UNSTED STATES SECURITIES AND EXCHANGE COMMISSJON 20549 DIVISIONDF 14d. January 26. 1998 CORPORATION FINANCE Jo Anne Murphy Counsel Banhn Exxon Corporation Hum . 5959 Las Colinas Boulevard Irving, Texas 75039?2298 ?g?gb?m /c9?(9 652 Re: Exxon Corporation 1? Incoming letter dated December 18, 1997 Dear Ms. Murphy: This is in response to your letter of December 18, 1997 concerning a shareholder proposal submitted by the Province of the Capuchin Order and several co~proponents {collectively the ?Proponents"}. We have also received correspondence from the Proponents dated January 1998 and January 19, 1998 and from their attorney, Mr. Paul M. Neuhauser dated January 23, 1998. Our response is attached to the enclosed photocopy of your correspondence. By doing this, we avoid having to recite or summarize the facts set forth in the correspondence. Copies of all of the correspondence also will be provided to the Proponents. - In connection with this matter, your attention is directed to the enclosure, which sets forth a brief discussion of the Division?s informal procedures regarding shareholder proposals. Sincerely, Catherine T. Dixon Chief Counsel Enclosures .em_vr.mu_. II .1 i . cc: Paul M. Neuhauser 3485 Richard Circle, Iowa City, Iowa 52240 DIVISION OF CORPORATION FINANCE INFORMAL PROCEDURES REGARDING SHAREHOLDER PROPOSALS The Division of Corporation Finance believes that its responsibility with respect to matters arising under Rule Ida-8 as with other matters under proxy rules, is to aid those who must comply with the rule offering informal advice and suggestions and to determine, initially, whether or not it may be appropriate in a partclular matter to recommend enforcementaction to the Commission. In connection witha shareholder proposal under Rule 14a-8, the Division?s staff considers the information furnished to it by the Company in support of its intention to exlcude the proposals from the Company?s proxy material, as well as any information furnished by the proponent or the proponent?s representative. Although Rule 14a-8(d) does not speci?cally provide for any communications from shareholders to the Commission?s staff, the staff will always consider information concerning alleged violations of the statutes administered by the Commission, including argument as to whether or not activities proposed to he taken would be violative of the statute or rule involved. The receipt by the staff of such information, however, should not be construed as changing the staff?s informal procedures and proxy review into a formal or adversary procedure. It is important to note that the staff?s and Commission?s no-action reponses to Rule 14a-8(d) submissions reflect only informal views. The determinations reached in these no action letters do not and cannot adjudicate the merits of a company?s position with respect to the proposal. Only a court such as a U.S. District Court can decide whether a company is obligated to include shareholder proposals in its proxy material. Accordingly, a discretionary determination not to recommend or take Commission enforcement .action, does not preclude a proponent, or any shareholder of a company, from pursuing any rights he or she may have against the comapny in court, should the management omit the proposal from the company?s proxy material. The Commission staff's role in the shareholder process is explained futher in this statement of the Division?s Informal Procedures for: Shareholder Proposals. EKON CORPORATION JO ANNE MURPHY 5959 LAB COLINAS BOULEVARD, IRVING, TEXAS 75039-22293 Counsel Telephone: (51'2) 444-1 Facsimile: [972] 444?1432 December 18, 199? Of?ce of Chief Counsel Division of Corporation Finance Securities and Exchange Commission 450 Fifth Street, NW. Judiciary Plaza Washington, D.C. 20549 Re: Exxon Corporation - Omission of Shareholder Proposal Under SEC Rule -- Report on Global Warming Dear Sir or Madam: Exxon Corporation has received- ?'om the Province of St. Joseph of the Capuchin Order and several eo?proponents :a proposal requesting a report on global warming and a statement in support thereof for inclusion in the proxy material for its 1998 annual meeting of shareholders. As Exxon intends to omit the proposal and statement from such proxy material, this letter and its enclosures are being sent to the Commission pursuant to paragraph of Rule 14a-8. . Enclosure No. 1 is a copy of the shareholder proposal and supporting statement, as received ?om the proponent. The proposal requests that Exxon?s Board of Directors create a committee of outsidedirectors to independently review and report to shareholders on the impact of climate change on Exxon?s present policies and practices. Exxon believes it may omit this proposal and statement from its 1998 proxy material pursuant to paragraph (3) of Rule l4a?3 for the reasons set forth below. To the extent such reasons are based .on matters of law, this letter represents the opinion of the undersigned- The proposal is contrary to Rule l4a?9, which prohibits false or misleading statements .in proxy statements, and therefore may be omitted under paragraph of Rule l4a?8. While the Of?ce of Chief Counsel Securities and Exchange Commission December 13, 1997 Page 2 supporting statement accurately quotes statements that appeared in a Business Week article, those statements as quoted and discussed by the proponent in the supporting statement are highly misleading in a number of respects. For your reference, Enclosure No. 2 is a copy of the relevant Business Week articie. First and foremost, the entire supporting statement implies a scienti?c certainty on climate change which, in fact, does not exist, as will be demonstrated below. Second, the ?rst quoted paragraph of the supporting statement, which attempts to summarize Mr. Raymond?s comments on climate change in a speech on October 13, 1997 before the 15th World Petroleum Congress in Beijing, misstates Mr. Raymond?s remarks. For your reference, Enclosure No. 3 contains the text of that speech. The supporting statement implies that Mr. Raymond simply offered the three bare-bones conclusions contained in that paragraph. In fact, he posed three basic questions. Is the earth really warming? Does burning fossil fuels cause warming? Do we have a reasonable scienti?c basis for forecasting temperature? Mr. Raymond then offered detailed comments on each question, which re?ected and were consistent with the unsettled state of science today, including material in the ?glltext of the 1995 Second Assessment Report of the United Nation?s Intergovernmental Panel on Climate Change issued last year. That report (the Report?), unlike the supporting statement, cites the uncertainties and caveats surrounding these questions and is the authoritative source referred to in the supporting statement and extensively cited in the Business Week article. For your reference, Enclosure No. 4 contains relevant excerpts from the IPCC Report. A careful reading of the Report shows that signi?cant scienti?c uncertainty remains. By carefully selecting a single sentence from the summary of the IPCC report balance of evidence suggests a discernible human in?uence on global climate?) without explaining the signi?cant caveats to that statement contained in the Report, the supporting statement would mislead shareholders about the true state of scienti?c knowledge. Dr. Benjamin Santer of Lawrence Livermore National Laboratory, the lead author of the Report chapter on the detection of greenhouse warming, was quoted in the May 16, 1997 edition of the respected scienti?c publication Science as saying: ?It?s unfortunate that many people read the media hype before they read the chapter.