092 35.5 35 37.84 36 45, 51 I-ID 201091 36.1 38 35.37 36 44, 47.5 HD 95735 54.7 53 28.5 (1) present paper, (2) Baliunas et al. (1996a), (3) Donahue et a1. (1996), mean periods, (4) Frick et a1. (2004, (5) Baliunas et a1. (2006) records, fundamental results on decadal variability have been published. Ours is not the ?rst attempt to ?nd not only cycle but also rotational periods from that dataset. However, we wished to compare, using the method here, the results from photometry of stellar photospheres with those from Ca 11 spectroscopy, which re?ects the variability of chromospheres. 5. 1. Rotational periods The ?rst thorough analysis of rotational modulation from the Ca II index data was carried out by Baliunas et al. (1996a). That work is based on records of 112 stars, among them are all the six objects of the present investigation. All rotational periods found by Baliunas et al. (1996a) agree well with the present re- sults, which is not always the case in later studies of other stars in the sample. Fig. 3 shows the period analysis of the six stel- lar records, all of which were prewhitened with the long-term variability (trends and cycle periods). The results are presented in Table 4 together with periods found in the literature. In all cases typical patterns arising from Fourier analysis of non-stable signals are seen: differential rotation of the stars produces signi?cant power at frequencies around the main rotational fre- quency of the amplitude spectra. We split the records into three nearly equal parts in time and repeated the period search. The results are dillcrent among the three sets, primarily because of the effects of differential rota- tion, but short lived structures appearing and disappearing at dif- ferent positions may modify the derived periods as well. The values of periods detected in the subsets of the records are given in the electronic Table 5. Taken into account these results plus those from the literature We estimate uncertainties in the mean derived rotational periods of at least a few tenths of a day in case of shorter periods, and about 1.5-2 days for stars with rota? tional periods of a few weeks. Seasonal periods determined by Donahue et al. (1996) show a larger range of periods than seen in our analysis because of a multi-seasonal average in our anal? ysis of several seasons combined as one subset. Baliunas et al. (1996a) and Donahue et a1. (1996) used datasets 10 years shorter, and Frick et a1. (2004) and Baliunas et al. (2006) the same length records as we use in the current paper. In Fig. 3 (lower panel) we show two examples, HD 131156A and HD 131156B in which the amplitude spectra of the entire record and part of it yield di?erent periods. Note that in the presence of strong differential rotation but without knowing the rotational pro?le and the incli- nation of the star, it is nearly impossible to derive a precise mean rotational period. Olah et 31.: Changing cycles results 7 1m 1252:): Ilmurl 1 . HD 100180 [errEiLlnm-t'. dalml taxis-r: "are; - a . l?iJJq?i?-il: i i 1 Ill5mm. I an nu um can an! Cull Index (dLlM Call 1min (slut) on mutual. wail-111nm dulnul norii. ?fg. 3 13043.05 - i. 1, i cm man ram.) cm 1314!: ram.) l-[D 201091. dnLnacl. HID 15111002 [urn-?hurried dill-gay. is fie-i=1. EL. I1: g: i Few-din. th?lril. :lit -1. 1?:1 '1 3 :5 +42. I I I 4m Haul am Index (arm a: 11-1: ?c .I . 201.1, J-ni . v. If, - 1:300 ?out HIM. . mu 1: arm: Call hide: (lull? Indus (dLl'lJ Fig.3. Rotational periods from Ca II index data. Upper panel, from left to right: HD 95735, I-ID 100180, HD 201091, 11]) 201092. Lower panel, from left to right: 1131156A and detail, HD 1311563 and detail. Each segment consists of three panels: top: data prewhitened with decadal-scale variatious, middle: power Spectra with the rotational period marked, bottom: data folded on the derived period. 5.2. Time-frequency results Below we present the new results of the cycles of each star indi- vidually, and compare them with the previous results. HD 131156A {5 B00 A. Variability on two time scales are present. The shorter cycle is ~5.5?yr and exhibits a high ampli? tude in the beginning of the record; the cycle length decreases to ~3 .9-yr at the record?s end. A longer?period variability is also apparent, with a characteristic timescale of ~11 yr in the middle of the record. Baliunas et a1. (1995) marked the decadal variabil? ity of the record as "var," which means "signi?cant variability without pronounced periodicity. on timescales longer, than l-yr but much shorter than and this is in accordance with our results for the record that is ten years longer; also, Baliunas et a1. (2006) ?nd a long cycle of 13.2 years. HD 13115613 15 B00 B. Only one long?term periodicity is found, and only in the beginning of the record, approximately ~4.3-yr, although a secular change is also evident. Baliunas et a1. (1995) remark as ?long?, signifcant variability is found, which is longer than 25 years. HD 100180 88 Leo. We derive a more or less constant cy- cle of ~3.5-yr, which persists nearly through the entire record, but whose amplitude is strong in the ?rst half of the record and decreasing thereafter. Another variable cycle of ~13.7-yr appears in the beginning of the record; the period decreases to ~8.6-yr by the end of the record, in agreement with the earlier results of Olah et a1. (2007). Our results in the beginning of the dataset are similar to those found by Baliunas et al. (1995), who found two cycles, ~3 56 and ~12.9-yr in the record shorter by ten years. Analyzing the full record, Frick et a1. (2004) found 3 Olah et 211.: Changing cycles results days dam HDIJIISGIIHGJ 5663 days dam HD days dull?11? . 5 51ml . .- . . . t; gun I- I - I rim-is . :1 E'lil I grin 3w} in?, If H: 5.1 ELI. - sma? swims l-Iiuz'l' sit-1m ??asr:u 1 ??Iaiimu su'mp Him?f twain-T "nra'ni?u mun] u-Ilm Time-frequency Thu-[mum dIsIrlbulJon dim-Ibullm nun? Han H'th I mm- - v.1! um:? - ha:- cw. NT umll- ?nuts I31 11cams 5.1! mm: - 5n l1913?? -L- . . . _.1rrl' mm mm: ?mun mu 42mm: mu "minimal: limo tuna mm 20l?92 =61Cyg? (rum :35: days Ill: [60 HD IDIIDI =61 A thy! dl: 160 95735 MI 1? days I-Intll?'13 1 33:1 n? - 5a; ?n ?Ilum TlliJu?' (?lial: a. lil' Tim-T" 4111?? ii line on.) um: In mu unl Time-Mu!an [m ?rm-frequency new. I 1m n- 'm LII um}. 11ml. . Iier 7 Is: amulr - - LBJ ?mm. I 'm I I I null? I nrn1?Jami 111' ?k mien 4.1mm ?Jim I mi?isrliis?mr u? litur- T?n?a 1?i?wp I Fig.4. The STFT of the Ca II sample, panels and colour codes are the same as in Fig. 2. Top: I-ID 131156A, HD 131156B, HD 100180,bottom: HD 95735. Details are in the text. periodicities of ~3.7 and ~10.1-yr. The short cycle is consistent with our result, the ~10.l?yr value may be an averaged mean value of the of the long variable cycle we detect. HD 201092 61 B. The record for this star also ex- hibits two activity cycles: one is ~4.7-yr and persists thoughout the record; the other has a timescale of ~10-13 years. Both peri? odicities have variable amplitudes. The long cycle was estimated in previous studies as 11.4, 11.1 and 11.7-yr by Baliunas et a1. (1995), Frick et al. (2004) and Baliunas et a1. (2006), respec- tively. HD 20109] 61 A. The high amplitude cycle seen in the record of this star has a mean length of 6.7-yr, which slowly changes between 6.2 and 7.2-yr. A shorter, signi?cant cycle is found in the ?rst half of the record with a characteristic timescale of ~3 .6?yr. Probably because of the changing and double nature of the main cycles, previous determinations are sligth dilferent but still consistent with our result: Baliunas et a1. (1995) derived 7.3-yr, Frick et a1. (2004) found 7.12-yr and ?nally Baliunas et (2006) estimated 5 .43-yr. However, the most interesting com? parison to our results forI-ID 201091 is that of Frick et a1. (1997, their Figure 5, bottom left panel, and Figure 6, lower panel). They derived a changing cycle between ~6.7 and 7.9-yr, while the record at that time was about ten years shorter than now. Comparing Figure 5 of Frick et a1. (1997) to Fig. 4, (bottom, middle), it is seen that the results are similar, though, because of our higher frequency resolution we could resolve two cycles in the beginning of the record. HD 95735 GJ 411. The stronger cycle of this star is ~3.9-yr, which is shorter (3 .4-yr) in the beginning of the record. A longer, ll?yr cycle is also present with a smaller amplitude. Baliunas et al. (1995) classify the record as ?Var? (see the explanation at HD 131156A). The long cycle was es- timated as ~1 1 .3-yr by Baliunas et a1. (2006). The rotational pe- riod found by Baliunas et a1. (2006) is 28.5-d and by us 54.7-d; the shorter result possibly arises from the non-sinusoidal nature of the rotational modulation caused by two distinct activity cen- tres at spatially separated stellar longitudes. 6. Discussion 6.1. Cycle The results of our analysis are summarised in Table 6, where the cycle (yr, and in the electronic Table 7, in d) are listed in order of increasing value. In cases of changing cycle periods, arrows show the direction of the change between the extreme values (increasing, decreasing, or both). An denotes a multi- decadal variation, which may occur in addition to the detected cycle(s); however, a timescale cannot be giVen because of the insu?icient length of the record to resolve the longer period. Most of the stars show multiple cycles. Those stars for which we list only one cycle usually show variability on longer timescales. The of the records are too short to verify ad- Ol?h et Changing cycles results 9 Table 6. Derived activity cycles. star Rotation period" Cycle (days) (years) AB Dor 0.515 3.3, LQ Hya 1.601 2.5, 3.6, 12.4 V833 Tau 1.788 2.2, 5.2, 27-30, V410 Tau 1.872 EI Eri 1.947 2993.1, 4.1H4.9, 14.0 FK Com 2.400 V711 Tau 2.838 3.3, 5.4, 8.8c?l7.9 UZ Lib. 4.768 3.1e?4.3, UX Ari 6.437 3.6c?4.6, HU Vir 10.388 5.7, 1L Hya 12.905 2.7?8.0. 4.4, XX Tri 23.969 3.8, HK Lac 24.428 5.4?6.9, 10.0?r13.3, 1M Peg 24.649 9.0, HD 131156A 6.25 11.0 HD131156B 11.05 4.3.L HD 100180 14.32 3.5, HD 201092 35.53 4.7, 11.5 HD 201091 36.06 3.6, HD 95735 54.7 3.4H3.9, 11.0 The cycle given in years, are listed in order of increasing value. For changing cycle periods, arrows show the direction of the change between the extreme values.?L? denotes multi-decada] variations. in the case of a binary, the orbital period is given ditional, longer cycles for AB Dor, I-IU Vir, XX Tri, IM Peg and HD 131156B (see Table 6). The cycle are not strictly constant or randomly vari- able, but show systematic changes during the period of obser- vations. We see two kinds of systematically changing cycles. The ?rst is characterized by a cycle length that oscillates be- tween a lower and an upper value, and the second by a cycle length exhibiting a secular variation, an increase or a de- crease. However, these continuously increasing or decreasing cycle may reverse over a longer observational period than currently available. In Paper I we studied the multi?decadal variability of the solar Schwabe and Gleissberg cycles during the last 250 years from Sunspot Number records. In Figure 2 (bottom panel) of Paper I we used the same method as for the stars in the present paper. The results look similar for the Sun, one cycle (Schwabe) varies between limits, while the longer one (Gleissberg) continually increases. However, the results for a longer time displayed on Figure 8 in Paper I indicated that the Gleissberg cycle of the Sun also varies between limits on a scale of ~50-200-yr, but to con?rm the variability of the Gleissberg cycle required a longer, 500-yr record. By analogy from the anal- ysis of the lenger solar record, the presence of a long?term trend may suggest an increasing or decreasing multidecadal cycle that is presently unresolved in the stellar records of short duration. Another feature of interest seen in our analysis of cycle in stellar records is that of the dominant (highest am- plitude) cycle rapidly switching to another one. In El Eri in the beginning of the dataset the long, ~l4?yr cycle is dominant, and the short cycle of about 4.5?yr is only weakly present. Around ID 2447000 the amplitude of this short cycle increases consider- ably, and that high-amplitude cycle persists for more than 12-yr short cycles). Later the longer cycle again dominates. (The beginning is uncertain because of sparse data, but the end of the record is well sampled.) A similar pattern is seen in LQ Hya. Interesting feature is found in V410 Tau, EI Eli and UK Ari: two cycles in each of them which are not harmonics of each other, are changing parallel. This ?nding points towards the com- mon physical origin of these cycles, that both cycles are pro- duced by the same dynamo. 6.2. Relations The cycle of two stars were excluded from the fur- ther analysis. The ?rst is the shortest period, ullrafast rotating (Pm 0.459 Ml-2 dwarf star EY Dra, which might show a cycle of around 1 year from observations made every two months with gaps of only 4 months (see Vida 2007), but that cycle is uncertain. The other is the star with the longest rotation period in our sample, HD 95735 54.7 a dwarf of spec- tral type M2, which is rotating ~120 times slower than EY Dra, but has the same spectral type. Those two stars are the only dwarfs which have reported activity cycles, and have the short- est (EY Dra) and longest (HD 95735) rotation periods, which are plotted in Fig. 5 and Fig. 6, but are excluded from the ?ts. Their internal structure may differ signi?cantly from the other stars in the sample. Graphical representation of the results are given in Fig. 5 on a log? log scale of l'i'cyc/Pmr vs. 1 [Fran as in the paper of Baliunas et al. (1996b, Figure The cycle period is expected to scale where is the slope of the relation and is the dynamo number. The slope for the results in Baliunas et al. (1996b) was 0.74 (dotted line in Fig. 5), and we get 0.81 d: 0.05 for all the re- sults from the present paper and from Messina Guinan (2002), and Frick et a1. (2004). The ?t to the shortest cycles determined by us gives a slope of 0.84 0.06. The log log representation of the cycle period normalised by the rotation period gives a direct relation to the dynamo num- ber. In comparison, the direct representation of the cycle vs. ro- tational periods in Fig. 6 has information about the tightness of the relation that is in Fig. 5 suppressed a little by the normaliza- tion and the scale. All symbols of Fig. 6 are the same as in Fig.5. The Gleissberg timescale of the Sun is not plotted, because even its shortest value (SO-yr, about 18000-d) would compress the ?gure along the y-axis. The slope of the relation for all the results of the present paper together with the values from the lit? erature is 39.7 10.6, whereas for the newly determined shortest cycles is 27.1 12.1 the relation is signi?cant for all the data considered together, and marginal when only the shortest cycles are considered. However, an important warning should be recalled from Paper 1: cycle periods of l-2?yr recovered from records with approximately yearly gaps are highly uncertain if not insignif? icant. Hence, the lower end of the diagrams in Fig. 5 and Fig. 6, at the shorter rotational periods, the shortest cycles simply cannot be recovered with certainty, thereby distorting the ?tted relations. Finally, we note that the stars studied in this paper and those from the literature that are plotted together in Fig. 5 and Fig. 6 represent a wide range of objects in the context of spec- tral types, characteristics of the observational data (photometry, spectroscopy, of the records), and that the sample con- sists of both single stars and stars in binary systems. Therefore, while we ?nd a general trend between the rotational periods and cycle timescales, a much tighter correlation cannot be expected. 10 Ol?h et al.: Changing cycles ?results -. -1.5 ?1 ?0.5 0 Fig.5. The quantity as in Baliunas et al. (1996b). The plot includes the results of the present inves- tigation, and also results from the literature: Messina Guinan (2002, circles), and Frick et a1. (2004, crosses). The results from the current photometric investigation are marked by ?lled dots, and ?lled triangles denote the results from the Ca 11 records (HD 95735 is excluded from the ?ts and the mean of its chang- ing cycles are plotted). The dilferent cycles of the Sun are repre- sented by stars. A hexagon in the far right shows the position of the fast rotator EY Dru [excluded from the The dotted line shows the original relation from Baliunas et al. (1996b), dashed line is the ?t to all results, including those from the literature, and the solid line is the relation for the shortest cycle determined by us. 7. Summary We analysed continuous photometric and spectroscopic records on multi-decadal periods of 20 active, cool stars using Short- Term Fourier Transform and studied the patterns of their activity cycles. We found that 15 stars show de?nite, multiple cycles, the others probably do, most of the cycle are found to change systematically, in three cases two cycles are changing parallel, and the relation between the rotational periods and cycle is con?rmed for the inhomogeneous set of active stars consisting of dwarfs and giants, single and binary systems. Moss et (2008) depict a complicated picture of symmetry properties derived from models of stellar magnetic ?elds. while expecting that properties of observed stellar cycles may offer progress on the problem. The present paper gives twenty differ- ent examples together with the Sun from Paper 1) of multiple cycles that can arise from "rt?ynamos operating in signi?cantly super-critical regimes The observed modulations in amplitude and cycle may originate from cyclic quadrupolar com- ponents superimposed on a dipole cycle. as suggested by Moss et' al. (2008). The predictability of the cycle patterns at the mo- ment is very limited: we are able to describe from observations with mathematical tools what happened in the past-yrs a Sr - . .0 I: 4000I1. 2000- 0_ I I I Fig.6. The rotational period-cycle period diagram, symbols are as in Fig. 5. The dashed line is the ?t to all results including the ones from the literature. and solid line is the relation for the shortest cycle determined by us. future only trends and tendencies can be suggested. Further de- tailed theory is necessary to model the complicated cycle struc- tures we observe in stars and in the Sun. Thorough modeling the observed cycle patterns may be for the present the only way to predict with success the future patterns of the Sun and the stars. Acknowledgements. K.O. acknowledges support from the Hungarian Research Grants OTKA T-048961 and T-068626. SB. acknowledges support from JPL 12700064. Smithsonian Institution Restricted EndOWment funds and NASA NNX07A1356. KGS appreciates the continuous support of the Vienna-MP in southern Arizona through the State of Brandenburg MWFK. References Aarum-Ulvas. V, Henry. G.W. 2003. 402, 1033 Baliunas. S.L..Donahue. R.A.. Soon, W. et authors) 1995, 438. 269 Baliunas. S.L., Sokoio?'. D.. Soon. W. 1996a. 457, L99 Baliunas. S.L., Nesme?Ribes, E.. Sokolo?', D., Soon, W. 1996b, 460, 848 Baliunas, S.L.. Friek. P.. Moss. D. Popova. E. Sokoloff, D, Soon, W. 2006. MNRAS 365. 181 Berdyugina, S.V., Henry, G. 2007. 659, L157 Berdyugina, S. V., Pelt. 1., 'Iluominen, I. 2002, 394, 505 Bushby. PJ..Tobias. SM. 2007. 661. 1289 Cameron. R., Schtissler, M. 2007, Ap] accepted Dette. L. 1971. Inf. Bull. Var. Stars, No. 520 (Editor's note) Donahue. RA.. Saar. S.H.. Baliunas, S.L. 1996. 466. 384 Dumey, BR.. Sten?o, .0. 1972, 15, 307 Frick. P.. Baliunas, S.L.. Galyagin. D., Sokolofl?, D, Soon, W. 1997. 483, 426 Frick, P.. Soon. W.. Popova, Baliunas. S. 2004, New Astronomy 9.599 H.-E., Knoll. P.. Strassmeier, K.G.: 2006. 454. 295 Geyer. E. 1978, Asuonomische Mitteilungen. no. 43, p. 187 Hamnann, L.. Dussault, M., Noah. P. V., Klimke, A., Bopp. B. W. 1981, Apl 249, 662 Henry. Hamer, L, 1995, 97. 513 Innis, 1.. Budding, E., Olah, K., Jarvinen. S. P.,Coates, D. W.. Messina, S., Kaye, T. G. 2008. IBVS No. 5832 Jlirvinen. S. P.. Berdyugina, S. V., Tuominen. I.,Cutispoto. G., Bos. M. 2005. 432. 657 Kjurkchieva. D.P. Marchev. D.V. 2005, 434. 221 Kollath, Z., Ol?h. K. 2008, submitted (Paper I) K?vari, Zs.. Strassmeier, K.G.. Gmnzer, T., Weber. M., 01h, K., Rice, 1.3. 2004, 417, 1047- 1054 Kron, GE. 1947, PASP. 59. 26] Lanza, A.F.. Piluso. N.. Rodono. M., Messina, S., Cutispoto, G. 2006, 455. 595 01611 et 20.: Changing cycles results Lockwood, G.W., Skiff, BA.. Henry. G.W., Henry. S., Radick. RR., Baliunas. S.L., Donahue, R.A., Soon, W. 2007, 171, 260 Messina, S., Guinan . BE 2002. 393, 255 Messina. S., Guinan, EF. 2003, 409, 1017 Moss, D.. Saar. S.S.. Sokolo?', D. 2008, MNRAS 388, 416 01611, K. 2007, in: Highlights of Astronomy, IAU XXVL GA. 14?25 August 2006, Prague, ed. Karel A. van der Hucht. Vol. 14. p. 283 Ol?h, K.. Korhonen. H., K?v?ri. Zs ., Forg?cs?Dajka. E., Strassmcier, K. G. 2006, 452,303 Ol?h, K., K?v?ri, Zs.,Bartus, J..et a1. 1997. 321, 811 01511 K., Koll?th Z., Shassmeier K.G.: 2000, 356, 643 Ol?h K., Strassmeier K.G.: 2002. in: K.G. Snassmeier Proc. Is: Potsdam Hinlu?hap an Sunspors and Stars-pols, AN 323, 3/4, 361 Ol?h, K., Strassmeier, K.G., K?v?ri, Zs., Guinan, ER 2001, 372, 119 Ol?h, K.. Strassmeier, K.G., Granzer, T. 2002, AN 323, 5, 453 Ol?h, K., Strassmeier. K.G.. Granzer, T., Soon, W., Baliunas, S.L.2007, AN 328, 1072 Radick, R.R., Lockwood, G.W.,Skilf, BA.,Baliunas, S.L. 118, 239 Rib?rik, G., Ol?h, K. Snassmeicr, K.G. 2003. AN 324, 202 Saar, S.H., Brandenburg. A. 1999. 524, 295 Schwabe, H. 1843, AN 20, 284 Stelzer B, Fem?ndez, M, Costa, VM. et 2003, 411, 517 Su'assmeier, K.G. 2002, AN 323 309 Strassmeier, K. G., Boyd, L. ., Epand. D. 11., Granzer, Th. 1997a, PASP 109, 697 Strassmcier, K. G., Bartus, J., Cutispoto, G., Rodono, M. 1997b, 125. 1] Valenti. J.A., Fischer, D.A. 2005, 159, 141 Vaughan. A.H., Pleston, G.W., Wilson, 0.C. 1978, PASP 90, 267 Vida, K. 2007. AN 328, 817 Vogt, S.S.. Hangs. A.P., Misch, A. 1999, 121, 547 Washuettl, K?v?ri, 25., Foing. B.H., 01511, K., Vida. K. et :11. (+29 authors) 2008, submitted Wilson, 0.C., 1968, 153, 221 Wilson, 0.C. 1978, 226, 379 11 ATTACHMENT I SOLAR CLIMATE VARIATION ON MU TO CENTENNIAL TIMESCALES: EMPIRICAL EVIDENCE, MECHANISTIC EXPLANATION, AND TESTABLE CONSEQUENCES Willie W.-H. Soon Harvard-Smithsonian Center for Astrophysics Cambridge, Massachusetts 021 38 Abstract: Soon (2005) showed that the variable total solar irradiance (TSI) could explain, rather surprisingly, well over 75% of the variance for the decadally smoothed Arctic-wide surface air temperature over the past 130 years. The present paper provides additional empirical evidence for this physical connection, both through several newly published high-resolution paleo?proxy records and through robust climate-process mod- eling outputs. This paper proposes a mechanistic explanation, involving: (1) the variable strength of the Atlantic meridional overturning circulation (MOC) or thermohaline circu? lation (2) the shift and modulation of the Inter-Tropical Convergence Zone (ITCZ) rainbelt and tropical Atlantic ocean conditions; and (3) the intensity of the wind?driven subtropical and subpolar gyre circulation, across both the North Atlantic and North Pacific. A unique test of this proposed solar TSl-Arctic thermal?salinity?cryospheric cou- pling mechanism is the 5- to 20-year delay effect on the peak Atlantic MOC ?ow rate centered near and on sea surface temperature (SST) for the tropical Atlantic. The solar Arctic?mediated climate mechanism on multidecadal to centennial timescales pre- sented here can be compared with and differentiated from both the related solar and UV irradiance forcing on decadal timescales. The ultimate goal of this research is to gain sufficient mechanistic details so that the proposed solar-Arctic climate connection on multidecadal to centennial timescales can be confirmed or falsified. A further incentive is to expand this physical connection to longer, millennial-scale variability as motivated by the multiscale climate interactions shown by Braun et al. (2005), Weng (2005), and Dima and Lohmann (2009). [Key words: solar?Arctic climate connection, total solar irradiance, Atlantic meridional overturning circulation, climate variability.) THE SOLAR CLIMATE VARIATION ON MULTIDECADAL TO CENTENNIAL TIMESCALES Paleoclimatic proxies show ubiquitous, multidecadal to centennial-scale vari- abilities that may ultimately be associated with the persistent forcing by solar irradi- ance variability as properly projected and amplified through the annual progression of the Earth around the Sun (Table A1, Appendix). The present study indirectly assumes the optimal climatic response filter of the Earth ocean?atmosphere-ice sys- tem to peak around such multidecadal to centennial scales, which can be taken to be roughly 50 to 500 years much less than 1000 years). The challenge of this research, then, must lie in the identification of relevant and/or dominant centers of climatic action Table 1 lists acronyms used in this paper) and interactions among those COAs (Christoforou and Hameed,, 1997; Rodionov et al., 2005; Huth et al., 2006; Lim et al., 2006). Huth et al. (2006) found a general tendency for atmo- spheric circulation modes1 to be more zonal", with COAs covering wider areas and 144 Physical Geography, 2009, 30, 2, pp. 144?184. Copyright 2009 by Bellwether Publishing, Ltd. All rights reserved. DOI: 10.274710272-3646.30.2.144 CLIMATE VARIATION 145 Table 1. List of Acronyms Used in This Paper Acronym Definition total solar irradiance UV ultraviolet BP Before Present COAS centers of (Climatic) action SST sea surface temperature SLP sea level air pressure EPG equator-to?pole surface temperature gradient AMO Atlantic Multidecadal Oscillation NPMO North Pacific Multidecadal Oscillation PDO Pacific Decadal Oscillation NAO North Atlantic Oscillation ENSO El Nine?Southern Oscillation MOC meridional overturning circulation THC thermohaline circulation ISOW Iceland?Scotland Overflow Water GIN Seas Greenland-lcelandic-NonNegian Seas ITCZ lntertropical Convergence Zone SPCZ South Pacific Convergence Zone GISPZ Greenland Ice Sheet Project 2 MIS marine isotope stage GCM general circulation model Coupled Model lntercompariSOn Project Phase 3 NCAR National Center for Atmospheric Research United Nations intergovernmental Panel on Climate Change teleconnection among different regions spanning longer distances when solar activ? ity is strong. The hard task of separating the dynamics of the teleconnection from the actual physical mechanisms at COAs must be kept in mind as well. In this paper, climate refers to the systematic persistence of weather patterns and fluctuations that involve: (1) seasonal and annual cycles notjust time-averaged weather statistics); (2) local and regional air pressure systems; (3) topography, land- scape, and the storage and exchange of heat/energy through atmospheric and oce? anic circulation; and (4) delayed actions. All these persistent local and regional actions and variations take place prior to any global mean radiative forcing or any cohesive global mean temperature and precipitation responses. In other words, the weather?mediated climate variation and change will be viewed as local and regional "inter-seasonal? variations that cover time intervals from months and years to tens of millennia. The basic mechanisms involved are not unlike the original orbital theory of climate change by Milutin Milankovitch, published in the early 19405, which emphasized high-latitude, light-sensitive COAs to explain global- scale glaciation and deglaciation events and transitions. A key emphasis of this insolation?weather?climate framework are the differential responses at different lat~ itudes to insolation changes (Davis and Brewer, 2009) in addition to responses aris- ing from effects of the four seasons. Thus, it is suggested that persistent insolation forcing, when maintained over multidecadal to centennial timescales, accounting for both the systematics of the Sun?Earth orbital geometry (Loutre et al., 1992) and 146 WILLIE W-H. SOON the irradiance variability intrinsic to magnetic variation of the Sun Soon, 2007), is both necessary and sufficient to explain the observed climatic variation on multidecadal to centennial timescales. It can be further added that an all-inclusive theory of climate change should also account for the newly proposed theory of independent hemispheric responses to solar forcing by Huybers and Denton (2008), whereby a Northern Hemispheric response is sensitive to both the local summer insolation intensity and the latitudi? nal insolation and temperature gradients (Davis and Brewer, 2009), while a South- ern Hemispheric response is more sensitive to local summer duration. In the framework of climatic forcings and responses, an understanding of both the spectral peaks and the seemingly gap-less continuum of weather?climate oper- ation will be sought. Huybers and Curry (2006) recently re?initiated such research by seeking to connect the annual and Milankovitch cycles to the in-between con- tinuum temperature variability in terms of the response to deterministic insolation forcing. Coincidentally, the multidecadal to centennial timescales discussed in the present study are similar to the recognized transitional timescale of Huybers and Curry (2006), who proposed that the annual cycle, with assistance from the ocean- storage delays, served to extend the continuum temperature variability from months to decades, while the Milankovitch orbital forcing cycles, with assistance from non? linear ice?sheet dynamics, drove the continuum temperature variability to higher- frequency timescales of millennia. This View of local and regional origins of wide spatial climatic co?variations and responses is consistent with the emphasis on relatively high net solar radiation reaching the surface at various locations in the Pacific Ocean Stanhill and Cohen, 2008), or at other warm-pool regions Pavlakis et al., 2008), as a driv- ing force for the fast-coupled air-sea responses that are coherent over broad spatial extent (Meehl et al., 2008; van Loon and Meehl, 2008). Finally, an important prac? tical concept of the ?modulated annual cycle," which accounts for the intrinsic nonlinearity of the weather?climate forcings and feedbacks, has been recently developed by Wu et al. (2008). Both d?Orgeville and Peltier (2007) and Zhang and Delworth (2007) showed the intimate multiscale coupled interactions and connections among dominant times? cales and patterns of climate variability involving the Pacific Decadal Oscillation (PDO), the related North Pacific Multidecadal Oscillation (N PMO), and the Atlantic Multidecadal Oscillation (AMO). Although d?Orgeville and Peltier (2007) did not commit to any particular mode as the leading variable, Zhang and Delworth (2007) suggested a lagged North Pacific response to the AMO forcing of about 13 years that is connected through a chain of dynamical atmospheric teleconnections (induced first from AMO-related northward oceanic heat transport) and then ampli? fied by the positive local air?sea feedback over the central and western North Pacific. Zhang and Delworth (2007) further deduced that a regime shift of the North Pacific opposite to the 1976?1977 shift might be expected soon, following the switch of the AMO to a positive phase around 1995. Three inter-related causes support a strong control of natural multidecadal?to- centennial scales of climate variation through a solar?Arctic connection mecha- nlsm: CLIMATE VARIATION I47 Cause A: A persistent and systematic variation of the solar T51 and related insolation gradient modulates the atmospheric heat transport from the tropics to the Arctic, and hence modulates the Arctic tempera? ture change itself with little or no delay. Cause B: Thermal perturbations lead to both natural modulation of the Arctic sea ice and transport of fresh water through the Bering Strait, and from the Arctic through both the Greenland Sea and Denmark Strait and the Canadian Arctic Archipelago pathways to deep water formation sites spread across the North Atlantic from the Greenland?icelandic? Norwegian (GIN) Seas to the east and at the Labrador Sea in the west. Cause C: Further effects are: (1) thermal, freshwater, and salinity pertur? bation of the Atlantic (2) the delayed connection of about 5 to 20 years with the tropical Atlantic SST and the lnterTropical Conver- gence Zone and (3) coupling of the affected tropical Atlantic processes feeding back to the It is important to note that current climate models are not yet able to account for all the empirical and proxy evidence and relations noted here Zhang and Delworth, 2007; Alexander, 2009). Kravtsov and Spannagle (2008), for example, make use of the fact that the AMO signals contained in the difference (suggesting that climate models have failed to account for the AMO) between the observed SST and the multimodel ensemble-mean SST from the CMIP3 (Coupled Model Inter- comparison Project Phase 3) database, suggesting that AMO is a natural climatic signal plausibly related to the oceanic thermohaline circulation (THC). Davis and Brewer (2009) pointed out that climate models may overemphasize the seasonal response to insolation changes when compared