-.. I think the caveats are there. We say quite clearly that few scientistswould say the attribution issue was a done deal.? See Enclosure No. 5 for a copy of the relevant Science article titled ?Greenhouse Forecasting Still Cloudy.? Even the cited Business Week article states that ?[t]he IPCC report admits there are uncertainties in the science.? See Enclosure No. 2. The proponent, however, chose not to include even that small caveat acknowledged by Business Week. Accordingly, the supporting Of?ce of Chief Counsel Securities and Exchange Commission December 18, 1997 Page 3 statement is misleading in that it presents anunbalanced description of the IPCC's views as expressed in the Report. The supporting statement also states that the past decade global warming itself has become one of the most exhaustively debated subjects in science. The result is a solid consensus on the scienti?c facts." In fact, no such consensus exists. See, a July 1997 paper .by the leaders of the climate change group at the Massachusetts Institute of Technology, which states that scientists have been skeptical about the ?balance of evidence? statement ?om the beginning. This group has grown'substantially over See page 4 of Enclosure No. 6. The supporting statement also Raymond?s statement that only 4% of the carbon dioxide entering the atmosphere is Item human activity. While the supporting statement misleadingly implies that this is Mr. Raymond?s ?contention,? in fact this is one of the few undisputed statistics bearing on climate change and is supported by data from "the IPCC Report. Enclosure No. 7 is an excerpt train the IPCC Report containing a chart that .shows the components of the global carbon cycle. The four upward-pointing arrows at the top of the chart show the carbon dioxide ?ows entering the atmosphere. 0f the total (157.1 the portion due to human activity (7.1 comprised of 5.5 ??om fossil ?re! and cement production and. 1.6 GtCe?yr from changing land?use) is Finally, the supporting statement is misleading in that it dismisses the question :of whether the world is warming or not. This question is relevant because of the three major data sets compiled for this purpose, one shows warming and two do not. The most globally comprehensive temperature data available are from U.S. government satellites which show no warming since 1979 when they began operating. See Enclosure No. 8. The proposal is also so vague and inde?nite as to be misleading within the meaning of Rule Uta-9. The staff has the position that a registrant may omit a proposal that is so vague and inde?nite that neither the shareholders voting on the proposal nor the registrant would be able to determine with any reasonable certainty exactly what action or measures would be taken in the event the proposal were approved. The proposal requests a ?full report about the impact of climate change on our company?s present policies and practices.? Yet, as demonstrated above, there is great uncertainty surrounding the issue of ?climate change.? While the proposal suggests two topics to be covered by the report, it leaves the balance of the contents of this ?full report? to the imagination of the company and the shareholders. Is the proposal asking the company to prepare a full report on all of the differing views offered on this subject by the scienti?c community and others? Is the proposal asking that the company simply accept the views expressed in the cited Business Week article and act under the assumption that climate change is a certainty? If so, what degree of climate changeover what Of?ce of Chief Counsel Securities and Exchange Commission December 18, 1997 Page 4 period of time and with what effects should the company assume for this purpose? The hypothetical possibilities are innumerable. Under these circumstances, neither the company nor any shareholder asked to vote on this proposal could possibly know what the company would be expected to cover in or assume for this ?full report.? Consequently, it is unclear what a voting shareholder would expect the company to do if the proposal were adopted and any actions taken or assumptions made by the company could be signi?cantly different from those envisioned by voting shareholders. Accordingly, the proposal and supporting statement may be omitted under paragraph of Rule l4a?8. Pursuant to paragraph of Rule 14a?8, enclosed are ?ve additional copies of this letter and six copies of the aforementioned enclosure, and a copy of this letter is being sent to the proponent. If you desire further information with respect to this matter, please telephone me at (972) 444-1477, or in my absence, Richard E. Gutman, at (972) 444-1480. Very truly yours, 0 Anne Murphy c: Province of St. Joseph of the Capuchin Order dot? .jarn Enclosure No. 1 Corporate Responsibility Of?ce Province of St. Joseph of the Capuchin Order 1015 North Ninth Street no: 2 0 ?97' FAX: 4142271?0637 November 17,199? m: 1m; 3% LmRRaymondCW RECEIVED NOVZU 1997 ?iig??im?m . ?997 the? Irving, Texas 75039-2293 PETER TOWNSEND Dear Mr. Raymond: I appreciated the opportunity to talk with Frank Sprow and Peter Townsend. last Monday about our concerns related to your Beijing speech and the efforts of Exxon to keep any signi?cant policy change from taking place in the upcoming negotiations on global warming. The Province of St. Joseph of the Capuchin Order is bene?cial .owner of 100 shares of Exxon Corporation stock. Underseparate cover you will receive proof of our ownership. We will retain shares through the annual meeting. I am authorized to notify you of our intention to sponsor the enclosed resolution for consideration by the stockholders at the next annual meeting. I hereby submit it for inclusion in the proxy statement, in accordance Rule 14a-8 of the General Rules and Regulations of the Securities and Exchange Act of 1934. i If you Should; for reasomidesiireto oppose the adoption of this by the stockholders, I pleaseinclude'in-the corporation's proxy material the attached statement of the security holder, submitted in support of this proposal, as required by the aforesaid rules and regulations. Even though our dialogue did not achieve mutual agreement, I do want you to know I much prefer to dialogue than go to an annual meeting- I would hope continued infonnation ?ows might ?nd us determining reason to withdraw our resolution. Sincerely, RECEIVED NUU z; 199? OFFICE OF IHE SECRETARY AFFIL. DOCS. SVCS. 8c PROXY (Rev) Michael H- Crosby, OF Corporate Responsibility Agent EXXON Global Warming WHEREAS, a Business Week article:??lobal Warming: Is There Still Room for Doubt? Sorry, Skeptics- Scientists $213r They Have the Smoking Gun" (1 [103197) stated: On Oct- 13, the CEO of Exxon Corp, Lee R. Raymond, told the 15" World Petroleum- Congress in Beijing three things: First, the world isn't scanning. Second, even if it more, oil and gas ?wouldn't-be the cause. Third;.no one can the liker [unite Mute rise; and would likeljr With Raymond?s make choices about global climate policies, we need an open debateon the science. an analysis oftne risks, ands careful consideration ofthecosts andbene?ts 111:: call for scienti?c debate is 10 years too late. The costs ofdmling with global wanningare emcertaim but in mningitselfhasbecomeoneof?ie most exhaustide debated subjects in science. "the result is a. solid museums on the scienti?c