to the differential latitudinal response which, in turn, translates into an incorrect representation of the latitudinal temperature gradient that is fundamental for capturing climate dynamics (Lindzen, 1994; Jain et al., 1999). Such a zero-order climate modeling barrier has been recently reframed by Rind (2008, p. 855) as "the consequences of not knowing low- and high-latitude climate sensitivity.? Finally, two further assumptions regarding the multidecadal? to centennial-scale solar~Arctic connection mechanism. First, the solareArctic mechanism borrows from recent studies by van Loon and Meehl (2008) and Meehl et al. (2008), which focus on coupled surface responses in the Pacific region to the Sun?s decadal peaks, giving rise to their hypothesized solar-induced, hydrology-amplified climatic responses. Multidecadal to centennial responses could represent the envelope of the responses to the solar decadal peaks. However, the mechanism proposed here specifies responses in the Arctic and Atlantic basins and postulates equivalent and related responses elsewhere. Second, the solar?Arctic mechanism assumes the importance of a significant coupled thermal-5alinity?cryospheric interaction involving the Arctic and many other climatic COAs around the world. Behl and Kennett (1996) discussed the connections of anoxic events in the Santa Barbara basin with the Dansgaard?Oeschger warm interstadials recorded in the GISP2 core 148 WILLIE W-H. SOON for the past 60 kyr, and Wang et al. (2008) discussed the strong coupling between the East Asian Summer Monsoon system and the North Atlantic, especially during the cold glacial interval between 75 to 10 BP. However, different regions may not be so optimally teleconnected at a much warmer interglacial time when there is little Arctic or Northern Hemisphere volume of ice, such as during the MIS stage 5d (Zhou eta?L, 2008).2 This paperioffers support for the proposed solar?Arctic mechanism for climate variations on multidecadal to centennial timescales. The mechanism can also be compared and contrasted with two other promising Sun?climate connection sce- narios via the decadal solar UV and mechanisms reviewed below. THE DECADAL SOLAR UV AND TSI MECHANISMS Kodera (2004) showed a dynamical response of the Indian summer July4 August) monsOon that perhaps can be traced to the downward?propagating effects of wave?mean flow interactions through the forcing by relatively stronger solar dec? adal UV radiation in the mesosphere and stratosphere. Kodera and Shibata (2006), using a unique and powerful new diagnostic technique, showed how enhanced solar heating in the tropical lower stratosphere, while suppressing convective activ- ity in the equatorial region, enhances convective activity in off?equatorial regions and ultimately produces a change in the meridional circulation in the tropical tro? posphere. Theianalyses by Claud et al. (2008) show added spatial complexity to this dynamical coupling of the Indian summer monsoon system to the 11-year solar activity cycle.?Such a proposed top-down response to solar decadal UV forcing on the Indian summer monsoon offers a remote teleconnection to both the climatic variations and trends in the Pacific equatorial cold tongue region and the North Atlantic (Selten et al., 2004; Compo and Sardeshmukh, 2008) through what is known as the circumglobal wave teleconnection mechanism Branstator, 2002; Watanabe, 2004; Ding and Wang, 2005). Van Loon and Meehl (2008) and Meehl et al. (2008), through the powerful com? bination of data analyses and climate modeling experiments, showed how the fast and closely coupled surface responses to changing solar surface radiation between solar activity maxima and minima could add to the top?down responses through the solar UV mechanism proposed by Kodera and colleagues. Specifically, van Loon and Meehl (2008) showed that the coupled surface responses in the Pacific to TSI variation is such that in solar peak forcing years, the sea level air pressure (SLP) is above normal in the Gulf of Alaska and south of the equator, which in turn pro? duces stronger southeast trade winds across the equatorial Pacific and causes increased upwelling and hence cooling SST tendencies broadly across the Pacific Ocean. Using two GCMs at NCAR, Meehl et al. (2008) sketched a coupled response to peaks of solar decadal TSI forcing that involves increased latent heat flux and evaporation, which in turn is carried to the Pacific Intertropical conver- gence zone (ITCZ) and South Pacific convergence zone (SPCZ) to intensify both precipitation regimes. The resulting solar response patterns resemble La i?a-like events, but yet are distinct from them, mainly by virtue of their different vertical pro- file of responses, especially in the stratosphere (van Loon and Meehl, 2008). ARCTIC-MEDIATED CLIMATE VARIATION I49 Modeling experiments conducted by Emile?Geay et al. (2007) suggest that the pow- erful ENSO coupled air-sea interaction system can serve as a mediator3 of the per? sistent solar TSI forcing on climate over the Holocene. Their sensitivity calculations, with the intrinsic solar TSI variation ranging from 0.05% to 0.2% to 0.5% over the Holocene, coupled with the orbital forcing effect, generated El Nino?like SST anomalies at times of decreased TSI, which is consistent with the empirical results by van Loon and Meehl (2008). 'In summary, Kodera and colleagues proposed a top?down forcing-response sce- nario for a Sun?climate decadal connection that more directly involves the decadal solar UV forcing, while van Loon, Meehl and colleagues sketched a bottom-up forcing-response scenario for their Sun?climate connection on decadal timescale, which invokes decadal solar TSI forcing. The strength of the bottom-up scenario of coupled surface air-sea responses to a persistent solar TSI forcing by van Loon, Meehl, and colleagues is that it offers a bet? ter explanation for why the multidecadal- to centennial-scale variability can be found in such a diverse range of climate-proxy archives from the bottom of the sea to high mountain tops (Table A1, Appendix). In contrast, it is harder to conceive of a spatially coherent and temporally persistent near-surface response over long dis? tances, wide geographical conditions, and different topographic settings if the ini- tial meteorological and climatic impact centers are rooted in the tropical mesosphere and stratosphere as a response to the decadal solar UV forcing. Although the decadal signal for solar irradiance forcing of global-averaged sur- face temperature has recently been confirmed by Tung and Camp (2008), the actual scenario for a physical connection has not been identified. Lim et al. (2006) found that solar irradiance modulation of local and regional relative humidity, in combi? nation with the related climatic distribution of clouds and water vapor over the tropical Atlantic, is sufficient to explain the observed tr0pica Atlantic decadal oscillation. RESU LTS AND DISCUSSION Empirical Evidence and Mechanistic Explanation for lnterrelated Causes and Responses A, B, and Soon (2005) showed evidence of natural climate variations on multidecadal to centennial timescales through a solar?Arctic connection mechanism. Figure A1 (Appendix) updates the previously published solar TSI?Arctic surface air tempera- ture correlation in Soon (2005).4 The results presented by Kauker et al. (2008) strongly support the multidecadal variations of the Arctic surface temperature from the Arctic Atlantic, to the Arctic Pacific and then to the Arctic Greenland/Iceland sectors (the chart is available upon request). Figure 1 shows that the solar Arctic-wide temperature correlation can also be found on a much smaller regional scale, as demonstrated by similar TSI?temperature correlations for two coastal sta- tions of southern Greenland: Godthab Nuuk in the west and Ammassalik to the east. It is important to note that available oceanographic data at Fyla Bank off Godthab Nuuk show that the early 20th century surface thermometer warming was 150 WILLIE SOON 1369 2 Solar irradiance 1368 1367 - 1366 Flu 1365 .. 1364 Solar total irradiance (Wlmz) 1363' 5.5 .4 Annual mean Godthab Nuuk surface temperature 1362 NASA GISS Godthab Nuuk surface air temperature (2008} 1880 1 900 1920 1940 1960 1980 2000 1369 I . . 2 Solar irradiance 1368 7 1367 1366 1365 '2 1364 Solar total irradiance (Wlmz) 1363 4 Annual mean Ammassalik surface temperature 1362 NASA Ammassalik surface air temperature (2008) 1880 1900 1920 1940 1960 1980 2000 Year Fig. 1. The annual mean estimates of total solar irradiance (solid line) compared with the surface temperature records from two coastal Greenland stations: Godthab Nuuk (detted curve; top panel) and Ammassalik (clotted curve; bottom panel) from about 1881 to 2007 (from "after homogeneity adjust? ment? records in This result adds regional details to the TSl-Arctic-wide surface temperature correlation identified in Soon (2005). detected around 1920 at the surface ocean (Holland et al., 2008). Such a consistent pattern of correlations on different spatial domains and scales is an important ingre- dient for a physical solar?Arctic connection. Another important update and extension is the new result by Jiang et al. (2005) showing the consistent role for solar irradiance forcing in triggering and ARCTIC-MEDIATED CLIMATE VARIATION 151 maintaining the multidecadal to centennial variation of the SSTs around the North Icelandic Shelf over the last 2000 years. Jiang et al. (2005) further noted the rela? tively stronger temperature responses during winters than summers, which is con- sistent with the result outlined in Soon (2005). High-resolution proxy annual-mean and wintertime SSTs from a coral record at Bermuda (Goodkin et al., 2008a, 2008b) showed enhanced multidecadal?scale variability during the late 20th century when compared to variability near the end of the Little Ice Age. These results, together with the evidence on the sensitivity of Arctic Ocean ice cover and thickness to atmospheric poleward energy flux by Soderkvist and Bjork (2004) in their coupled ocean?ice?atmosphere column model, support the proposed solar?Arctic connec- tion Cause A. The enhanced poleward atmospheric transport scenario is supported by the consistent increases in wind stress trends over the Arctic basin shown for annual mean, winter, and summer values for 1948-2006 (Hakkinen et al., 2008). Indirect evidence for variable poleward heat transport for earlier periods before 1950) can be found in the multidecadal variation of the equator?to-pole (EPG) surface temperature gradient as well as the multidecadal-scale modulation of the phase of the EPG annual cycles (Jain et al., 1999). Graversen et al. (2008), in their close examination of the vertical pattern of recent Arctic warming, concluded that much of the observed Arctic warming aloft is related to changes in poleward atmospheric heat and moisture transports rather than from near?surface snow and ice albedo feedbacks, as has been modeled and suggested in climate model experiments with increased atmospheric C02. This result is consistent with the theoretical and modeling studies by Alexeev et al. (2005), Winton (2006), and Cai and Lu (2007), where poleward heat transports, plausibly linked to differential latitudinal response to insulation changes, are shown and argued to be more important in explaining polar warming than direct surface snow and ice albedo feedbacks. Smedsrud et al. (2008) showed that indeed the poleward atmospheric energy flux to the Arctic has increased overall for the last 50 years, from 1956 to 2006, which is consistent with solaruArctic connection Cause A, but they emphasized that the tendency for a net increase over more recent decades has slowed. L?Heureux et al. (2008), Overland et al. (2008), Serreze et al. (2008), Zhang et al. (2008), and Lindsay et al. (2009) all provided updated data series up to 2007 and discussion of the key role played by the recent shift in spatial patterns of atmospheric forcing and the strengthened poleward atmospheric heat transport directly or indirectly reaching the central Arctic. Polyakov et al. (2005) showed evidence for the enhanced North Atlantic warm water intrusion into the Arctic Ocean and Barents Sea, while Shimada et al. (2006) documented the influx of warm Pacific summer waters into the Arctic Ocean via the Bering Strait in order to account for the observed rapid changes in the Arctic climate system. Serreze et al. (2008) argued that the near surface?based Arctic amplification signal through snow and ice albedo feedbacks may soon be emerging if the Arctic Sea continues to lose its ice, and emphasized that their results are not in conflict with those of Graversen et al. (2008). Finally, direct hydrographic data from the northeast North Atlantic and Nordic Seas in Holliday et al. (2008) showed not only the reversal of the 1960 to 19905 freshening trend but also seem to offer practical short-term forecasts for temperature 52 WILLIE SOON and salinity around the Fram Strait region for Atlantic inflow conditions to the Arctic Ocean. A similar forecast based on short-term hydrographic tendencies for Labrador Sea regions has also been proffered by Yashayaev (2007) and Yashayaev et al. (2007), but Yashayaev and Loder (2009) recently reported a sudden atmospheric cooling and enhanced production of Labrador Sea water in the fall?winter 2007? 2008 season, which disrupted the steady warming around the region since 1994. Similarly, Vage et al. (2009) documented a surprising return of winter deep convec- tion to the subpolar gyre in both the Labrador and lrminger Seas, apparently with? out going through a phase of preconditioning. This most up-to-date situation in the Labrador Sea should points to the need for caution when attempting to forecast any near? or long-term changes in the northern North Atlantic and Arctic.5 Empirical evidence supporting the solar?Arctic connection Cause may be found in the important of observational data in Polyakov et al. (2008), demonstrating the multidecadal variability of climate variables in the Arctic and their interconnections, which include the Arctic surface air temperature, upper 150-m Arctic Ocean freshwater content, fast ice thickness, intermediate Atlantic water core temperature of the Arctic Ocean, and upper 300 North Atlantic water salinity. Here, one might interpret that a warmer Arctic (detected in both air and ocean-water temperatures) led to above-normal melting of Arctic sea ice and excess flushing of Arctic freshwater to the Nordic seas and the subpolar North Atlantic basins. The observational data of Polyakov et al. (2008) are consistent especially with the newly reconstructed freshwater content data series over the northern Atlantic by Pardaens et al. (2008). Dima and Lohmann (2007) independently sketched a dynamically consistent framework for the AMO, and were able to fill in some important feedbacks and delay factors. They show the hemispheric wavenumber?l sea level air pressure pat- tern to be related to the Fram Strait sea ice export, which, in turn, affects the MOC oceanic circulation and hence the sea surface conditions in both the North Atlantic and North Pacific. Dima and Lohmann (2007) spelled out the role of the THC adjustment to freshwater forcing, the Atlantic SST response to MOC, and the oceanic adjustment in the North Pacific as key delays in the chain, while the ocean- atmosphere-sea ice interactions in the Atlantic, Pacific, and Arctic oceans served as the crucial negative feedbacks to sustain the AMO oscillation on timescales of about 70 years. Jungclaus et al. (2005) also proposed a scenario of Arctic?North Atlantic interactions with the multidecadal variability of Atlantic based on the outputs of their BOO-year GCM control, unforced, run. The important exten? sion of climate modeling experiments by Grosfeld et al. (2008) shows that, in addi- tion to attributing the origin of the 60?70 year scale oscillation to the Atlantic Ocean, there is possibly a separate and distinct scale of about 80?100 years that is intrinsic to the Pacific Ocean. The paper?s argument for a multidecadal- to centennial-scale variability adOpts and accepts most of the detailed physical processes outlined in Dima and Lohmann (2007), but the solar?Arctic connection picture given here also includes a more direct emphasis on climatic modulation by the Arctic the call for direct involvement of Arctic?wide surface temperature and sea ice and fresh water in the Arctic basin, with emphasis on pathways for freshwater exchanges and transports, CLIMATE VARIATION I 53 rather than merely sea?ic'e export from?the?Frarn Strait); and a Wider range of spatial? temporal scales beyond the more limited 70- to 80-year variability set in the Dima and Lohmann (2007) framework, because the memory and turning points for the multiscale oscillation in this solar?Arctic connection picture appear to be decided more by the external TSI forcing. For this reason, the phrase "multidecadal to cen? tennial timescales?6 is used throughout this paper. Adding to these processes is the current emphasis on the effects of influx of low- salinity Pacific water through the Bering Strait (Aagaard et al., 2006; Shimada et al., 2006; Keigwin and Cook, 2007) in perturbing the ice and freshwater environment over the Arctic Ocean. These effects are non-negligible and may at times have played a more prominent role than at present Wadley and Bigg, 2002; Yang, 2005, 2006; Dickson et al., 2007). Finally, Peterson et al. (2006) and Serreze et al. (2006) confirmed the roles of net precipitation, river discharge, and sea ice attrition as important freshwater sources, compared to the relatively minor contributions of glacial melt. The modeling study by Wu and Wood (2008) suggests that the recent freshening trend over the subpolar North Atlantic can be explained by a redistribur tion of freshwater within the Arctic and subpolar North Atlantic and that the redis- tribution was probably carried out by a perturbed ocean circulation in the subpolar seas and triggered by deep convection in the Labrador Sea. Both the Arctic sea ice extent data derived by Zakharov (in Johannessen et al., 2004) and the Icelandic sea ice extent (Zhang and Vallis, 2006) provide evidence in support of the inverse relation between Arctic temperature and sea ice extent for about the past 100 years, as proposed in Cause of this solar?Arctic connection. Quantitative reconstruction by Kauker et al. (2008) also support a significant ice loss over the Arctic basin from 1916 to 1955, although they suggested that the ice loss of this period was somewhat less extensive than the recent loss from mid?19605 to mid-19905. In the context of the proposed solar Arctic?mediated climate variation mecha? nism, it is assumed that the sea ice export is not exactly the same as freshwater export from the Arctic and that more sea ice export from the Arctic basin may be more related to colder conditions within the Arctic, not unlike the notable ice- rafting events and episodes seen throughout the Quaternary Bond et al., 1997, 1999; Vidal et al., 1997; Darby and Zimmerman, 2008; Hill et al., 2008; or counter?views and interpretations by Andrews et al., 2006), or in the summer of 1695 recently re-interpreted by Gil et al. (2006), as well as the great salinity anom- aly events of the 19705, 19805, and 19905, modeled and discussed in Zhang and Val Iis (2006). But importantly, it is recognized that extra exports of sea ice, episodic or otherwise, or a more continuous nature of the freshening and flushing of water from the Arctic basin to the northern North Atlantic basins Condron et al. 2009) would serve as key negative feedbacks for the oscillation. Dickson et al. (2007) highlighted the potentially greater importance of combined ice and freshwater outflows from the Arctic Ocean basin through the Canadian Arc- tic Archipelago/Nares Strait/Baffin Bay/Davis Strait pathways under a warm Arctic and Iow-ice?volume climatic regime.7 Detailed computer modeling (Proshutinsky et al., 2002; Moon and Johnson, 2005; Dukhovskoy et al., 2006; Condron et al., 2009) shows how the oscillations between the anticyclonic and cyclonic 54 WILLIE SOON Linear oonelation coef?cient r= -0.45 (n 109/3; biennial sampling between 1770-2004) I 1367 5 I 15.6 A 55? 1366 1364 ,5 to g; 16.0 1363 .3 I: I . d. Three point - smoothed grain size - Iceland - Scotland- 1362 . servos water. {sesame at all . . 16.2 1800 1850 1900 1950 2000 Yea Fig. 2. The annual mean estimates of total solar irradiance (solid line) correlated with the three-point smoothed mean grain-size index (dotted line) of Boessenkool et al. (2007) from about 1770 to 2004. The grain-size index is a proxy for the flow speed of the near-bottom lceland?Scotland Overflow Water (ISOW) which is, in turn, related to the deep water formation in the Labrador Sea to the west. Smaller mean grain size suggests slower ISOW, and larger grain size implies faster ISOW. circulation regimes, involving contraction and expansion of the Beaufort Gyre, are affecting how Arctic sea ice and freshwater are stored and released to the northern North Atlantic Ocean. Figure 2 recOrds plausible evidence for a connection between solar forcing in producing the effects on deep?ocean flow of the northern North Atlantic for the full 1770?2004 A.D. interval. It uses new mean grain-size data from Boessenkool et al. (2007) that represents the near-bottom flow speed of Iceland?Scotland Overflow Water (ISOW). It should be further noted that Bossenkool et al. (2007) suggest that the vigor of ISOW is controlled by the trans? port and characteristics of the Labrador Sea water farther to the west (Jungclaus et al., 2005). The correlation between and the three-point smoothed grain-size index shown in Figure 2 has a correlation coefficient of -0.45, which even with the reduced degrees of statistical freedom would still constitute a significant corre- lation. In comparison, Boessenkool et al. (2007) showed a correlation between the grain size with a seven?year smoothed NAO index for the selected (rather than the full data shown in Fig. 2) interval of 1885?2004 with an value of only -0.-42 (n 55). Although it is not the intent of this paper to explain the correlation, but merely to demonstrate plausibility, the apparent correlation in Figure 2 suggests a slower ISOW flow speed with increasing TSI. Finally, the solar?Arctic connection in Cause ARCTIC-MEDIATED CLIMATE VARIATION 55 . "r "7 -- Atlantic THC eight future projections (Knight et al.. 2005) Solar irradiance shifted forward by 10 years Atlantic THC anomaly (Sv) Detrend solar total irradiance anomaly (\me2) I 1900 1950 2000 Year Fig. 3. The detrended total solar irradiance anomaly Series shifted forward by 10 years (thick solid line; see also the same shift in Fig. 4) to show correlation with the maximum of the zonal mean of the Atlantic Meridional Overturning Circulation at deduced by Knight et al. (2005) (dotted grey lines with the upper and lower bounds as the "uncertainty" limits). Grey diamond symbols connected with thin solid lines are the eight-member forecasts for the 35 years offered by Knight et al. (2005). A detrended solar series was used in order to compare more fairly with the normalized measures of SST and THC anomalies used in Knight et al. (2005). See Kravtsov and Spannagle (2008) for a discussion of the details of the detrending of datasets for the construction of AMO-related SST changes, and Vellinga and Wu (2004) for a discussion of why the maximum MOC index is a useful proxy for the Atlantic THC for the study of AMO, but the index is clearly not useful for assessing interannual THC variability. outlined here may also find support from the specific documentation of the 75? to 80?year period from the Holocene history of drift ice within the northern North Atlantic region by Moros et al. (2006). In the search of a physical mechanism and understanding of a Sun?climate con? nection, one need not be automatically hunting for maximum possible statistical correlations between any two variables Soon et al., 2000). For example, Zhang et al. (2007) showed how an equally good fit of the observed detrended Northern Hemispheric temperature time series can be achieved with relatively high correlations, and yet each of the good fits was obtained under dramatically different heat flux redistribution and transport scenarios. Such a reality suggests that high correlations between variables do not imply correct identification of a physical mechanism given that multiple physical processes could well be responsible for establishing a quasi-mean state or any deviation from the mean. Figures 3 and 4 show the empirical support for the proposed solar?Arctic connec? tion Cause C. In Figure 3, the maximum MOC index centered around I56 WILLIE W-H. SOON I I I :3 3 Tropical Atlantic SST response? 3 1363 If Contemporaneous irradience 0 5 5 t0 - 2 1367 1366 1364 -o.5 '3 1363 IE1880 1900 1920 1940 1960 1980 2000 Year 1369 Solar irradiance shifted 10 years forward 1368 to Illustrate the delayed correlation A ?f'g 1367 a - 1366 .E (D a I 5 . 1365 1364 1: =5 8 ID 3 2 5-9: 1363 1362 1830 1900 1920 1940 1960 1980 2000 Year Fig. 4. The annual-mean estimates of total solar irradiance (TSI) versus the tropical Atlantic SST at from 1870 to 2004 (top panel), and with the solar TSI advanced fonNard by 10 years (bottom panel) in order to illustrate the delayed connection of the tropical Atlantic SST to solar TSI forcing effects initiated first within the Arctic and North Atlantic basins. roughly 1000 to 2000 below surface) as deduced from the SST distribution by Knight et al. (2005) is plotted with the TSI series shifted forward by 10 years, corre- Sponding to the estimated delayed response in lower latitudes. Figure 4 shows a sim- ilar comparison with tropical Atlantic SST at around The chosen delay time of 10 years is only a rough estimate for the thermal?cryospheric-salinity and mechanical wind stress effects occurring within the Arctic and northern North ARCTIC-MEDIATED CLIMATE VARIATION 57 Atlantic basins to propagate southward. But it is clear from both empirical evidence (Curry et al., 1998; Molinari et al., 1998) and careful ocean modeling (Yang, 1999) that a physical delay of some 5 to 20 years is reasonable. Yang (1999), for example, pointed out a five-year delay for decadal variations in the Labrador Sea and the trop? ical Atlantic dipole index set by coastally trapped waves, rather than the probably longer advection time through the Deep Western Boundary Current (Goodman, 2001). in the AMO framework of Dima and Lohmann (2007), a delay of 10?1 5 years was deduced for the time it will take for the freshwater forcing on both the North East Atlantic deep water and Labrador Sea deep water convection sites to affect the MOC circulation. Jungclaus et al. (2005) deduced an optimal lead time of 12 years for changes in the convection intensity in the Labrador Sea to affect the Finally, Latif et al. (2006) offered evidence and argument for the atmospheric NAO index to lead the Atlantic Dipole Index8 by 5 to 20 years, where this index is pro? posed as a good proxy for circulation. Based on inland temperature proxy data, the finding by Eichler et al. (2009) of 10- to 30?year delays between the solar forcing proxy and Siberian Altai Mountain region temperature throughout the 1250?2000 AD period is consistent with the proposed solar?Arctic and Atlantic MOC-mediated mechanism. Additional Mechanistic Explanation for Interrelated Causes and Responses A, B, and The wind-driven subtropical and subpolar gyre circulations both across the Pacific and Atlantic Oceans may be also important for the plausible solar?induced feedbacks and delays to help sustain the multidecadal to centennial variations Wu et al., 2003; Zhang and Vallis, 2006; Hasegawa et al., 2007,- Qiu et al., 2007; DiLorenzo et al., 2008; Guan and Huang, 2008; Alexander, 2009; Saenko, 2009). The modeling study by Wu et al. (2003), for example, shows that in the North Pacific, the multidecadal memory may be rooted in the slow adjustment of the subtropical/subpolar gyre in response to wind stress imposed in the central North Pacific and the slow growth/decay of the SST anomalies that propagate eastward in the Kuroshio Extension region. Saenko (2009) showed the important climatic impacts of wind stress, especially those around regions poleward of 30?, with oce? anic heat transport accounting for only a small fraction of total poleward energy transport, and where, if one were to remove that wind stress forcing, surface tem- peratures at high-latitude regions could drop by more than with the mean position of the simulated sea ice edge moving equatorward and reaching latitude The important study by (Juan and Huang (2008), which emphasizes mechan- ical energy in order to sustain the THC, shows how adding the wind-driven gyre not only leads to a more complete modeling of the physical processes related to THC, but also changes the threshold value of the THC dynamical bifurcation property greatly. Therefore, both fresh water and wind forcing will be key elements for the current solar? Arctic connection picture. It may not be straightforward to explain the seemingly counterintuitive relation- ship of stronger Atlantic maximum MOC with increased TSI forcing indicated in Figure 3. But the plausibility of a decreased equator-to-pole surface density 158 WILLIE W-H. SOON from an enhanced thermal and fresh water perturbation and modulation of the con- vective sinking regions for deep water formation spread across the North Atlantic with increased TSI forcing) leading to stronger, rather than weaker, thermohaline circulation was studied by Nilsson and Walin (2001) and Nilsson et al. (2003). The theory of ilsson, Walin and colleagues viewed the slow upwelling of dense water overall in the low latitudes and the Southern Ocean (see also Visbeck, 2007; Toggweiler and Russell, 2008), rather than high?latitude production and sinking of dense water as the rate-limiting branch of THC.9 The Nilsson et al. theory showed, with a reasonable model of interval wave mixing, that the vertical diffusivity would increase with decreasing surface equator-to-pole density contrast, and that would deepen the thermocline and, in turn, lead to a stronger THC. The proposed Cause mechanism must necessarily include coupling with the multidecadal? to centennial-scale variations of the Atlantic Intertropical Conver- gence Zone, as noted in several proxy archives (Nyberg et al., 2001; Poore et al., 2004; Peterson and Haug, 2006; Black et al., 2007) and in climate modeling exper- iments (Vellinga and Wu, 2004; Chiang and Bitz, 2005; Zhang and Delworth, 2005). in general, these studies have highlighted a robust shift of the south- ward during North Atlantic cooling and slower and a northward shift during the opposite phase of stronger and warmer North Atlantic- Arctic conditions. The most important aspect of these studies that focused on the tropical Atlantic is the related feedbacks to the itself. Vellinga and Wu (2004) placed a greater emphasis on the role of low?latitude fresh water through the variability on a centennial timescale for feeding back to the THC circulation. This emphasis may also be ultimately related to the THC variability theory of Nilsson and Walin (2001) and Nilsson et al. (2003). There is little doubt that both the and inter-hemispheric SST gradient proposed by Vellinga and Wu (2004) are dominant weatheraclimate processes operating on seasonal and interannual timescales that can feed back and couple to the solar TSI?induced Arctic?high lati- tude processes emphasized in this paper. But it is harder to find justification in avail- able data that ?sustained salinity anomalies slowly propagate toward the subpolar North Atlantic at a lag of 5?6 decades? to maintain the centennial?scale variability of Atlantic THC, as seen in model outputs by Vellinga and Wu (2004, p. 4498). More Related Consequences and Impacts Several important and related consequences and connections of multidecadal? to centennial-scale variations of the Atlantic have recently been pointed out by Dong et al. (2006), Goswami et al. (2006), Knight et al. (2006), Lu et al. (2006), Li and Bates (2007), Sutton and Hodson (2007), Timmermann et al. (2007), Chang et al. (2008), Denton and Broecker (2008), Feng and Hu (2008), Li et al. (2008), Ting et al. (2008), and Wang et al. (2009). In Figure 11 of their study of the effects of a weakening Atlantic on the coupled ENSO system, Timmermann et al. (2007) made the remarkable obser- vation that, during the positive phase of AMO, the annual cycles of the Nino-3 SST are intensified, while the ENSO?scale Ge, 2 to 8 years) SST variability is relatively more muted, and the inverse occurs for the opposite phase of AMO. Such a CLIMATE VARIATION I 59 nonlinear multidecadal modulation of the annual?cycle and ENSO signals, which was clearly noted by White and Liu (2008b), may ultimately be consistent with the new insight they offered concerning the non-linear alignment of El Ni?o/La Ni?a episodes with the combined signals from the 11-year solar cycle?generated 3rd (3 .6- year) and 5th (22-year) harmonics. On millennial timescales, proxy data (Stott et al., 2002) from the western Pacific warm pool region suggest