facts. According to the consensus, Raymond's three assertions on: ?song. The article shows that, in critical areas, Mr. Raymond's contentions contradict conclusions of the Intesrgovermnental Panel on Clima?ie Change The was established in 1983 by the World Meteorological Organization and the United Nations Environment Program to assess the-science of climate change, determine its imch on the en?romncm and society, and formulate suategies to respond..Akey poiinisthecontentionthatonlym from human activity This led Mr. to cut this tiny silVer ofthe greenhouse . de?es common sense.? Ham, according to BW: the 30% rise in aunosPhetic carbon-dioxide am is dueto hmnao activity and isresponsible form: warningso?r. Incaudouslangnagegenerated dismssiom?tciPCC "?rebalanceof cvirhnce suggests a discernible human in?uence. on global climate?. John Em Gmup chicr?munvc ofBritish Petrolemn ?The time to consider the policy' dimensionsof climate change is . . . when the possibility cannothe discounted. - We in BP have reached that point." Such an admission from an execntive of a major competitor in the fossil fuel indusn'y believed to be a signi?cant the problem - invites us to seriously reexamine the issue. RESOLVED that shareholders reqoest the Boardto create a committee of its outside directors to independently review (at reasomhlc cost and attaining moprietary information) and make available to shareholders byAugust; [998 clung-on our company's present policies and prac?ces Amongissuesto betroated newcomendthatthese include i) any anticipated liabilities our company may incur from its possible contribution to the problem; 2) what the company can do to reduce carbon dioxide emissions from our fossil ?lcls lt'you agree, please vote Wes." JOHN atnri'rr?lonr STONE rumors -.. -- nu-n-u 4-1- .. Boulder, Colo, a lead author for one of the report?s chapters. ?Literally thou- sands of people wind up reading these . It?s the consensus view of just about everyone who?s chosen to become involved." In June, some 2,400 scien- tists signed a letter saying they en- dorsed the ?ndings. The basics of global warming are sim- ple. Sovcalled greenhouse gasos?in- eluding carbon dioxide and methane? build up in the atmosphere. Carbon dioxide is the most important of the greenhouse gases generated by human activity. The gases trap the sun's heat, likeacarparkedinthe sun with the windows closed. Couple that with a ba- sic factz?eananmt emblem since The implication is that temperatures are ris- ing, and that?s what the was charged with studying. Hum ms. In his speech, Ray- mond acknowledged that some measure- ments show warming but added that ?satellite measurements have shown no wanning trend since the late 1970s. In fact," he concluded, ?the earth is cooler today than it was 20 years ago-" In its 1995 report, the ECG disagreed. It said the temperatln'e at the'earth's surface, where it matters, has increased from onehalf to 1 degree since the late 19th century. 'Ihe 20th century' has been at least as warm as any other century since 1400 A. D., and recent years have been among the warmest on rocord. On the causes of skeptics make the argument that roost of the greenhouse effect comes from wa- ter vapor carbon dioxide entering theatmosphere is due ,_m_human_a_cut?jty. "Leaping to cut this tiny sliver ofthe greenhouse pie. ..de?es common sense,? as Raymond put it Notso, said the WOO?the 30%rise in atmospheric carbon dioxide during the in dustrial era is due to human activity and is responsible for the wanning so far: In cautious language generated by extensive discussion, the true produced what sci- anoe of evidence'suggests a discernible human immense on global climate." Last, skeptics say that predictions of future weaning are notoriously inaccu- rate. Raymond agreed, but the IPcc doesn't. Continuing improvement of computer projections ?has undressed our con?dence in their use for projection of ?lture climate change,? it said. The IPCC concluded that by 2100, temperatures could rise 2 to 5 degrees F, depending partly upon how fast carbon dioxide lev- els'rise. That could lead to a sea-level frequency of drought and ?ooding. The less admits there are uncertain- ties in the science But that doesn?t un? dercut the tree?s conclusions ?This is an ongoing research problem,? says Meehl. As the Kyoto meeting draws near, many in the business community are campaigning vigor-oime against limits on fossil fuel use, saying such curbs could sti?e the economy. Enviiomnentalists are lobbying just as hard for binding limits on greenhouse gas emissions. That?s a political issue, not a scienti?c one. And there the IPCC has no answers. By Paul Rue?bum in New York The Climate Consensus These are the conciusr'ohs reached by the United Nations intergovernmentai Pane! on Cilimate Change UPCC), the authority on global warming.- HOW WARM WILL THE EARTH The IFCC estimates that global surface temperatures could rise 2 to 6 degrees by 2100. Temperatures have risen about half a degree to 1 degree since the late 19th century. ARE HUMANS "The balance of evidence suggests a discernible human influence on global climate," according to the IPCC's 1995 report. WHAT ARE GLOBAL Crop yields will fluctuate, improving in some areas and plummeting in others. Overall global agricultural production probably won't change. Sea level is projected to rise 6 to 38 inches. A possible increase in extreme weather could batter coastlines and cost lives. The warming would cause ?significant loss of life." HOW 60511.? WOULD IMPROVEMENTS Improvements in energy efficiency of 10% to 30% are feasible at little or no cost. Gains of 50% to 60% are possible in some areas. There are manyr options for reducingemissions of greenhouse gases, but some depend on lowering the cost of alternative technologies. The no. 215/4008 scrics. Pack: up to 4.6 times I?an [or snull plan-?1' BUSINESS WEEK ii NOVEMBER 3. 19%? 159 Enclosure No. 4 The pages included in this Enclosure No. 4 are excerpts the Second Assessment of the Intergovernmental Panel on Climate Change (the Report). The Report was completed in 1995 and issued in 1996. These excerpts illustrate the uncertainty and caveats surrounding the issue of global climate change. The IPCC Report is in three volumes; the following pages are attached: lemme I Science Pages 5, 180, 253, 345, 389, 411, 412, 414, 416, 423, 43? II Economicand Social Dimensions Pages 329, 830 Impacts, Adaptation, and Mitigation. Pages 22, 26, 29, 159, U9, 379 Climate Change 1.995 The Science of Climzlte Change Edited by .T. Enlighten, L..G. Meira Filho, BA. .Callander, N. Harris, A. Kattenberg and K. Maskell Produdiun EditmtJA. Lakeman. Contribution of WGI to the Second Assessment Report of the Intergovexmnenml Panel on Climate Change Swnmotyfor Policymakch change in climate is highly unusual in a statistical sense. but does not provide a reason for the change "Attribution" is the of establishing cause and effect relations. including the testing of competing hypotheses. Since the 1990 IPCC Report. considerable progress has. been made in attempts to distinguish between natural and antluopogenic in?uences on climate. This progress has been achieved by including effects of Sulphate aerosols in addition to gremhottse gases. thus leading to more realistic estimates of human-induced radiative forcing- These have then been used in climate models to provide more complete of the human-induced climate-change "signal". In addition. new sirnulations with coupled atmosphere-ocean models have providedirnportant