that El Ni?o conditions correlate with cooler-stadial conditions around Greenland and the North Atlantic, while La Ni?a conditions tend to correlate with warmer interstadials. Benton and Broecker (2008) demonstrated the non?obvious connection between AMO and the retreating and advancing activity of 38 selected glaciers in the Swiss Alps with little or only slight delays in the glacier response. Such a tight coupling between glacier activity in the Swiss Alps and AMO was suggested to arise from the effects of AMO on European summertime temperatures. Chang et al. (2008) showed the active role played by in explaining abrupt climate events in the tropical Atlantic, including the rapid reduction of summer monsoonal wind and rainfall over West Africa. Knight et al. (2006) and Ting et al. (2008) showed the wide-ranging climatic impacts of the AMO, including rainfall over the Sahel and sea surface temperature over the main development region of Atlantic hurricanes (Fig. 4). Goswami et al. (2006), Lu et al. (2006), Sutton and Hodson (2007), Feng and Hu (2008), and Li et al. (2008) found multidecadal modulation of Indian summer mon- soon rainfall through empirical data analyses and modeling experiments. Although the AMO?Indian monsoon rainfall relationship is not fully robust, the general ten- dency is such that a positive phase, via both persistent tropo- spheric and near-surface response pathways, leads to more summer rainfall with modulated delay responses until the months of September and October. Li and Bates (2007) showed atmospheric GCM results that yielded relatively uniform, warmer winters in East China but a dipolar north?south positive?negative pattern of precipitation responses during the positive AMO phase and inversely for the nega- tive AMO phase. Earlier, Tan et al. (2004) showed an interesting correlation between the warm-season temperature proxy for Beijing and the North Atlantic Drift ice index of Bond et al. (2001) covering the last 2650 years, but they did not offered a working mechanism. Wang et al. (2009) emphasized the influences of AMO on Asian monsoonal climate in all four seasons, producing weakened winter monsoons but enhanced summer monsoons related to AMO?modulated tropo- spheric heating anomalies. Finally, the works by Braun et al. (2005), Weng (2005), and Dima and Lohmann (2009) support the present proposal by showing how the various key intrinsic times? cales and physics related to this solar-Arctic connection can interact and connect dynamically from annual cycles to the noted millennial?scale oscillation of about 1470 years of the Dansgaard?Oeschger events noted during glacial intervals.10 Weng (2005), using both ocean temperature data and a toy model, illustrated how in a nonlinear weather?climate regime that even a "small" change in TSI forcing will effectively interact with and couple to the seasonal forcing to generate and sus- tain climate responses and variations of multidecadal to centennial timescales.11 Braun et al. (2005), using the Potsdam Institute?s intermediate complexity coupled 160 WILLIE SOON climate system model, showed how the i470?year glacial climate cycle could be robustly and realistically generated solely from the periodic forcing of freshwater input into the North Atlantic Ocean in cycles of 87 and 210 years, which were identified by the authors as the solar Gleissberg and deVries/Suess activity cycles, respectively. ?2 Dima and Lohmann (2009) suggest that, instead of being the of the two basic solar cycles or any amplification of a weak direct 1500-year forcing of unknown origin by THC, the origin of the 1500-year cycle is best viewed as the rectification of an external solar forcing through dynamical connection to a thresh- old internal response of the THC. Their work emphasizes that observed millennial variability in paleo-proxy records should be considered as a derived dynamical mode of the climate system without physical processes on a fixed millennia times- cale, regardless of whether this timescale is rooted in the Sun or in Earth climate system. This possibility certainly adds another layer of complexity in the study of the Sun?climate connection. CONCLUSION This paper proposes three interrelated causes for natural climate variations on multidecadal to centennial timescales through a solar?Arctic connection mecha- nism. The first, Cause A, is that a persistent and systematic variation of the solar TSI and related insolation gradient modulates the atmospheric heat transport from the tropics t0 the Arctic, and hence modulates the Arctic temperature change itself with little or no delays. The second, Cause B, is that thermal perturbations lead to both natural modula- tion of the Arctic sea ice and to transport of fresh water through the Bering Straits and from the Arctic through both the Greenland Sea and Denmark Strait and the Canadian Arctic Archipelago pathways to deep water formation sites spread across the North Atlantic from the Greenland~lcelandic?Norwegian (GIN) Seas to the east and at the Labrador Sea to the west. The third, Cause C, is that further effects are: (1) thermal, freshwater, and salinity perturbation of the Atlantic (2) the delayed connection of about 5 to 20 years with the tropical Atlantic SST and the lnterTropical Convergence Zone and (3) coupling of the affected tropical Atlantic processes feeding back to the MOC-THC. This three?part solar?Arctic cli? mate variation mechanism emphasizes plausible physical arguments rather than statistical correlations. The proposed solar?Arctic connection chains from Causes have good empirical support, and this mechanism appears to explain the operation of coupled air?ocean?ice responses over broad areas connecting the Arctic and North Atlantic to other locations on multidecadal to centennial timescales. This proposal offers the opportunity for a rejectable scientific hypothesis of a physical Sun?climate connec? tion. The new should be viewed as a step forward in the long quest to understand how the full weather?climate continuum varies on multidecadal to cen- tennial timescales by highlighting the role of solar irradiance forcing upon the Arctic region, in not only sustaining and amplifying the natural climatic oscillation CLIMATE VARIATION I 61 and variation, but also in the selectivity or specification of the broadband nature of the spatial and temporal scales of the climatic responses involved. Acknowledgments: I thank two referees for their constructive comments and edits that improved the paper. I also thank all colleagues whose works are cited here, and especially those who have allowed access to their hard-earned data series: Nicola Scafetta, Karin Boessenkool, Igor Yashayaev, Igor Polyakov, Mihai Dima, Lars Smedsrud, Jeff Knight, Rob Allan, Daniel Hodson, David Holland, Mads Ribergaard, Frank Kauker, and John Fasullo. I thank Scott Gene Avrett, Sallie Baliunas, Dan Botkin, Bob Carter, Shaun Cheok, Susan Crockford, Bob Ferguson, Dave Fettig, Kesten Green, Joe Kunc, Keith Lockitch, Christopher Monckton, Lubos Motl, Jane Orient, Eric Posmentier, Art Robinson, Mitch Taylor, Bin Wang, and the late Robert Jastrow for their encouragement, and Gene Avrett and Steve Cranmer for their editorial help. I further thank Than, Lien and Julia Pham, Chiew-See Chua, as well as Benjamin and Franklin Soon for motivation. This new was based on a presentation at the 33rd International Geological Congress held in Oslo, Norway, August 6?14, 2008, as well as by another presentation at the International Symposium on Climate and Weather of the Sun?Earth System held at Jakarta, Indonesia, November 24?26, 2008. The views expressed herein are solely those of the author and are independent of sources providing support. REFERENCES Aagaard, K., Weingartner, T. J., Danielson, S. L., Woodgate, R. A., Johnson, G. C., and Whitledge, T. E. (2006) Some controls on flow and salinity in Bering Strait. Geophysical Research Letters, Vol. 33, Adkins, J. F., Ingersoll, A. P., and Pasquero, C. (2005) Rapid climate change and conditional instability of the glacial deep ocean from the thermobaric effect and geothermal heating. Quaternary Science Reviews, Vol. 24, 581-594. Alexander, M. A. (2009) Extratropical air?sea interaction, SST variability and the Pacific Decadal Oscillation (PDO). AGU Geophysical Monograph Series (Cli- mate Dynamics: Why Does Climate Vary?), in press. Alexeev, V. A., Langen, P. L., and Bates, J. R. (2005) Polar amplification of surface warming on an aquaplanet in "ghost forcing? experiments without sea ice feed? backs. Climate Dynamics, Vol. 24, 655?666. Alley, R. B. (2007) Wally was right: Predictive ability of the North Atlantic "Con? veyor Belt? hypothesis for abrupt climate change. Annual Review of Earth and Planetary Sciences, Vol. 35, 241?272. Andrews, J. T., Jennings, A. E., Moros, M., Hillaire?Marcel, C., and Eberl, D. (2006) Is there a pervasive Holocene ice?rafted debris (IRD) signal in the northern North Atlantic? The answer appears to be either no, or it depends on the proxy! PAGES News, Vol. 14, No. 2, 7?9. Behl, R. J. and Kennett, J. P. (1996) Brief interstadial events in the Santa Barbara basin NE Pacific, during the past 60 kyr. Nature, Vol. 379, 243?26. Black, D. E., Abahazi, M. A., Thunell, R. C., Kaplan, A., Tappa, E. J., and Peterson, L. C. (2007) A 8-century tropical Atlantic SST record from the Cariaco Basin: Baseline variability, twentieth-century warming, and Atlantic hurricane fre? quency. Paleoceanography, Vol. 22, doizl Boessenkool, K. R, Hall, I. R., Elderfield, H., and Yashayaev, I. (2007) North Atlantic climate and deep-ocean flow speed changes during the last 230 years. Geophys- ical Research Letters, Vol. 34, doiz?IOJ 162 WILLIE W-H. SOON Bond, G. and 9 co-authors (1997) A pervasive millennial?scale cycle in North Atlantic Holocene and glacial climate. Science, Vol. 278, 1257?1266. Bond, G. and 9 co?authors (1999) The North Atlantic?s 1?2 climate Relation to Heinrich events, Dansgaard/Oeschger Cycles and the Little Ice Age. AG Geophysical Monograph Series Mechanisms of Global Climate Change at Millennial Time Scales), Vol. 112, 35?58. Bond, G. and 9 co?authors (2001) Persistent solar influence on North Atlantic cli- mate during the Holocene. Science, Vol. 294, 2130?2135. Branstator, G. (2002) Circumglobal teleconnections, theJet Stream waveguide, and the North Atlantic Oscillation. journal of Climate, Vol. 15, 1893?1910. Braun, H.,.Christl, M., Rahmstorf, 5., Ganopolski, A.-, Mangini, A., Kubatzki, C., Roth, K., and Kromer, B. (2005) Possible solar origin of the 1,470-year glacial climate cycle demonstrated in a coupled model. Nature, Vol. 438, 208?21 1. Braun, H., Ditlevsen, P., and Chialvo, D. R. (2008) Solar forced Dansgaard? Oeschger events and their phase relation with solar proxies. Geophysical Research Letters, Vol. 35, Cai, M. and Lu, 1. (2007) Dynamical greenhouse?plus feedback and polar warming amplification. Part II: Meridional and vertical asymmetries of the global warm- ing. Climate Dynamics, Vol. 29, 375?391. Chang, P. and 8 co?authors (2008) Oceanic link between abrupt changes in the North Atlantic Ocean and the African monsoon. Nature GeoScience, Vol. 1, 444?448. Chiang, J. C. H. and Bitz, C. M. (2005) Influence of high latitude ice cover on the marine lntertropical Convergence Zone. Climate Dynamics, Vol. 25, 477?496. Christoforou, and Hameed, S. (1997) Solar cycle and the Pacific ?centers of action.? Geophysical Research Letters, Vol. 24, 293?296. Claud, C., Duchiron, B., and Terray, P. (2008) On the associations between 11-year solar cycle and the Indian summer monsoon system. journal of Geophysical Research, Vol. 113, Compo, G. P. and Sardeshmukh, P. D. (2008) Removing ENSO?related variations from the climate record. journal of Climate, submitted. Condron, A., Windsor, P., Hill, C., and Menemenlis, D. (2009) Simulated response of the Arctic freshwater budge to extreme NAO wind forcing. journal of Climate, in press, Cuny, J., Rhines, P. B., and Kwok, R. (2005) Davis Strait volume, freshwater and heat fluxes. Deep Sea Research I, Vol. 52, 519?542. Curry, R. G., McCartney, 5., and Joyce, T. M. (1998) Oceanic transport of subpolar climate signals to mid?depth subtropical waters. Nature, Vol. 391, 575?577. Darby, D. A. and Zimmerman, P. (2008) lce-rafted detritus events in the Arctic the last glacial interval, and the timing of the lnnuitian and Laurentide ice sheet calv- ing events. Polar Research, Vol. 27, 114?127. Davis, B. A. S., and Brewer, S. (2009) Orbital forcing and role of the latitudinal insolation/temperature gradient. Climate Dynamics, Vol. 32, 143?165. Denton, G. H. and Broecker, W. S. (2008) Wobbly ocean conveyor circulation dur- ing the Holocene? Quaternary Science Reviews, Vol. 27, 1939?1950. ARCTIC-MEDIATED CLIMATE VARIATION 63 Dickson, R., Rudels, B., Dye, 5., Karcher, M., Meincke, J., and Yashayaev, (2007) of- seas. Progress in Physical Oceanography, Vol. 73, 210?230. Di Lorenzo, E., Schneider, N., Cobb, K. M., Franks, P. J. 5., Chhak, K., Miller, A. J., McWilliams, J. C., Bograd, S. J., Arango, H., Curchitser, E., Powell, T. M., and Riviere, P. (2008) North Pacific Gyre Oscillation links ocean climate and ecosys- tem change. Geophysical Research Letters, Vol. 35, doi:1 0.1 02 9/ 2007GLO32838. Dima, M. and Lohmann, G. (2007) A hemispheric mechanism for the Atlantic mul? tidecadal oscillation. journal of Climate, Vol. 20, 2706?2719. Dima, M. and Lohmann, G. (2009) Conceptual model for millennial climate vari? ability: A possible combined solar?thermohaline circulation origin for the ~1,500-year cycle. Climate Dynamics, Vol. 32, 301?311. Ding, Q. and Wang, B. (2005) Circumglobal teleconnection in the Northern Hemi- sphere summer. journal of Climate, Vol. 18, 3483?3505. Dong, B., Sutton, R. T., and Scaife, A. A. (2006) Multidecadal modulation of El Nine?Southern Oscillation (ENSO) variance by Atlantic Ocean sea surface tem- peratures. Geophysical Research Letters, Vol. 33, d?Orgeville, M. and Peltier, W. R. (2007) On the Pacific decadal oscillation and the Atlantic multidecadal oscillation: Might they be related? Geophysical Research Letters, Vol. 34, Dukhovskoy, D., Johnson, M., and Proshutinsky, A. (2006) Arctic decadal variability from an idealized atmosphere-ice-ocean model: 2. Simulation of decadal oscil- lations. journal of Geophysical Research, Vol. 111, Eichler, A., Olivier, 5., Henderson, S. K., Laube, A., Beer, J., Papina, T., Gaggler, H. W., and Schwikowski, M. (2009) Temperature response in the Altai region lags solarforcing. Geophysical Research Letters, Vol. 36, Emile-Geay, J., Cane, M., Seager, R., Kaplan, A., and Almasi, P. (2007) El Ni?o as a mediator of the solar influence on climate. Paleoceanography, Vol. 22, Feng, S. and Hu, Q. (2008) How the North Atlantic Multidecadal Oscillation may have influenced the Indian summer monsoon during the past two millennia. Geophysical Research Letters, Vol. 35, Giampapa, M. 5., Hall, J. C., Radick, R. R., and Baliunas, S. L. (2006) A survey of chromospheric activity in the solar-type stars in the open cluster M67. The Astro- physical journal, Vol. 651 444?461. Gil, l. M., Abrantes, F., and Hibbeln, D. (2006) The North Atlantic Oscillation forc? ing through the last 2000 years: Spatial variability as revealed by high?resolution marine diatom records from and SW Europe. Marine Micropaleontology, Vol. 60,113?129. Goodkin, N. F., Hughen, K. A., Curry, W. B., Doney, S. C., and Ostermann, D. R. (2008a) Sea surface temperature and salinity variability at Bermuda during the end of the Little Ice Age. Paleoceanography, Vol. 23, Goodkin, N. F., Hughen, K. A., Doney, S. C., and Curry, W. B. (2008b) Increased multidecadal variability of the North Atlantic Oscillation since 1781. Nature Geoscience, Vol. 1, 844?848. 164 WILLIE SOON Goodman, P. J. (2001) Thermohaline adjustment and advection in an OGCM. jour? nal of Physical Oceanography, Vol. 31 1477?1497. Goswami, B. N., Madhusoodanan, M. S., Neema, C. P., and Sengupta, D. (2006) A physical mechanism for North Atlantic SST influence on the Indian summer monsoon. Geophysical Research Letters, Vol. 33, Graversen, R. G., Mauritsen, T., M., Kallen, E., and Svensson, G. (2008) Vertical structure of recent Arctic warming. Nature, Vol. 541, 53?56. Greene, C. H., Pershing, A. J., Cronin, T. M., and Ceci, N. (2008) Arctic climate change and its impacts on the ecology of the North Atlantic. Ecology, Vol. 89 Supplement, $24?$38. Grosfeld, K., Lohmann, G., and Rimbu, N. (2008) The impact of Atlantic and Pacific Ocean sea surface temperature anomalies on the North Atlantic multidecadal variability. Tel/us, Vol. 60A, 7284741. Guan, Y. P. and Huang, R. X. (2008) Stommel?s box model of thermohaline circula- tion revisited?the role of mechanical energy supporting mixing and the wind? driven gyration. journal of Physical Oceanography, Vol. 38, 909?917. Hakkinen, S., Proshutinsky, A., and Ashik, l. (2008) Sea ice drift in the Arctic since the 19505. Geophysical Research Letters, Vol. 35, Hall, J. C., Henry, G. W., and Lockwood, G. W. (2007) The sun?like activity of the solar twin 18 SCORPII. Astronomical journal, Vol. 133, 22064208. Hall, J. C. and Lockwood, G. W. (2004) The chromospheric activity and variability of cycling and flat activity solar-analog stars. The Astrophysical journal, Vol. 614, 942-946. . Hall, J. C., Lockwood, G. W., and Skiff, B. A. (2007) The activity and variability of the Sun and sun?like stars. I. Synoptic Ca ll and observations. Astronomical journal, Vol. 133, 862?881. Hasegawa, T., Yasuda, T., and .Hanawa, K. (2007) Multidecadal variability of the upper ocean heat content anomaly field in the North Pacific and its relationship to the Aleutian Low and the Kuroshio Transport. Papers in Meteorology and Geo- physics, Vol. 58, 155?166. Hill, J. C., Gayes, P. T., Driscoll, N. W., Johnstone, E. A., and Sedberry, G. R. (2008) Iceberg scours along the southern US. Atlantic margin. Geology, Vol. 36, 447? 450. Holland, D. M., Thomas, R. H., de Young, 3., Ribergaard, M. H., and Lyberth, B. (2008) Acceleration of Jakobshavn lsbrae triggered by warm subsurface ocean waters. Nature Geoscience, Vol. 1, 659?664. Holliday, N. P., Hughes, S. L., Bacon, 5., A., Hansen, 3., Lavin, A., Loeng, H., Mork, K. A., Osterhus, S., Sherwin, T., and Walczowski, W. (2008) Reversal of the 1 9605 to 19905 freshening trend in the northeast North Atlantic and Nordic Seas. Geophysical Research Letters, Vol. 35, Holzkamper, S., Mangini, A., Spot], C., and Mudelsee, M. (2004) Timing and pro- gression of the Last lnterglacial derived from a high alpine stalagmite. Geophys- ical Research Letters, Vol. 31, Hoyt, D. V. and Schatten, K.H. (1993) A discussion of plausible solar irradiance variations, 1700?1992. journal of Geophysical Research, Vol. 98, 18,895? 18,906. ARCTIC-MEDIATED CLIMATE VARIATION 165 Huth, R., Pokorna, L., Bochnicek, J., and Hejda, P. (2006) Solar cycle effects on modes of low-frequency circulation variability. journal of Geophysical Research, Vol. 111, Huybers, P. and Curry, W. (2006) Links between annual, Milankovitch, and contin- uum temperature variability. Nature, Vol. 441 329?332. Huybers, P. and Denton, G. (2008) Antarctic temperature at orbital timescales con? trolled by local summer duration. Nature Geoscience, Vol. 1, 787?792. IPCC (2007) Chapter2 by Forster, P., Ramaswamy, V., Artaxo, P., Berntsen, T., Betts, R., Fahey, D. W., Haywood, J., Lean, J., Lowe, D. C., Myhre, G., Nganga, J., Prinn, R., Raga, (3., Schulz, M., and Van Dorland, R., Changes in atmospheric constituents and in radiative forcing. In S. Solomon, D. Qin, M. Manning, Z. Chen, M. Marquis, K. B. Avery, M. Tignor, and H. L. Miller, eds., Climate Change 2007: The Physical Science Basis. Cambridge, UK and New York, NY: Cambridge University Press, Contribution of Working Group 1 to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Jain, 5., Lall, U., and Mann, M. E. (1999) Seasonality and interannual variations of Northern Hemisphere temperature: Equator?to-pole gradient and ocean?land contrast. journal of Climate, Vol. 12, 1086?1100. Jiang, H., Eiriksson, J., Schulz, M., Knudsen, K.-L., and Seidenkrantz, (2005) Evidence for solar forcing of sea?surface temperature on the North lcelandic Shelf duing the late Holocene. Geology, Vol. 33, 73?76. Johannessen, O. M., L., Miles, M. W., Kuzmina, S. L, Semenov, V. A., Alekseev, G. V., Nagurnyi, A. P., Zakharov, V. F., Bobylev, L. P., Pettersson, L. H., Hasselmann, K., and Cattle, H. P. (2004) Arctic climate change: Observed and modeled temperature and sea?ice variability. Tellus, Vol. 56A, 328?341 . Jones, E. P., Swift, J. H., Anderson, L. G., Lipizer, M., Civitarese, G., Falkner, K. K., Kattner, G., and McLaughlin, F. (2003) Tracing Pacific water in the North Atlantic Ocean. journal of Geophysical Research, Vol. 108, 141. Jungclaus, J. H., Haak, H., Latif, M., and Mikolajewicz, U. (2005) Arctic?North Atlantic interactions and multidecadal variability of the meridional overturning circulation. journal of Climate, Vol. 18, 4013?4031. Kauker, F., Koberle, C., Gerdes, R., and Karcher, M. (2008) Modeling the 20th cen- tury Arctic ocean/sea ice system: Reconstruction of surface forcing. journal of Geophysical Research, Vol. 1 13, Keigwin, L. D. and Cook, M. S. (2007) A role for North Pacific salinity in stabilizing North Atlantic climate. Paleoceanography, Vol. 22, Knight, J. R., Allan, R. J., Folland, C. K., Velinga, M., and Mann, M. E. (2005) A signature of persistent natural thermohaline circulation cycles in observed cli? mate. Geophysical Research Letters, Vol. 32, Knight, J. R., Folland, C. K., and Scaife, A. A. (2006) Climatic impacts of the Atlantic multidecadal oscillation. Geophysical Research Letters, Vol. 33, doi:10.1029/ 2006GL026242. Kodera, K. (2004) Solar influence on the Indian Ocean monsoon through dynamical processes. Geophysical Research Letters, Vol. 31, 166 WILLIE W-H. SOON Kodera, K. and Shibata, K. (2006) Solar influence on the tropical stratosphere and troposphere in the northern summer. Geophysical Research Letters, Vol. 33, Kravtsov, S. and Spannagle, C. (2008) Multidecadal climate variability in observed and modeled surface temperatures. journal of Climate, Vol. 21, 1104?1 121. Kwok, R. (2006) Exchange of sea ice between the Arctic Ocean and the Canadian Arctic Archipelago. Geophysical Research Letters, Vol. 33, doi:10.1029/ 2006GL027094. Kwok, R. (2007) Baffin Bay ice drift and export: 2002?2007. Geophysical Research Letters, Vol. 34, Latif, M., Boning, C., Willebrand, J., Biastoch, A., Dengg, J., Keenlyside, N., Sch?w?ecke'ndiek, U., and Madec, G. (2006') is the thermohaline circulation changing? journal of Climate, Vol. 19, 4631?4637. Latif, M., Roeckner, E., Esch, M., Haak, H., 'Hagemann, S., Jungclaus, J., Legutke, 5., Marsland, 5., Mikolajewicz, U., and Mitchell, J. (2004) Reconstructing, monitor- ing, and predicting multidecadal?scale changes in the North Atlantic thermoha? line circulation with sea surface temperature. journal of Climate, Vol. 17, 1605? 1614. Lean, J. (2000) Evolution of the Sun?s spectral irradiance since the Maunder Mini- mum. Geophysical Research Letters, Vol. 27, 2425?2428. Lean, J., Beer, J., and Bradley, R. (1995) Reconstruction of solar irradiance since 1610: Implications for climate change. Geophysical Research Letters, Vol. 22, 3195?3198. L?Heureux, M. L., Kumar, A., Bell, G. D., Halpert, M. 5., and Higgins, R. W. (2008) Role of the Pacific?North American (PNA) pattern in the 2007 Arctic sea ice decline. Geophysical Research Letters, Vol. 35, Li, S. and Bates, G. T. (2007) Influence of Atlantic multidecadal oscillation on the winter climate of East China. Advances in Atmospheric Sciences, Vol. 24, 126? 135. Li, S., Perlwitz, J., Quan, X., and Hoerling, M. P. (2008) Modeling the influence of North Atlantic multidecadal warmth on the Indian summer rainfall. Geophysical Research Letters, Vol. 35, Lim, G.-H., Suh, Y.-C. and Kim, B.-M. (2006) On the origin of the topical Atlantic decadal oscillation based on the analysis of the ICOADS. Quarterly journal of the Royal Meteorological Society Vol. 132, 1139?1 152. Lindsay, R. W., Zhang, J., Schweiger, A., Steele, M., and Stern, H. (2009) Arctic sea ice retreat in 2007 follows thinning trend. journal of Climate, Vol. 22, 165?176. Lindzen, R. S. (1994) Climate dynamics and global change. Annual Review of Fluid Mechanics, Vol. 26, 353?378. Lockwood, G. W., Skiff, B. A., Henry, G. W., Henry, S., Radick, R. R., Baliunas, S. L., Donahue, R. A., and Soon, W. (2007) Patterns of photometric and chromospheric variation among sun?like stars: A 20- year perspective. The Astro- physical journal Supplementary Series, Vol. 171, 260?303. Loutre, M. F., Berger, A., Bretagnon, P., and Blanc, (1992) Astronomical fre- quencies for climate research at the decadal to century time scale. Climate Dynamics, Vol. 7, 181?194. ARCTIC-MEDIATED CLIMATE VARIATION 67 Lu, R., Dong, B., and Ding, H. (2006) Impact of Atlantic multidecadal oscillation on the Asian summer monson. Geophysical Research Letters, Vol. 33, doi:10.1029/ 20066L027655. Meehl, (3. A., Arblaster, J. M., Branstator, (3., and van Loon, H. (2008) A coupled air?sea response mechanism to solar forcing in the Pacific region. journal of Cli- mate, Vol. 21, 2883?2897. Molinari, R. L., Fine, R. A., Wilson, W. 0., Curry, R- (3., Abell, J., and McCartney, 5. (1998) The arrival of recently formed Labrador Sea Water in the Deep Western Boundary Current at Geophysical Research Letters, Vol. 25, 2249-2252. Moon, S. and Johnson, M. A. (2005) The relationship between isotherm and atmospheric forcing in the Arctic Ocean. Geophysical Research Letters, Vol. 32, Moros, M., Andrews, J. T., Eberl, D., and Jansen, E. (2006) Holocene history of drift ice in the northern North Atlantic: Evidence for different spatial and temporal modes. Paleoceanography Vol. 22, Munchow, A., Malling, H., and Falkner, K. K. (2006) An observational estimate of volume and freshwater flux leaing the Arctic Ocean through Nares Strait. journal of Physical Oceanography, Vol. 36, 2025?2041. Nilsson, J., Brostrom, (3., and Walin, G. (2003) The thermohaline circulation and vertical mixing: Does weaker density stratification give stronger overturning? journal of Physical Oceanography, Vol. 33, 2781?2795. Nilsson, J. and Walin, (3. (2001) Freshwater forcing as a booster of thermohaline circulation. Tel/us, Vol. 53A, 629?641 . Nyberg, J., Kuijpers, A., Malmgren, B. A., and Kunzendorf, H. (2001) Late Holocene changes in precipitation and hydrography recorded in marine sediments from northeastern Caribbean Sea. Quaternary Research, Vol. 56, 87?102. Overland, J. E., Wang, M., and Salo, S- (2008) The recent Arctic warm period. Tel/us, Vol. 60A, 587?597. Pardaens, A., Vellinga, M., Wu, P., and lngleby, B. (2008) Large?scale Atlantic salin? ity changes over the last half-century: A model?observation comparison. journal of Climate, Vol. 21, 1698?1720. Park, W. and Latif, M. (2008) Multidecadal and multicentennial variability of the meridional overturning circulation. Geophysical Research Letters, Vol. 35, doi:1 0.1 02 L035 779. Pavlakis, K. (3., Hatzianastassiou, N., Matsoukas, C., Fotiadi, A., and Vardavas, l. (2008) ENSO surface shortwave radiation forcing over the tropical Pacific. Atmo- spheric Chemistry and Physics, Vol. 8, 5565?5577. Peterson, B. J., McClelland, J., Curry, R., Holmes, R. M., Walsh, J. E., and Aagaard, K. (2006) Trajectory shifts in the Arctic and subArctic freshwater cycle. Science, Vol. 313, 1061?1066. Peterson, L. C. and Haug, (3. H. (2006) Variability in the mean latitude of the Atlantic Intertropical Convergence Zone as recorded by riverine input of sediments to the Cariaco Basin (Venezuela). Palaeogeography Palaeoclimatol- ogy, Palaeoecology, Vol. 234, 97?113. Polyakov, I. V., Alexeev, V. A., Belchansky, G. Dmitrenko, l. A., Ivanov, V. V., Kirillov, S. A., Korablev, A. A., Steele, M., Timokhov, L. A., and Yashayaev, I. 168 WILLIE SOON (2008) Arctic ocean freshwater changes over the past 100 years and their causes. journal of Climate, Vol. 21, 364?384. Polyakov, l. V., A., Carmack, E. C., Dmitrenko, I. A., Fahrbach, E., Frolov, I. E., Gerdes, R., Hansen, E., Holfort, J., Ivanov, V. V., Johnson, M. A., Karcher, M., Kauker, F., Morison, J., Orvik, K. A., Schauer, U., Simmons, H. L., Skagseth, O., Sokolov, V. T., Steele, M., Timokhov, L. A., Walsh, 0., and Walsh, J. E. (2005) One more step toward a warmer Arctic. Geophysical Research Let- ters, Vol. 32, Poore, R. 2., Quinn, T. M., and Verardo, S. (2004) Century-scale movement of the Atlantic lntertropical Convergence Zone linked to solar variabilty. Geophysical Research Letters, Vol. 31, Portis, D. H., Walsh, J. E., Hamly, M. E., and Lamb, P. J. (2001) Seasonality of the North Atlantic Oscillation. journal of Climate, Vol. 14, 2069?2078. Prinsenberg, S. J. and Hamilton, J. (2005) Monitoring the volume, freshwater and heat fluxes passing through Lancaster Sound in the Canadian Arctic Archipelago. AtmOSphere-Ocean, Vol. 43, 1?22. Proshutinsky, A., Bourke, R. H., and McLaughlin, F. A. (2002) The role of the Beaufort Gyre in Arctic climate variability: Seasonal t'o decadal climate scales. Geophysical Research Letters, Vol. 29, Qiu, B., Schneider, N., and Chen, S. (2007) Coupled decadal variability in the North Paci?c: An observationally constrained idealized model. journal of Cli- mate, Vol. 20, 3602?3620. Radick, R. R., Lockwood, G. W., Skiff, B. A., and Baliunas, S. L. (1998) Patterns of variation among Sun-like stars. The Astrophysical journal Supplementary Series, Vol. 118, 239-258. Rind, D. (2008) The consequences of not knowing low? and high?latitude climate sensitivity. Bulletin of the American Meteorological Society, Vol. 89, 855?864. Rodionov, S. N., Overland, J. E., and Bond, N. A. (2005) The Aleutian Low and winter climatic conditions in the Bering Sea. Part I: Classification. journal of Cli- mate, Vol. 18, 160?177. Saenko, O. A. (2009) On the climatic impact of wind stress. journal of Physical Oceanography, Vol. 39, 89?106. Scafetta, N. and Willson R. (2009) ACRlM-gap and trend issue resolved using a surface magnetic flux proxy model. Geophysical Research Letters, in press, Selten, F. M., Branstator, G. W., H. A., and Kliphuis, M. (2004) Tropical origins for recent and future Northern Hemisphere climate change. Geophysical Research Letters, Vol. Serreze, M. C., Barrett, A. P., Slater, A. G., Woodgate, R. A., Aagaard, K., Lammers, R. B., Steele, M., Moritz, R., Meredith, M., and Lee, C. M. (2006) The large-scale freshwater cycle of the Arctic. journal of Geophysical Research Letters, Vol. 1 1 1, doi:10.1 029/2005] C003424. Serreze, M. C., Barrett, A. P., Stroeve, J. C., Kindig, D. N., and Holland, M. M. (2008) The emergence of surface?based Arctic amplification. The Cryosphere Discussion, Vol. 2, 601?622. ARCTICHMEDIATED CLIMATE VARIATION 169 Shimada, K., Kamoshida, T., ltoh, M., Nishino, S., Carmack, E., McLaughlin, F., Zimmermann, 5., and Proshuntinsky, A. (2006) Pacific Ocean inflow: Influence on catastrophic reduction of sea ice cover in the Arctic Ocean. Geophysical Research Letters, Vol. 33, Smedsrud, L. H., Sorteberg, A., and Kloster, K. (2008) Recent and future changes of the Arctic sea?ice cover. Geophysical Research Letters, Vol. 35, doi:10.1029/ 20080L034813. Soderkvist, J. and Bjork, G. (2004) Ice thickness variability in the Arctic Ocean between 1954?1990, results from a coupled ocean-ice-atmosphere column model. Climate Dynamics, Vol. 22, 57?68. Soon, W. (2005) Variable solar irradiance as a plausible agent for multidec- adal variations in the Arctic?wide surface air temperature record of the past 130 years. Geophysical Research Letters, Vol. 32, Soon, W. (2007) Implications of the secondary role of carbon dioxide and methane forcing in climate change: Past, present, and future. Physical Geography, Vol. 28, 97?125. Soon, W., Baliunas, 5., Posmentier, E. S., and Okeke, P. (2000) Variations of solar coronal hole area and terrestrial lower tropospheric air temperature from 1979 to mid-1998: Astronomical forcings of change in Earth?s climate? New Astron- omy, Vol. 4, 563?579. Stanhill, G. and Cohen, S. (2008) Solar radiation changes in Japan during the 20'h century: Evidence from sunshine duration measurements. journal of the Meteo? rological Society of japan, Vol. 86, 57?67. Stott, L., Poulsen, C., Lund, S, and Thunell, R. (2002) Super ENSO and global cli? mate oscillations at millennial time scales. Science, Vol. 297, 222?226 (with erratum published in the October 25, 2002 issue). Sutton, R. T., and Hodson, D. L. R. (2007) Climate response to basin?scale warming and cooling of the North Atlantic Ocean. journal of Climate, Vol- 20, 891?907. Tan, M., Hou, J., and Liu, T. (2004) Sun-coupled climate connection between east- ern Asia and northern Atlantic. Geophysical Research Letters, Vol. 31, doi: Timmermann, A., Okumura, Y., An, Clement, A., Dong, B., Guilyardi, E., Hu, A., Jungclaus, J. H., Renold, M., Stocker, T. F., Stouffer, R. J., Sutton, R., Xie, S.