information aboutdecadetocenorry time-settlement! internal climate variability. A furthermajor mean changes to comparisons of modelled and observed spatial and temporal patterns of climate change. The most important results related to the issues of detection and attribution are: - The limited available evidence from proxy climate indicators suggests that the 20th century global mean tomperatute is at least as warm as any other century since at least 1400 AD. Data prior to 1400 are too sparse to allow the reliable estimation of global mean temperature. - Assessments of the statistical signi?cance of the observed global mean surface air temperature trend over the last century have used a variety of new estimates of natural internal and externally forced variability- These are derived from instrumental data. palaeodata, .simple and complex climate models. and statistical models ?tted to observations. Most of these studies have We signi?cant change and show that the observed warming trend is unlikely to be entirely natural in origin. - More convincing recent evidence for the attribution of a human effect on climate is emerging from pattern-based studies. in which the modelled climate response to combined forcing by greenhouse gases and anthropogenic sulphate aerosols is compared with observed geographical. seasonal and vertical patterns of atmospheric temperature-change. These studies show that such pattern correspondences increase with time. as one would expect as an anthropogenic signal increases in strength. Furthermore. the probability is very low that these 5 correspondences could occur by chance as a result of natural internal variability only. The vertical patterns of change are also inconsistem with those expected for solar and volcanic forcing. - Our ability to quantify the human in?uence on global climate is currently limited because the expected signal is still emerging from the noise of natural variability, and because there are uncertainties in key factors- These include the magnitude and patterns of long term natural variability and the time-evolving pattern of forcing by. and response to. changes in concenUations of greenhouse gases and aerosols. and land surface changes. Nevertheless. the balance of evidence suggests that there is a discernible human in?uence on global climate. Climate is expected to continue to change in the future The has developed a range of scenarios. of future greenhouse gas and.aerosol precursor emissions based on assumptions concerning population and economic growth. land-use. technological changes. energy availability and fuel mix during the period 1990 to 2160. Through understanding of the global carbon cycle and of atmospheric chemistry. these emissions can be used to project atmospheric concentrations of greenhouse gases and aerosols and the perturbation of natural radiative forcing. Climate models can then be used to develop projections of future climate. - The increasing realism of simulations of current and past climate by coupled atmosphere-ocean climate models has increased our con?dence in their use for projection of future climate change- important uncertainties remain. but these have been taken into account in the full range of projections of global mean temperature and sea level change. a For the mid?range emission scenario, [392a, assuming the ?best estimate? value of climate sensitivity1 and including the effects of future increases in aerosol. models project an increase in In IPCC reports. climate sensitivity usually refers to the long term (equilibrium) change in global mean surface temperature following a doubling of atmospheric equivalent concentration. More generally. it refers to the equilibrium change in surface .air temperature following a unit change in radiative forcing ("Wmil I80 in? .- millennium Hydrological indicators was" w- am" (amusement?tacos] . sit-W ?oorwocliveclouds {MG-Wmtg?l 1.. {10%deoeasart?3?94} 3, . Observed Cit'mafe l?oriabit'ity and firm ge 'l Figure 3.23: Schematic of observed variations of temperature. Schematic of observed variations of the hydrologic cycle. observed warming cannot be attributed to urbanisation since it is also found in ocean temperatures and re?ected in indirect temperature measurements. The increased precipitation is re?ected in increased streamflow. Despite this consistency, it' should be clear from the earlier parts of this chapter that current data and systems are inadequate for the completedescription of climate change. Virtually every monitoring system and data set requires better data quality and continuity. New monitoring systems, as well as improvements on current systems and studiesto reduce quality historical data, are required. Such improvements are essential, if we are to answer conclusively the questions posed in this chapter. Enormous amounts of meteorological data have been collected and archived over the past century. Even greater amounts will be collected, using new observing systems in Climate Models Evaluation the simulations for a number of common climate variables for both summer and winter in each hemisphere. Of particular note are the relatively large RMS deviations of the cloud radiative forcing (CRF) and of the net ocean surface heat ?ux in the summer hemisphere. which are illustrative of the need for flux adjustment in coupled climate models. We also note the relatively large RMS di?erences in the surface air temperature over land in the winner hemisphere, which emphasise the need for improved parametrization of the surface heat ?uxes and land-surface processes- The simulation of clouds and their seasonal variation remains a major source of uncertainty in atmospheric models. Simulation ofvariabiliry and trends Here we consider atmospheric simulation of variability on a variety of time-scales and their portrayal of recent trends. as a supplement to the discussion of the mean scasonal climate given in Section 5.3.1-1. Table 5.6: Hemispheric mean seasonal madman-square diferencet between observations and the man of the models. IMF HA Variable NH SH NH SH MeansealevelpressurefhPa} 1.4 L4 1.3 2.4 Surface air temperature 2-4 1.6 1-3 2.0 (over land} mar wind furs") (at 100 hPa} 2.4 La 1.3 2.4 Precipitation (mm lday} use on: not on? Gmdiness 10 2t 14 is Dragoing long-wave radiation 2.3 3-2 2.9 5.5 (our: (Wm-i) Cloud radiative forcing (Wm'zj 9.1 20.5 16.2 6.5 Surface flux (Wm-1} 22.5 27.3 30.5 11.2 (over ocean) See Supplementary Table for identi?cation of the models. formeansealcvel pressure 1992} for to 1988. diagnosesof Schemm eral. {1992) and Schubert er a1. {1993) for surface air unperature and precipitation over land for 19?? to 1988. MSU data of Spencer (1993} for precipitation over the oceans for 1979-1983. COADS data of [la Silva et at. U994) for net surface hea?lux over the ocean. and ECMWF TOGA analysesas summarised by Schubert et al. {1992} for the 200 hPa zonal wind for 1985 to 1991. The observed data sources for cloudiness. outgoing long?wave radiation and cloud radiative forcing are as in Figures 5.10. 5.11 and 5-12. 25.? 5.3.l.2.l Diurnal and seasonal ranges Recent observations have shown a reduction in the diurnal range of surface air temperature {nominally at 1.5 or) over the continents (Karl et at, 1993; Horton. 