-P., and Yin, J. (2007) The influence of a weakening of the Atlantic meridional over- turning circulation on ENSO. journal of Climate, Vol. 20, 4899?4919. Ting, M., Kushnir, Y., Seager, R., and Li, C. (2008') Forced and internal 20lh century SST trends in the North Atlantic. journal of Climate, doi:10.1 Toggweiler, J. R. and Russell, J. (2008) Ocean circulation in a warming climate. Nature, Vol. 451, 286?288. Tung, K. K. and Camp, C. D. (2008) Solar cycle warming at the Earth?s surface in NCEP and data: A linear discriminant analysis. journal of Geophysical Research, Vol. 113, Usoskin, l. G., Berdyugina, S. V., Moss, D., and Sokoloff, D. D. (2007) Long-term persistence of solar active longitudes and its implications for the solar dynamo theory. Advances in Space Research, Vol. 40, 951?958. 1 70 WILLIE W-H. SOON Vage, K., Pickart, R. S., Thierry, V., Reverdin, 6., Lee, C. M., Petrie, B., Agnew, T. A., Wong, A., and Ribergaard, M. H. (2009) Surprising return of deep convection to the subpolar North Atlantic Ocean in winter 20074008. Nature Geoscience, Vol. 2, 67?72. . van Loon, H. and Meehl, G. A. (2008) The response in the Pacific to the sun?s dec- adal peaks and contrasts to cold events in the Southern Oscillation. journal of Atmospheric and Solar-Terrestrial Physics, Vol. 70, 1046?1055. Vellinga, M. and Wu, P. (2004) Low?latitude freshwater influence on centennial variability of the Atlantic thermohaline circulation. journal of Climate, Vol. 17, 4498?451 1. Vidal, L., Labeyrie, L. Cortijo, E., Arnold, M. Duplessy, Michel, E., Becque, 5., and van Weering, T. C. E. (1997) Evidence for changes in the North Atlantic Deep water linked to meltwater surges during the Heinrich events. Earth and Planetary Science Letters, Vol. 146, 13-27. Visbeck, M. (2007) Power of pull. Nature, Vol. 447, 383. Wadley, M. R. and Bigg, G. R. (2002) Impact of flow through the Canadian Archipelago and Bering Strait on the North Atlantic and Arctic circulation: An ocean modeling study. Quarterly journal of the Royal Meteorological Society, Vol. 128, 2187?2203. Wang, Y., Cheng, H., Edwards, R. L., Kong, X., Shao, X., Chen, 5., Wu, J., Jiang, X., Wang, X., and An, Z. (2008) Millennia? and orbital?scale changes in the East Asian monsoon over the past 224,000 years. Nature, Vol. 451, 1090?1093. Wang, Y., Li, S., and Luo, D. (2009) Seasonal response of Asian monsoonal climate to the Atlantic Multidecadal Oscillation. journal of Geophysical Research, in press. Wang, Lean, J. L, and Sheeley, N. R., Jr. (2005) Modeling the Sun?s magnetic field and irradiance since 1713. The Astrophysical journal, Vol. 625, 522?538. Watanabe, M. (2004) Asian jet waveguide and a extension of the North Atlantic Oscillation. journal of Climate, Vol. 17, 4674?4691 . Weng, H. (2005) The influence of the 11 yr solar cycle on the interannual? centennial climate variability. journal of Atmospheric and Solar-Terrestrial Phys- ics, Vol. 67, 793?805. White, W. B. and Liu, Z. (2008a) Resonant excitation of the quasi-decadal oscilla- tion by the 11-yr signal in the Sun?s irradiance. journal of Geophysical Research, Vol. 113, White, W. B. and Liu, Z. (2008b) Non-linear alignment of El Nine to the 11-yr solar cycle. Geophysical Research Letters, Vol. 35, Winton, M. (2006) Amplified Arctic climate change: What does surface albedo feedback have to do with it? Geophysical Research Letters, Vol. 33, doi:10.1029/ 20056L025244. Wu, L., Liu, Z., Gallimore, R., Jacob, R., Lee, D., and Zhong, Y. (2003) Pacific dec- adal variability: The Tropical Pacific mode and the North Pacific mode. journal of Climate, Vol. 16, 1101?1120. Wu, P. and Wood, R. (2008) Convection induced long term freshening of the sub- polar North Atlantic ocean. Climate Dynamics, Vol. 31, 9414356. ARCTIC-MEDIATED CLIMATE VARIATION 71 Wu, Z., Schneider, E. K., Kirtman, B. P., Sarachik, E. S., Huang, N. E., and Tucker, C. J. (2008) The modulated annual cycle: An alternative reference frame for climate anomalies. Climate Dynamics, Vol. 31, 823?841. Yang, J. (1999) A linkage between decadal climate variations in the Labrador Sea and the tropical Atlantic Ocean. Geophysical Research Letters, Vol. 26, 1023-? 1 026. Yang, J. (2005) The Arctic and subarctic Ocean flux of potential vorticity and the Arctic Ocean circulation. journal of Physical Oceanography, Vol. 35, 2387? 2407. Yang, J. (2006) The seasonal variability of the Arctic Ocean Ekman transport and its role in the mixed layer heat and salt fluxes. journal of Climate, Vol. 19, 5366? 5387. Yashayaev, I. (2007) Hydrographic changes in the Labrador Sea, 1960?2005. Progress in Physical Oceanography, Vol. 73, 242?2 76- Yashayaev, I., Bersch, M., and van Aken, H. M. (2007) Spreading of the Labrador Sea water to the lrminger and Iceland basins. Geophysical Research Letters, Vol. 34, doi:10.102912006GL028999- Yashayaev, I. and Loder, J. W. (2009) Enhanced production of the Labrador Sea water in 2008. Geophysical Research Letters, Vol. 36, Zhang, Q., Soon, W. H., Baliunas, S. L., Lockwood, G. W., Skiff, B. A., and Radick, R. R. (1994) A method of determining possible brightness variations of the Sun in past centuries from observations of solar?type stars. The Astrophysical journal Letters, Vol. 427, Zhang, R. and Delworth, T. L. (2005) Simulated tropical response to a substantial weakening of the Atlantic thermohaline circulation. journal of Climate, Vol. 18, 1 85 3?1 860. Zhang, R. and Delworth, T. L. (2007) Impact of the Atlantic multidecadal oscillation on North Pacific climate variability. Geophysical Research Letters, Vol. 34, doi: Zhang, R., Delworth, T. L., and Held, I. M. (2007) Can the Atlantic Ocean drive the observed multidecadal variability in Northern Hemisphere mean temperature? Geophysical Research Letters, Vol. 34, Zhang, R., and Vallis, G. K. (2006) Impact of Great Salinity Anomalies on the low- frequency variability of the North Atlantic climate. journal of Climate, Vol. 19, 4704482. Zhang, X., Sorteberg, A., Zhang, J., Gerdes, R., and Comiso, J. C. (2008) Recent radical shifts of atmospheric circulations and rapid changes in Arctic climate system. Geophysical Research Letters, Vol. 35, Zhou, H. and 8 co-authors (2008) Decoupling of stalagmite?derived Asian summer monsoon records from North Atlantic temperature change during marine oxygen isotope stage 5d. Quaternary Research, Vol. 70, 315?321. Zweng, M. M. and Munchow, A. (2006) Warming and freshening of Baffin Bay, 1916?2003. journal of Geophysical Research Letters, Vol. 111, doi:10.1029/ 2005JC003093. 172 WILLIE W-H. SOON NOTES 'Huth et al. (2006) cautioned that it may not be appropriate to use spatially ?xed indices, such as the North Atlantic Oscillation see however the study of Portis et al., 2001 for a new NAO index with spatially evolving domains) to study plausible solar activity responses since the majority of COAs change their locations depending on the solar cycle phases. In addition, Rodionov et al. (2005), in their careful classi?cation of ?ve types of atmospheric circulation for anomalously warm months and another five types for anomalously cold months in the Bering Sea, found that changes in the position of the Aleutian Low are more important than changes in its central pressure. 2The high alpine stalagmite 6130 record of Holzkamper et al. (2004) covering the Eemian, however, still shows evidence for spectral peaks at 197, 109, and 21 years that can be associated with the Suess/ deVries, Gleissberg, and Hale cycles of solar activity variations. 3White and Liu (2008b) recently reported how the 11-year solar radiative forcing drove not only the quasi-decadal signal in the tropical Pacific sea surface temperature (White and Liu, 2008a) but also was responsible for the 3.6?year ENSO signal and the 22-year quasi-biennial?oscillation signal. The 3.6- year and 22?year SST signals are interpreted and modeled as the third and fifth harmonics of the first- harmonic 11-year period quasi-decadal response to the 11-year solar radiative forcing. Usoskin et al. (2007) provided additional discussion and consideration of intrinsic solar activity variations on the interannual timescale, including the persistence of the third harmonics of the 11-year solar cycles. ?One should note that the absolute level of TSI since 1979 has been measured by satellite-borne cavity radiometers with values ranging from 1360 to 1375 W/mz, and the resolution of this indetermi- nacy requires new measurements with radiometers with more precisely determined pinhole area versus surface area of the cavity radiometers. The absolute value of TSI used for this paper has been arbitrarily tuned so that the mean value for the 1979-present interval is roughly 1366.3 W/m2 (N. Scafetta, 2007, private comm; Scafetta and Willson, 2009). To further comment on the estimates of forcing, the IPCC (2007) AR4 WC 1 reports Section 2.7 188) has recently claimed: The estimates of long-term solar irradiance changes used in the TAR Hoyt and Schatten, 1993; Lean et al., 1995) have been revised downwards, based on new studies indicating that bright solar faculae likely contributed a smaller irradiance increase since the Maunder Minimum than was originally suggested by the range of brightness in Sun- like stars (Hall and Lockwood, 2004; M. [sic] Wang et al., 2005). Figure 2.17 on p. 190 of the IPCC (2007) WC 1 report provides a graphical summary that contrasts the previous estimate by Lean (2000) to the new estimate by Wang et al. (2005; including Lean as co-author). The comparison shows that the older estimate was 3.8 times larger for the deduced inCrease of radiative forcing from the Maunder Minimum to contemporary solar activity minima. But is the quoted claim correct? Several facts clearly suggest that those statements from IPCC AR4 are neither accurate nor authoritative. First, it muSt be pointed out that, although Y.-M. Wang et al. (2005) may have given the impression that their paper actually gives a constraint on how large or small the brightness of the Sun should be, it does not. Their paper was based primarily on the so-called mag- netic flux transport model that was never meant to model any irradiance change or any assessment of the energy budget of the whole Sun. The flux transport model does not even contain any radiative trans- fer calculation. A similar limitation can be noted in the IPCC reference to Hall and Lockwood (2004), which was primarily a paper on solar and stellar magnetic activity rather than on how magnetism and light outputs of the Sun and sunlike stars are linked. Furthermore, it is somewhat puzzling that the following related papers were not cited or discussed: (1) Radick et al. (1998); (2) Giampapa et al. (2006); (3) Hall, Henry et al. (2007); 4) Hall, Lockwood, et al. (2007); and (5) Lockwood et al. (2007). The AR4 WG Chapter 2 authors also ignored the key result published in Zhang et al. (1994) [previously cited in IPCC (2001) Third Assessment Report] that is clearly not outdated or superceded. Therefore, the AR4 quote highlighted here is not a defensible summary of the high-quality scientific research that has been done on TSI forcing. ARCTIC-MEDIATED CLIMATE VARIATION I 73 5See the discussion in Vage et al. (2009) on the multiple factors contributing to the return of deep convection at the Labrador and lrminger seas during the winter season of 2007?2008 despite a fairly low or neutral NAO index and an increased flux of Arctic sea ice to the North Atlantic subpolar basin. 6?Based on a study of the unforced internal variability of the Kiel Climate Model, Park and Latif (2003) recently proposed the separation of their "multidecadaI-scale? with peak spectral powers at roughly 50?100 years) and "multicentenniaI-scale? with peak spectral powers at roughly 300?400 years) variability of the Atlantic MOC, in that the former can be shown to originate in the North Atlantic whereas the latter is driven in the Southern Ocean. The current state of ocean proxy?observation and modeling does not meaningfully warrant such a distinction at this time. 7See also discussion and estimates in Wadley and Big (2002), Jones et al. (2003), Cuny et al. (2005), Prinsenberg and Hamilton (2005), Kwok (2006, 2007), Munchow et al. (2006), Serreze et al. (2006), Zweng and Munchow (2006), Greene et al. (2008), and Condron et al. (2009). athis dipole index was defined as the difference of the annual-mean SSTs from the 60- box and 40?60?5, box in Latif et al. (2006), and is different from a previous def- inition of the difference between the box and 0? box by Latif et al. (2004). 9See Guan and Huang (2008) for additional clari?cation on the key role played by wind stress and tidal dissipation as the external mechanical sources needed to support the and see Adkins et al. (2005) for a suggestion of the thennobaric effects and geothermal heating in explaining the rapid change and instability observed for glacial deep ocean. 10See Bond et al. (1997) and Bond et al. (1999) for the discussion of the muted Dansgaard/ Oeschger?like mode during the Holocene. See also the distinction and clari?cation for rapid and abrupt oscillations during glacial times and the Holocene in Alley (2007) and Benton and Broecker (2008). 11Weng (2005) was referring to the ?80?90 Gleissberg cycle? timescale, but I agree with her that it is probably dif?cult or even pointless to be too speci?c because the objective of our common task is to understand not only any particular spectral featureslcharacteristics, but also the broader scales of the weather?climate continuum. l2See Braun et al. (2008) for additional supports and arguments for their original paper. 74 WILLIE W-H. SOON APPENDIX 1 1369 I I i 5? 1368 If1367 - L1366 I a 0 ?9 .- 8 1365 1363 i, 0 Original Hansen and Lebedeif {193?} values in their Figure 7 Solar radiation NASA GISS Arclir: surface air temperature (20081880 1900 1920 1940 1960 1980 2000 Year Fig. A1. Updated annual mean Arctic surface air temperature anomaly (dotted line) time series (from NASA GISS) correlated with the estimated total solar irradiance (solid line) of Hoyt and Schatten (1993) from 1880?2007. It should be noted that an updated time series from Polyakov et al. (2003) is unavail- able at this time (Polyakov, private cemm., July 24, 2008), so the NASA GISS Arctic temper- ature series is adopted for convenience. Although not strongly affecting the current study on multidecadal to centennial variabil- ity, there are apparent discrepancies between the relative highs of the Arctic temperature values for 1937 and 1938 in the current NASA GISS database compared to previously published values (marked by the open diamond symbols) from Hansen and Lebedeff (1 987). In contrast, the value for the cool year at 1887 remained similar (closed diamond symbol) from the old and new NASA GISS records. It is not obvious how the urban heat island effect can play a dominant role in either the Arctic surface air tem? perature record of Polyakov et al. (2003) or NASA GISS. The vertical dashed line around the year 2000 marks the and year for the previouslypublished result in Soon (2005). APPENDIX 2 ARCTIC-MEDIATED CLIMATE VARIATION 175 Table A1. References on Sun-Climate Oscillation Scales Detected in Multiproxy Archives with a Focus on the BO-Year and (and the necessary 1500uyear scale) Solar Variability Reference Data Proxy Time intervals Scales of vari- ability detected (0) Solar proxies and theories Pipin, 1999 Wagner et al., 2001 and Damon, 2003 Vonmoos eta ., GRIP 2006: Vonmoos, 2005 thesis Horiuchi et al., 2008) Dynamo theory Solar "?Be ice-core 14 - A tree-ring chronology 108a 1033 magnetism IOBE produc- tion rate 14C production rate production rate "?Be concen- tration flux 1) The related 1500?yr scale (broadband 1000-2500 yr) Bond et al., 2001 Bianchi and McCave, 1999 Farmer et al., 2008 Andrews et al., 2009 Hu et al., 2003 Wollenburg et al., 2007 Kim et al., 2007 Arctic Ocean; Off NW Africa Hermatite- stained grains Sortable silt grain size (10?63 pm) Mg/Ca ratios in C. bquor'des X-ray diffraction analysis of <2 mm sediment fraction Biogenic silica Fischer 0t Alkenone, a bulloides) Ice-rafting events Iceland- Sco?and Overflow water SST Drift ice Aquatic productivity Biodiversity of benthic foraminifera SST, upwelling intensity, subtropical gyre General 20-50 B.P. Last 12,000 304?9315 yr B.P. 700-1900 AD. Holocene Holocene Last 12 Last 12 Last 15 Last 24 Last 10 Gleissberg and longer scales 205?yr 88-yr, 208?yr, 2304-yr 88-yr, 205-yr ~200-yr ?1 1 500?yr 670-yr 1500-yr?, 950- yr, 195-yr 1.57?kyr, 0.76? (1 .1 6?kyr, 0.54-kyr Holocene) 2-3-kyr (table continues) 176 WILLIE W-H. SOON Table A1. (Continued) Scales of vari- Reference Location Data Proxy Tlme intervals ability detected Moy et al., Lake Red-color ENSO activity Holocene 1500-yr, 2002 Pallcacaocha intensity 2000-yr (S. Ecuador) (2) North Atlantic Greenland Iceland Stuiver et al., GISP2 5 18O in ice Surface Holocene part 210-yr, 70-yr, 1995; temperature 11-yr, 6.3-yr Grootes and Stuiyerr 1997 Yiou et al., GRIP 5 ?30 in ice Surface Holocene part 2-kyr, 180-yr, 1997 temperature 150-yr, 120-yr Ram and Stolz, GISPZ Laser-light Atmospheric 92~1 4 B.P. 91-yr, 197-yr 1999 scattering dust distributions from ice Mayewski et GISPZ Polar circula- Atmospheric Last 110 1450-yr al., 1997 tion index circulation (glaciochemi- cal data) Fischer and North Na?r Atmospheric 1066?1993 10.4-yr, 62-yr Mieding, Greenland concentration circulation AD. 2005 Traverse (NGT) ice cores Andrews et al., North Iceland Sediment N. Atlantic Last 12 ~200-yr, 2003 magnetic oceano- 125-yr, 88-yr property, graphic grain size conditions Moros et al., North Iceland Quartz content Drift ice Last 12 1.3-kyr, 2006 75-30-yr Sicre et al., North Iceland Alkenones SST Last 2000 yr 20-25 yr 2003 (3) Northern Europe Europe Allen et al., Finnmark, Pollen geo- Vegetation Holocene 1810-yr, 1650- 2007 Norway chemical data history yr, 190-yr Knutz et al., NW Europe! Ice?rafted Glacial margin! 