1994; see also Chapter 3.3.2.4) and have suggested a reduction of the diurnal rangeof surface air temperature as carbon dioxide increases (Cao er al.. 1992?. Mitchell et al.. 1995b). In general. the patterns of observed and modelled diurnal range are similar, though modelled values exceed those observed in high northern latitudes in January and over northern continents and deserts in July (cg, Hansen er at, l994). The observed and modelled seasonal range of surface air temperature is shown in Figure 5.14 from the models, in which there is an overestimate of the seasonal amplitude of surface air temperature in the drier regions of the continents. There are, however. considerable differences among the models in higher latitudes. 5.3.1.22 Synoptic variability Figure 5.15 shows the climatology of observed and modelled synoptic variability in DJF and MA. as measured by the standard deviation of daily mean sea level pressure about each constitueni: mean. Simulated values are from the AMIP models. while observations are derived from the Climate Analysis Center Diagnostic Data Base for lQ?lQ to 1933. The models on average capture the broad pattern of the observed synoptic variability quite well, although the models? estimates are less than those observed in regions of maximum storminess. There is. however. considerable disagreement among the observed data themselves, and there are marked differences among the models away from the tropics (Hay et at. 1992; Hulme er al.. 1993}. The Maddenilulian Oscillation (M10) (Siingo and Madden. l991] is a major feature of tropical variability on time-scales of about 30-60 days. Since the physical causes of the M10 are not?fully understood (Roi and Wang, 1990). it is not surprising that analyses of nearequatorial winds at 200 hPa in the AMIP runs (Slingo er al. 1995) show only a modest skill in reproducing the observed MJO magnitude and frequency. 5.3. .13 international variability The ability of models to reproduce the observed mean interannual variability of climate is an important aspect of their performance. but in examining the ability of models to reproduce specific time sequences of interannual variability it must be borne in mind that not all interannual variations are forced by SST) and that a fraction of this variability. often large in the extratropics. is internal to .. .- -.-I-.- - Climate Models Projections of Future Climate resolution information. of the order of a few tens of kilometres-or less. may be necessary to achieve high accuracy in regional and local change scenarios in areas of complex physiography. In the last few years. substantial progress has been achieved in the development of tools for enhancing GCM information. Statistical methods were extendedfrom the to the daily time?scale and nested model experiments were extended to the multi- year time-scale- variable resolution and high resolution global models have become available for use in time-slice mode. While RegCMs allow climate sensitivity experiments to he run at a higher regional resolution. variable and high resolution global models can be used to study possible feedbacks of mesoscale forcings on the generalcirculation. Regional modelling techniques. however. rely critically on the GCM performance in simulating largerscale circulation patterns at the regional scale. because these are a primary input to both empirical and physically based regional models. Although the regional performance of coarse resolution GCMs is still somewhat poor. there are initiations that features such as positioning of storm track and jet stream core are better simulated as the model resolution increases Harrell et at. 1993). The latest nested and variable resolution model experiments. which employed relatively high resolution GCMs and were-run for long simulation times (up to 10 year's) show an improved level of accuracy. Therefore. as a new gecration of higher resolution GCM simulations become available. it is expected that the quality of simulations with regional and local downscaling models will also rapidly improve. In addition. the movement towards coupling regional atmospheric models with appropriately scaled ecological. hydrological. and mesoscale ocean models will not only improve the simulation of climatic sensitivity. but also provide assessments of the joint response of the land surface. atomspl'tcm audior coastal systems to altered forcings. Reducing Uncertainties. Future Model Capabilities and Improved Climate Change Estimates Recent Progress and Anticipated Climate Model Improves:an 6.11.1 . Improvements in the modelling of clouds and associated radiative processes The single largest uncertainty in determining the climate sensitivity to either natural or anthropogenic changes are clouds and their effects on radiation and their role in the hydrological cycle. Although there are many important 345 unresolved issues relating to the basic physics of cloud- radiation interactions and their parametrization in climate models, parametrizations of radiation and cloud optical properties cannot produce realistic radiative ?uxes and heating rates unless they are provided with a realistic distribution of cloudiness. At the present time, weaknesses in the parametrization of cloud formation and dissipation are probably the main impediment to improvements in the simulation of cloud effects on climate. Efforts to overcome this problem have focused on the introduction of cloud microphysics into annoSpheric GCMS Sundqvist 1973; Le Trent and Li 1988?. Smith. 1990; 051?: 1993; Senior and Mitchell. 1993; Fowler et at. l996). There are many dif?culties. The basic .microphysical processes themselves are imperfectly understood. In addition. the large spatial. scale of GCM grid boxes means that microphysical processes occur primarily within sub- grid regions. such as individual cloud cells. whose properties must be determined somehow. For example. according to some microphysics parametrizations. the conversion of small cloud droplets to raindrops occurs when the local small-drop concentration exceeds some threshold. but in a GCM the relevant local concentration is not known; only the generally much smaller grid-box-mmn is available- The distinction between liquid and ice phases. is also of great importance for the inclusion of the commonly observed feature of supercooled water. and hence coexistence of liquid and ice particles in clouds. The difference in saturation vapour pressure then makes the ice particles grow at the expense of the liquid ones (the Berg-eron-Findeisen mechanism). This mechanism enhances not only the in situ rate of release of precipitation, but subsequently also that in layers beneath. due to an enhanced coalescence effect as well. Consequently. whethermodels consider the ice phase or not has a pronounced impact on the resulting water content throughout the cloud depth (Sundqvist. 1993). and. hence on the optical quality of the cloud. Furthermore. the form {rain or snow) in which precipitation reaches the Earth's surface. is in?uenced by the ice and liquid microphysical processes that generate the precipitation. Hence. these processes affect the albedo of land areas. This may be an important factor in positioning and possibly moving the snow line {glacier borders). Consequently. there may be a delicate feedback from enhanced mid-latitude precipitation that is inferred to accompany a warming climate. The distribution of cloud particle sizes is important because it affects both microphysical processes I Changes in Sea [fuel largely ceased- Overall. these results reinforce the conclusions of IPCC (1990) that a long-term ?sea level rise commitment" must accompany greenhouse-gas?induced warming. Thus, even if greenhouse gas concentrations were stabilised. sea level would continue to rise for many centuries because of the large inertia in the ocean-ice- atrrIOSphere climate system. I 7.5.5 Possible oftlre West Antarctic Ice Sheet The West Antarctic lee Sheet (WAIS) is a marine ice sheet - it rests on a bed well below sea level. It has long been argued (Wear-trash. [974) that the WAIS may be inherently unstable because the interior, grounded ice (inland ice} cannot respond fast enough to changes in thickness of the ?oating portions at their junction. the grounding line. It has also been argued (Thomas. 