10?27 B.P. 180?220-yr 2007 British Ice debris events meltwater Sheet surges! Atlantic MOC Haltia-Hovi at Lake Varve thickness Lake Last 2000-yr Match A 14C al.. 2007 Lehmilampi, sedimentation series E. Finland -hydrology Swindles et al., Fermanagh, N. Peat humifica? Hydrology 2850 yr BC to 265-yr others 2007 Ireland tion and plant 1000 AD. icrofossil (table continues) CLIMATE VARIATION 177 Table A1. (Continued) Scales of vari- Reference Location Data Proxy Time intervals ability detected Chambers and Four mire sites Wet-dry index Hydrology Last 2000~yr 80-yr, Blackford, in the UK 2001 Holzkamper et Spannagel 5 13O in Hydrology 131?1 18 197-yr, 109-yr, al., 2004 Cave, stalagmite B.P. 21?yr Austrian Alps Mangili etal., Pianico 6 18O in calcite Hydrology 15,500 yr 780-yr, 125 to 2007 paleolake, varve during 195?yr Southern Alps lnterglacial of 400 B.P. (4) North America Yu and Ito, Rice Lake, Mg/Ca ratio of Salinity/ Last 21 00-yr 400-yr, 1999 North Dakota ostracode drought 130-yr, shells frequency 100-yr Anderson, Elk Lake, Varve thickness Aeolian 2000-yr in 7.3 20 to 1992 Minnesota activity/wind kyr-5.3 B.P. 25-yr Dean, 1997 Elk Lake, %Na in Aeolian activity Last 1500?yr 400-yr, 84?yr Minnesota varved lake sediments Wang et al., Fox Hill and Lightness para- Persistent heat 30?14 B.P. 800 to 1 2003 Keller Farm meter, car? and moisture 450 to 550-yr, loesses bonate, %Fe supply 350 to 390-yr Fortin and Canadian Lacustrine varve Hydrology 1550-1986 64-yr, 20 to Lamoureux, Arctic and and boreal AMO A.D. 40?yr 2009 southeastern tree-ring width boreal regions series Schimmelmann Santa Barbara Six major grey Floods and Last 2000-yr et al., 2003 basin flood deposits droughts in varved cycle sediments Douglas et al., Gulf of Biogenic silica, Primary Last 10000-yr 150-yr, 2007 California carbonate, productivity, 350-yr TOC dissolution cycles Patterson et al., Vancouver Sediment color Hydrology, 1400-4700 yr ~75 to 90-yr 2004a, 2004b Island, (X-ray images), ocean B.P. among others NE Paci?c anchovy biological herring scales productivity Springer et al., West Virginia Sr/Ca ratios and Hydrology, Last 7000-yr 715?yr, 550-yr, 2008 8 13C values droughts 455-yr, in stalagmite 210-yr Hubeny et al., Pettaquamscutt Fossil pigment Modes of large- 1024-2004 95.9-yr, 38.5? 2006 River Estuary, Bchle (Bacte? scale climate AD. yr, 11.6-yr, Rhode Island riochloro? variations, 8-yr, 5.5-yr e) NAO AMO (table continues) 178 WILLIE W-H. SOON Table A1. (Continued) Scales of vari? Reference Location Data Proxy Time intervals ability detected Asmeron et al., Southwestern 5 130 values in Hydrology, Last 12,000?yr 1533-yr, 444? 2007 US. stalagmite circulation, yr, 146-yr, droughts 88-yr McCabe et al., Yellowstone Tree-ring and Hydrology, Last 820-yr ~60?yr, ~20?yr 2008 National Park instrumental precipitation, data drought Wilson et al., Gulf of Alaska Tree-rind width Temperature Last 1300-yr 18.7-yr, 50.4- 2007 yr, 90-yr, 38-yr, 24-yr, 14 to 1 5-yr, 9 to 11?yr Wiles et al., Gulf of Alaska, Tree-ring and Hydrology, Last 265-yr 1 16-yr, 76-yr, 2009 Lake Erie lake water Lake Erie 28 to 20?yr, level level 17 to 14~yr, 12-yr, 11.2-yr (5) Gulf of Mexico (GOM) Caribbean, Cariaco Basin Poore et al., Pigmy basin, Abundance of Atlantic Last 5000-yr 512-yr, 2004 northern G. sacculifer movements GOM Poore et al., Core RC12-10, Abundance of Atlantic ITCZ 7.4 to 2.8 550-yr, 21 O-yr 2004 western GOM C. sacculifer movements B.P. Hodell et al., Lakes Chichan- Bulk density Hydrology, Last 2600-yr 208-yr 2001 canab and and 6 18O drought (Yucatan Punta Laguna cycles Peninsula) Nyberg et al., SW Puerto Three mineral Hydrology, Last 2000-yr 217?yr 2001, 2002 Rico magnetic drought parameters, cycles, SST, 6 18O of planktonic foraminifera I Black et al., Cariaco Basin 3 "30 in SST and Last 300-yr 159?yr, 24?yr, 2004 planktic precip-related 10.9-yr G. bullor'des salinity Lund and South of Dry Planktonic Florida current Last 5200-yr 360?yr, 190-yr, Curry, 2004 Tortugas foraminiferal 130?yr, 100- 8 180 yr, and 80-yr (6) Equatorial Tropical Africa Russell and Lake Edward, Mg in calcite Salinity/water Last 5400-yr 1500-yr Johnson, Congo of lake balance/ITCZ 2005a sediment movements Russell and Lake Edward, biogenic Salinity/water Last S400?yr 725-yr, 125-yr, Johnson, Congo silica of lake balance/ITCZ 63-72-yr, and 2005b sediment movements others (table continues) ARCTIC-MEDIATED CLIMATE VARIATION Table A1. (Continued) I79 Scales of vari- Reference Location Data Proxy Time intervals ability detected Stager et al., Lake Victoria, Abundance of Aridity/lake Last 13 2350- to 2550? 1997 East Africa diatom levels yr, species in and others lake sediment KuhImann et Off NW Africa Potassium Proxy of terrig- Last 9 900?yr al., 2004 intensity in enous supply sediment core to marine sed- iment Hanebuth and Off NW Africa Dust supply] Turbidite Last 1 1 900 150-yr Henrich, 2009 (Mauritania) accumulation activity (7) Indian Monsoon Neff et al., Hoti Cave, 3 13O in dated Regional 9?6 B.P. 1018-yr, 226-yr, 2001 northern speleothems precipitation/ 28-yr, 10.7-yr, Oman Indian monsoon 205-yr, 87-yr (tuned) Agnihotri etal., Northeastern Biogenic Intensity of Last 1200-yr 200 20-yr, 2002 Arabian Sea proxies (Corg Indian 105 15-yr, and N) and monsoon 60 10-yr Gupta et al., Northwestern Abundance of Indian Last 11.1-kyr 1550-yr,152-yr, 2005 Arabian Sea, planktic monsoon 137-yr, 114? off Oman G. bulloides yr, 101 ?yr, 89, 83, and 79-yr Fleitmann et Qunf Cave, 8 18O in dated Regional Last 11 220?yr, 140?yr, al., 2003 southern speleothems precipitation] (with some 107-yr, 11- Oman Indian data gaps) and IO-yr monsoon (untuned) Burns et al., Salalah region, Layer Regional Last 780-yr 204-yr, 97?yr, 2002 Oman thickness, precipitation! 19.8-yr, 16.1 - 8 13C and monsoon yr, 12.8-yr, 8150 rainfall and 6.6-yr (in 5180 spectra) East Asia East Asian monsoon Wang et al., Dongge Cave, 6 13O in Regional Last 9000-yr 206-yr, 2 005 southern absoluter precipitation] 159-yr China dated strength of stalagmite Asian monsoon Cosford et al., LianHua Cave, 6 16O in Regional Last 7000-yr 220?yr, 83-yr, 2008 Hunan, China absolutely dated stalagmite precipitation! strength of Asian mon500n 50~yr (table continues) 180 WILLIE W-H. SOON Table A1. (Continued) Scales of vari- Reference Location Data Proxy Time intervals ability detected Zhang et al., WanXiang a 130 in Regional Last 1810-yr 170-yr, 10.5-yr, 2008 Cave, Gansu, absolutely precipitation! 6.4-yr, 5.5-yr China dated strength of slalagmite Asian monsoon Zhong et al., 8. Tarim Basin, Mean grain Hydrology! Last 4000-yr 120?yr, 2007 Xinjian, NW size and other wet-dry 90?yr, and China meaSUres cycles others Lim et al., 2005 Cheju island, Eolian quartz Hydrology! Last 6500-yr 1137-yr, 739- Korea flux Asian dust yr, 214-yr, 162, 137, 127, 1 11-yr Ji et al., 2005 Qinghai Lake, Visible reflec- Hydrology! Last 18 293-yr, 200?yr, Qinghai- tance (redness Asian and 163-yr, 123- Trbetan record/iron indian yr Plateau oxide content) monsoon ii at al., 2009 Qinghai Lake, Abudance of Productivity of Last 18 Durations of Qinghai? bacteriophae anoxygenic APB peaks: Tibetan ophytina phototrophic 60- to 70-yr, Plateau bacteria (APB) 90- to 100-yr, 1 30- to 140-yr, 160- to 1 70-yr, 200- to 210-yr Xu et al., 2006 Hongyuan, 5 18O in peat Temperatures Last 6000?yr Quasi 100-yr eastern cellulose Qinghai? Tibetan Plateau Tan et al., 2003 Shihua Cave, Staglamite Temperatures Last 2650?yr 206?yr, 325-yr Beijing growth layers Hong et al., Jinchuan, 8 18O and Temperature Last 6000-yr 207 (205)-yr 2000, 2001 northeastern 6 13C in peat and and other cen? China cellulose hydrology tennial to mil- lennial scales Wei et al., Beijing, China Instrumental Summer 1724?2005 70?yr, 31 -yr, 2008 rainfall A.D. 20-yr Shen et al., Eastern China Documentary Summer 1470?2000 75- to 115-yr, 2006 records? rainfall/PDO AD. 50- to 70-yr Drought! ?ood Index Chu et al., Eastern China! Documentary Snow events Last 2000 years 281-yr, 103-yr 2008 Korea records Raspopov et Trenshan Tree-ring width Summer 600?2000 A.D. ~200-yr al., 2008 Mountains temperature! and Tibetan precipitation Plateau (table continues?l CLIMATE VARIATION Table A1. (Continued) 181 Reference Location Data Proxy Time intervals Scales of vari? ability detected (9) Other regions and proxies (examples only) Eichler et al., 2009 Sano et al., 2009 Ruzmaikin et al., 2006 van Beynen et al., 2008 Dima et al., 2005 Gedalof et al., 2002 Agnihotri et al., 2008 Belukha Glacier, Siberian Altai Mountain region Northern Vietnam (Mu Cang Chai) Nile River Briars Cave, central Florida Rarotonga coral, Cook Islands, South Paci?c Pacific Ocean (north to south) Peru margin 8 180 from glacier ice core Tree-ring index Water level 6 13C Sr in stalagmite Sr/Ca in coral PCT from Multiproxy? tree rings corals Ti (10) Southern Ocean and Antarctica Lamy et al., 2001 Nielsen et al., 2004 Del monte et al., 2004 Core GeoB 3313-1, southern Chile Site TN057-17, Relative diatom Polar Front, East Atlantic Southern Ocean Vostok and Dome C, East Antactica iron content abundances coarse particles Temperature (March- November) Hydrology, droughts Hydrology Soil productivity/ precipitation SST PDO (Oct? Mar) proxy Ocean productivity Regional - precipitation! variability shift of southern westerlies Summer SST sea ice presence Hydrology] atmospheric circulation (dipolar oscillations) 1250?2000 AD. 1470?2004 AD. 622?1470 A.D. Last 4000-yr 1 727?1996 AD. 1840?1990 A.D. Last 2000?yr Last 7700-yr Last 12.5-kyr 9.3- to 3.5 B.P. 205-yr, 86-yr, 1 0.8-yr to 79-yr, 3.2-yr, 2.5-yr, 2.0-yr 88-yr, 260?yr 170- to 180-yr and other scales ~80-yr, ~25-yr ~85-yr, ~23-yr, ~20-yr 83-yr 22- to 24?yr, 11- to 9.4?yr' 1750 yr 134D-yr (ca. 1500-yr band), 9SU-yr 820?yr (ca. 900-yr band) 1220?yr, 1070- yr, 400-yr, 150-yr 180? to 210-yr(Dome 1 50- to 23 0- yr, 120- to 140?yr [Vostok) (table continues) 182 WILLIE W-H. SOON Table A1. (Continued) Scales of vari- Reference Location Data Proxy Time intervals ability detected Leventer et al., Palmer deep Multi-variables Temperatures Last quasi 1996 basins/ (including and 2500-yr Antarctic magnetic cycles Peninsula susceptibility, diatoms) Masson- EPICA Dome 5 in ice Site Last 5000-yr 833-yr, 220-yr, Delmotte et C, East temperature and 60?yr al., 2004 Antarctica Watanabe et Site 525, H202, Atmospheric 1 890?1 980 1 1-yr al., 1998, Mizuho 5042', circulation AD. 1999 Plateau! coastal East Antarctica Goodwin eta ., Law Dome/East Early winter Mid-latitude 1301?1995 10.5-yr, 3.9 2004 Antarctica sea salt (Na) winter A.D. 3.2-yr, 2.33 aerosol atmospheric 2.18-yr concentration variability McConnell et James Ross Aluminum Atmospheric 1832?1991 10.7? to 13.2- al., 2007 Island, concentration circulation, AD. yr, 21 .3-yr, Antarctic and flux aridity of dust 1.52-yr, 1.8? Peninsula source yr, 3.55- to regions 3.71-yr, 4.73- yr List of abbreviated references to Table A 1: Agnihotri et al. (2002) EPSL, Vol. 198, 521?527. Agnihotri et al. (2008) Vol. 9, Allen et al. (2007) QSR, Vol. 26, 1432?1453. Anderson (1992) Nature, Vol. 358, 51?53. Andrews et al. (2003) EPSL, Vol. 210, 453?465. Andrews et al. (2009) Holocene, Vol. 19, 71?77. Asmeron et al. (2007) Geology, Vol. 35, 1?4. Bianchi and McCave (1999) Nature, Vol. 397, 515?517. Black et al. (2004) Paleoceanography, Vol. 19, Bond et al. (2001) Science, Vol. 294, 2130?2136. Burns et al. (2002) ICR, Vol. 107, Chambers and Blackford (2001) Vol. 16, 329?338. Chu et al. (2008) CRL, Vol. 35, Cosford et al. (2008) EPSL, Vol. 275, 296?307. Dean (1997) Geology Vol. 25, 331?334. Delmonte et al. (2004) Climate Dynamics, Vol. 23, 427?438. Dima et al. (2005) Climate Dynamics, Vol. 25, 329?336. Douglas et al. (2007) QSR, Vol. 26, 115?129. Eichler et al. (2009) CRL, Vol. 36, Farmer et al. (2008) CB, Vol. 9, ARCTIC-MEDIATED CLIMATE VARIATION 1 83 Fischer and Mieding (2005) Climate Dynamics, Vol. 25, 65?74. Fleitmann et al. (2003) Science, Vol. 300, 1737?1739. Fortin and Lamoureux (2009) Climate Dynamics, in press, Gedalof et al. (2002) CRL, Vol. 29, Goodwin et al. (2004) Climate Dynamics, Vol. 22, 783?794. Grootes and Stuiver Vol- 102, 26455?26470. Gupta et al. (2005) URL, Vol. 32, Haltia?Hovi et al. (2007) QSR, Vol. 26, 678?689. Hanebuth and Henrich (2009) QSR, in press, Hansen and Lebedeff Vol. 92, 13345?13372. Hodell et al. (2001) Science, Vol. 292, 1367?1370. Holzkamper et al. (2004) CRL, Vol. 31, Hong et al. (2000) Holocene, Vol. 10, 1?7. Hong et al. (2001) EPSL, Vol. 185, 111?119. Horiuchi et al. (2008) QC, Vol. 3, 253?261. Hoyt and Schatten (1993) [02, Vol. 98, 18895-18906. Hu et al. (2003) Science, Vol. 301, 1890?1893. Hubeny et al. (2006) Geology, Vol. 34, 569?572. )1 etal. (2005) EPSL, Vol. 233, 61?70. )1 et al. (2009) QSR, in press, Kim et al. (2007) Geology Vol. 35, 387?390. Knutz et al. (2007) Paleoceanography, Vol. 22, Kuhlmann et al. (2004) CRL, Vol. 31, Lamy et al. (2001) EPSL, Vol. 185, 369-382. Leventer et al. (1996) GSA Bulletin, Vol. 108, 1626?1644. Lim et al. (2005). QR, Vol. 64, 12?20. Lund and Curry (2004) Paleoceanography, Vol. 1 9, Mangili et al. (2007) QSR, Vol. 26, 1725-1735. Masson?Delmotte et al. (2004) Holocene, Vol. 14, 145?151. Mayewski et al. (1997) IGR, Vol. 102, 26345?26366. McCabe et al. (2008) Ql, Vol. 188, 31?40. McConnell et al. (2007) PNAS, Vol. 104, 5743?5748. Moros et al. (2006) Paleoceanography, Vol. 21, Moy et al. (2002) Nature, Vol. 420, 162-165- Neff et al. (2001) Nature, Vol. 411, 290?293. Nielsen et al. (2004) Geology, Vol. 32, 317?320. Nyberg et al. (2001) QR, Vol. 56, 87?4 02. Nyberg et al. (2002) Vol. 183, 2541. Patterson et al. (2004a) Sedimentary Geology, Vol. 172, 67?84. Patterson et al. (2004b) Palaeontologia Electronica, Vol. 7, 1?17. and Damon (2003) Vol. 108, Pipin (1999) AM, Vol. 346, 295?302. Polyakov et al- (2003) 1. Climate, Vol. 12, 2067?2077. Poore et al. (2004) CRL, Vol. 31 doi:10.102912004GL019940. Ram and Stolz (1999) CRL, Vol. 26, 1043?1046 (and p. 1763 for correction). Raspopov et al. (2008) Vol. 259, 6?16. Russell and Johnson (2005a) CRL, Vol. 32, Russell and Johnson (2005b) QSR, Vol. 24, 1375?1389. Ruzmaikin et al. (2006) ICR, Vol. 1 11, Sano et al. (2009) Climate Dynamics, in press, 1 84 WILLIE W-H. SOON Schimmelmann et al. (2003) Holocene, Vol- 13, 763?778. Shen et al. (2006) CRL, Vol. 33, doi:10.1029000561024804. Sicre et al. (2008) EPSL, Vol. 268, 137?142. Soon (2005) CRL, V01. 32, Springer et al. (2008) CRL, Vol. 35, Stager etal. (1997) QR, Vol. 47, 81?89. Stuiver et al. (1995) QR, Vol. 44, 341?354. Swindles et al. (2007) Vol. 22, 667?679. Tan et al. (2003) CRL, Vol. 30, van Beynen et al. (2008) Q1, Vol. 187, 76?83. Vonmoos (2005) thesis, ETH Zurich, Diss. No. 16224. Vonmoos et al. (2006) IGR, Vol. 111, Wagner et a1. (2001) GRL, Vol. 28, 3034306. Wang H. et al. (2003) Geology, Vol. 31, 179-182. Wang Y. J. et al. (2005) Science, Vol. 308, 854~857. Watanabe et al. (1998) Vol. 76, 447?451. Watanabe eta]. (1999) Polar Meteorology and Glaciology, Vol. 13, 64?74. Wei et al. (2008) IGR, Vol. 113, Wiles et al. (2009) GRL, in press, Wilson et al. (2007) Climate Dynamics, Vol. 28, 425?440. Wollenburg et a1. (2007) Vol. 255, 195?222. Xu et (2006) Vol. 230, 155-?1 64. Yiou et al. Vol. 102, 26441?26454. Yu and Ito (1999) Geology, Vol. 27, 263?266. Zhang P. Z. et al. (2008) Science, Vol. 322, 940?942. Zhong et al. Vol. 22, 659?665. Acronyms employed: Astronomy Astrophysics; EPSL Earth and Planetary Sci- ence Letters; G3 Geochemistry, Geophysics, Geosystems; GRL Geophysical Research Let- ters; GSA Geological Society of America; journal of Geophysical Research; journal of the Meteorological Society of japan; lournal of Quaternary Science; PNAS Proceedings of the National Academy of Sciences of the Palaeogeography, Palae- oclimatology, and Palaeoecology; QG Quaternary Geochronology; Quaternary inter- national; QR Quaternary Research; QSR Quaternary Science Reviews.