1973. 1935) that the large abutting ice shelves create ?back pressure" which prevents the collapse of the inland ice. such that ice shelf thinning or break-up could cause the grounded ice to ?surge? another critical element eonu-ibuting to marine ice sheet instability. These actions are changing- It is now known that the activity of the WAIS is dominated by fast-flowing, wet- based ice streams whose characteristics blend gradually into those of the ?oating ice shelves and whose response times to changes in the gmundiug line appear to be very rapid (Alley and Whillans, 1991). However, the effects of these dynamic ice streams on the stability of the WMS is very much in dispute. In the view of some glaciologists. the ability of ice streams to transport ice rapidly from the interiorto the ocean. on a time-sale of the order of 100 years; indicates an enhanced capability for a drastically - accelerated discharge. A contrary view is that the short response time of ice streams removes the float imbalance at the grounding line so that the purported instability may not ettist. Recent theoretical work is equivocal. Several recent treatments support the idea that the transition zone between the grounded and ?oating ice does not act as a source of instability (Van der Veen. 1935; Herterich, 1987; Barcillon and MacAyeal. 1993, Lestingant. 1994). On the other hand, there is also support for the idea of instability. which may include the concept of ice shelf buttressing at the grounding line (NASA. 199?. A recent theoretical development that suggests dramatic instability of a marine ice sheet is the so-called ?binge-purge" cycle put forward to explain the massive outpourings of icebergs (Heinrich events) from Northern Hemisphere ice sheets during the last ice age (Alley and MacAyeal, 1994; MacAyeal. 1994). A model study of the WAIS over the last million years that i 389 incorporated ice streams and their slippery beds suggests that the ice sheet did collapse in the past but that the out?ow rates were only a few times faster than at present (Mac?ryeal. 1992). Recent observational work does not present a clear answer either. On the one hand. there is evidence suggesting unstable behaviour of the WAIS: lee Stream is currently ?owing too rapidly for a steady?state; the current growth of the Crary Ice Rise is affecting the regional velocity ?eld and perhaps reducing the discharge of Ice Stream and lee Stream has stagnated in the last years (Retalaff and Bentley. 1993). Furthermore, there is geologic evidence that this ice sheet has been largely or compleme absent at some time after its initial formntion (Seheret', 1991; Burckle, 1993), which suggests transient behaviour in this part of the Ross Ice Shelf system. On the other-hand. there is-evidencc that does not support the notion of WAIS instability: the steady flow for the last years {except for one pulse a few hundred years ago) as suggested by flow tracers in the Ross lee Shelf; the current growth. rather than collapse, of the glaciers feeding Pine Island Bay (which lost its ice shelf in the recent geologic past}: and the lack of evidence of drasticehange in the heightor?owofthe WAIS Station in the last 30.000 years (Whillans, 1976). Given our present knowledge. it is clearthat while the ice sheet has had a very dynamic history, estimating the likelihood of a collapse during the next century is not yet possible. [fcollapse occurs. it will probably be due more to climate changes of the last 10,000 years rather than to greenhouse-induced warming. Nonetheless. such a collapse' once initiated. would be irreversible. Our ignorance of the speci?c circumstances under which West Antarctica might collapse limits the ability to quantify the risk of such an event occurring, either in total or in part, in the next 100 to 1000 years. 1.6 Spatial and Temporal Variability 7.6.1 Geological and Geophysical E?ects The only globally coherent geological contribution to long- term sea level change about which we possess detailed understanding due to a detailed theory of the process is post?glacial rebound (Peltier and Tushingham, 1939: Lambeck, l990}- This is the process by which the solid Earth and the ocean have continued to adjust to the effects of deglaciation throughout the Holocene period {last 10,000 years}. Sea level changes due to longer time?scale geological processes sea floor spreading) are suf?ciently small to be of little interest to this report (cg. fill SUMNIARY Since the 1990 Scienti?c Assessment considerable progress has been made in attempts to identify an' anthropogenic effect on climate. The first area of signi?cant advance is that model experiments are now starting to incorporate the possibleclimatic effects of human-induced changes in sulphate aerosols and stratospheric ozone. The inclusion of these factors has modi?ed in important ways the picture of how climate might respond to human in?uences. Thus. the potential climate change ?signal? due to human activities is now . better defined. although important signal uncertainties still The second area of progress is in better de?ning the background natural variability of the climate system, a crucial aspect of the detection problem. ?Detection of change" is the processof demonstrating that an observed change in climate is highly unusual in a statistical sense. This requires distinguishing any human effects on climate from the background ?noise? of climate ?uctuations that are entirely natural in origin. Such natural ?uctuations can be either purely internal or externally driven. for example by changes. in solar variability or the volcanic dust loading of the atmosphere. ?Total? natural variability includes both internal and externally forced components. Estimating either component of natural climatic noise from observed data is a dif?cult problem. Recent multi-ccntury model experiments that assume no human-induced changes in anthropogenic forcings have provided important information about the possible characteristics of the internal component of total natural variability. However, large uncertainties still apply to current estimates of the magnitude and patterns of natural climate variability, particularly on the decadal~ to centuryaime?scales that are crucial to the detection problem. The third area of progress is in the application of patteme based methods in attempts to attribute some part of the observed changes in climate to human activities: to establish a cause-effect relationship. Most studies that have attempted to detect an anthropogenic effect on climate have used changes in global mean. annually averaged temperature. These investigations have compared observed changes over the past 10?100 years with estimates of internal or total natural variability noise derived from palace-data, climate models. or statistical models ?tted to observations. The majority of these studies show that the observed change in global mean, annually averaged temperature over the last century is unlikely to be due entirely to natural ?uctuations of the climatesystem. Although mete global mean results suggest that there is some anthropogenic component in the observed temperature record. they cannot be considered as compelling evidenceof a clear causeaand4effect link between anthropogenic forcing and changes in the Earth's surface temperature. It is dif?cult to achieve attribution of all or part of a climate change to a speci?c cause or causes using global mean changes only. The dif?culties-arise due to uncertainties in natural internal variability and in the histories and magnitudes of natural and human-induced climate forcings. so that many possible forcing combinations could yield the same curve of observed global mean temperature change. To better address the attribution problem. a number of recent studies have compared observations with the spatial patterns of temperature?change predicted by models in response to anthropogenic forcing. The argument underlying pattern?based approaches is that different forcing mechanisms have different patterns of response or characteristic "?ngerprints". particularly if the response is considered in three or even four dimensions temperature changes as a function of latitude, longitude. height and time}. Thus, a good match between observed and modelled mold-dimensional patterns of climate change increases the likelihood that the ?cause? (forcing change) used in the model experiment is in fact responsible for producing the observed effect. Several recent studies have compared observed patterns of temperature change with model patterns from simulations with simultaneous changes in carbon dioxide (C01) and anthropogenic sulphate aerosols. These comparisons have boon made both at the Earth?s surface 1' . - 1-. . hits-J:? .II. .L . Jar?. 412 and in vertical sections through the atmosphere. The results of these studies rest mainly on pattern similarities at the largest spatial scales, .for which model predictions are most reliable: for example at the scale of temperature differences between hemispheres. land and ocean, or the troposphere and stratosphere. While there are concerns regarding the relatively simple treatment of aerosol effects in model experiments that attempt to define an anthropogenic signal, all such pattern comparison studies show significant correspondences between the observations and model predictions. The pattern eorreSpondences increase with time. as one would expect as an anthropogenic signal increases in strength. Pattern correspondences using combined C02+aerosol signals are generally higher than those obtained if model predictions are based on changes in Ct)2 alone. Furthermore, the probability is very low that these correspondences could occur by chanceas a result of natural internal variability. The vertical patterns of change are also inconsistent with- the response patterns expected for solar and volcanic forcing. Increasing confidence in the emerging identi?cation of a human-induced effect on climate comes primarily from such pattern-based work. Detection nfoinmte Change and Attribution of Causes In addition to these quantitative studies. there are areas of qualitative agreement between observations and those model predictions that either include aerosol effects or do not depend critically on their inclusion. Model and observed commonalities in which we have most confidence, but which have not been looked at in detailed detection studies, include reduction in diurnal temperature range, sea level rise, high latitude precipitation increases, and water vapour and evaporation increases over tropical oceans- Viewed as a whole. these results indicate that the observed trend in global mean temperature over the past 100 years is unlikely to be entirely natural .in origin. More importantly. there is evidence of an emerging pattern of climate response to forcings by greenhouse gases and sulphate aerosols in the observed climate record. This evidence comes from the geographical, seasonal and vertical patterns of temperature change. Taken together. these results pointtowards a human influence on global climate. Our ability to quantify the magnitude of this effect is presently limited by uncertainties in key factors. such as the magnitude of longer-term natural variability and the time-evolving patterns of forcing and response to changes in greenhouse gases. aerosols and other human factors; 414 This hierarchy can be conveniently organised in terms of the spatial and temporal detail considered in comparing observed climate changes with those predicted by some physically based model in response to increasing greenhouse gases)- 8.i.2.i Stage i studies At the base of the hierarchy are studies that deal with the largest spatial scales. relatively coarse temporal information, and a single variable only. Stage 1 studies exist in a variety of different types {see Section 8.4.1). All - such studies to date deal with global or hemispheric mean data for a single variable only. usually annually averaged heartsurface temperature. Stage 1 studies consider whether the observed change in global mean temperature over some period of time (generally the last century or so) is unusual in a statistical. sense. To de?ne "unusual" requires some ?baseline? or ?yardstick? of usual behaviour. It is dif?cult to make direct use of the recent observed data themselves. since signal and noise are intertwined in a complex way. The yardstick most frequently applied is internally generated natural variability. either as simulated by some physically based model Wigley ct oi. 1989'. Wigley and Raper. 1990. 1991a.b; Stouffer at at. 1994). or as derived from a statistical model ?tted to observations Karl et at. 19913.; Bloom?eld and .1992: Woodward and Gray. 1993. 1995}. The observed global mean change is deemed to be signi?cant when it is judged highly unlikely to have resulted from internally generated natural variability alone. Most of the recent work in the detection ?eld has been this type of ?Stage 1? study. A number of these investigations {both pre- and post-[FCC (1990)) have claimed the detection of a highly signi?cant change in observed global mean temperature over the last 100 years. However. none of these studies has convincingly demonstrated that this change can be uniquely attributedto anthropogenic in?uences (see Section 8.4.1). 8.i.2.2 Stage 2 studies Stage 2 studies again involve a single variable. but now compare model and observed patterns of change. generally for temperature. The patterns of change may be de?ned at many locations on the Earth's surface (Barnett and Schlesinger. Santer at at. 1993, 1995a; Karl ct oi. [995$ Hegerl et all. l99?}. in a vertical section through the atmosphere (Karoly at at. 1994; Santer at at. 1995b}. or even in three spatial dimensions [Barnett and Schlesinger. 193i}. Often the model signal pattern has no information on time development. and is either taken from Detection ofCiinmte Change and Attribution of Causes an equilibrium response experiment or consists of a time? averaged "snapshot" from a transient response experiment. In some cases the time-dependent information is contained in a pattern of linear trends (l-lasselmann et at. 1995: Karl er al.. 1995a; Hegerl et aL. 1996}. The observed data always have information on the time evolution of patterns. The next step is to compute measure of the similarity between the modelled and observed patterns of change. The problem is then to determine whether changes .in this statistic with time are large relative. to changes obtained in the absence of any external forcing. As in Stage 1 studies. various yardsticks are used to arrive at a probability level for detection-ode predicred signal in the observed data. Detection of a significant observed change in a Stage 2 study generally implies that even hemispheric- or sub- continental-scale features of observed and model-predicted patterns of change show a level of time-increasing agreement well beyond that expected due to natural variability alone'. if such a pattern correspondence were observed for a particular hypothesised set of causal factors. then (because this requires simultaneous agreement in a number of spatially separated regions} it is less likely that similar agreement could be obtained for a different set of causal factors. Nevertheless. to claim attribution Convincingly still. requires that alternative non- anthropogenic explanations for the observed changes he tested in a rigorous way. as described in Section 8.1.1. In a purely subjective sense. however. a scientist would probably feel more confidence in the attribution of observed changes to a speci?c cause or causes after ?successful detection" in a Stage 2 study than in a Stage I study. The justi?cation for this line of thinking is the implicit assumption that different ?causes? have different climate response patterns. This need not be the case. however. For example. some evidence suggests that the patterns of surface temperature change due to solar forcing and CD2 doubling may be similar. since both climate. forcings operate via similar feedback mechanisms. and these feedbacks dictate the spatial character of the response [Wigley and Jones. 1981: Hansen et al.. [984}. Yet the detailed three-dimensional structure of the thermal response to these two different forcings may well be different Wedierald and Manabe. 1975). In the case of combined forcing by C01 and anthrOpogenic sulphate aerosols. it is unlikely that the spatially distinctive surface temperature response could be produced by other causes. Except in cases where it can be shown that the pattern correspondence statistic largely provides information on global mean changes. 416 lb?hile some pre?lPCCiUEll-i?) studies: had attempted to I estimate the climate effects of andrropogenic aerosols and non-CO2 greenhouse gases in global mean terms Wigley. 1989; Wigley et at. 1989), it is only recently that GCMs have been used to estimate their detailed spatial signatures- Recent model simulations have also provided initial information about the background natural internal variability against which an anthropogenic signal must be detected. This is the seeded major area of progress since IPCC Two modelling groups have recently completed long (grotto years) control runs. which contain a wealth of information on the magnitude. patterns and time-scales of internal climate variability generated by AOGCMS (Delworth et at, 1993; Stouffer et at, 1994; Hasselmann er tel.- 1995). These model-generated noise estimates are now being used as the ?yardsticks? for judging signi?cance in climate-change detection .studies (see Section 8.4), and ?rst attempts are underway to check the consistency of decadal to century time-scale natural variability noise estimates derived from models and palaeodata (Barnett not, 1996); Additionally. information. from palaeoclimatic proxy recordsiis now. being used: to estimate'tlre natural variability of surface temperature-on: hemispheric or even global spatial scales- Statistical methods are the third main area in which advances have been made since IPCC (I990). The last ?ve years have seen an increase in the number and sophistication of statistical techniques that have been applied to the problem of identifying anthropogenic change. Two important examples are the application of Singular Spectrum Analysis (SSA) to partition signal and noise in global mean temperature data Gbil and Vantard. Schlesinger and Ramankutty. 1994) and the definition and application of ?optimal detection" techniques (Hasselmann, 1999. 1993; Bell, 1982; Nordic: at. 1995; North and Kim. 1995). Full optimal detection methods provide real promise for dealing with some :of the statistical problems that will be encountered in Stage 3 and Stage 4-type studies (see Section There has also been significant progress in the application of pattern-based methods to the detection problem. and in the incorporation of re?ned signal and noise estimates in detection studies. Up to 1990. most detection studies were of the Stage 1 type, focusing on global mean temperature- The main exceptions were the pattern-based studies by Barnett (1936} and Barnett and Schlesinger (1937}. During the last ?ve years, a number of research groups have published results from Stage 2 and Stage 3 investigations. in addition to the enhanced Detection of Climate Change and Attribution of Causes greenhouse effect-(Santa ct 1993'. 'Hegcrl (IL. 1996), the searcheddor patterns now include: the signals due to anthropogenic sulphate aerosols '(Karl at at, 19953) and the combined effect of greenhouse gases and sulphate aerosols {Hasselmann et at. 1995; Mitchell at at. 1995a; Santer er at. l995a-b). Although important uncertainties remain in these studies, they have yielded initial evidence for the existence of an anthropogenic effect on climate. Furthermore, we have now started to see pattern-based studies that directly address the attribution question, and try to rule out various non-anthropogenic forcing mechanisms as explanations for some observed pattern of climate change Karon ct cl. l994; Hansen et al, 1995b}. 8.2 Uncertainties in Model Projections of Anthropogenic Change There is no direct historical or palaeoclimatic analogue for the rapid change in atmospheric CO2 that has taken place over the last century (Crowley. l99l}. This means that unless we assume that the signal pattern is independent of the spatial characterof ;the- forcing (see, Hoffert and ?ovey. il992}; we cannot- use palaeodimatic data or instrumental records as analogues for the regional and seasonal patterns and rate of climate change over the next century. We must therefore rely on models for this information. The aim of the following section is to consider the main uncertainties involved in such projections of anthropogenic climate change- 8.2..l Errors in Simulating Current Climate in Uncertpled and Coupled Models - Model validation is one of the most important components in our efforts'to predict future global climate change. Although model performance has generally improved over the last decade. both coupled and uncoupled models still show systematic errors in their representation of the mean state and variability statistics of current climate (see Chapter 5, and also Gates at at, 199i)1 1992}. Such err-om reduce our confidence in the capability of AOGCMs to predict anthropogenic change. 8.2.2 Inadequate Representation of Feedbacks Realistic simulation ofthe present clinratc is probably a necessary- but not suf?cient condition to ensure successful simulation of future climate. To be con?dent that a model has predictive skill on time-scales of decades or longer, we would also have to be sure that it incorporates correcdy all of the physics and feedback mechanisms that are likely to be important as greenhouse gas concentrations or aerosol? Detection of Climate and Attribution of Center 0.10003 5 A . .. mm? is. narrow, 3. .. . - :a 0.0010; _l I c.0001 {1 2'0 40 60 an Trend length {years} Figure 8.3: Signi?cance of observed changes in global mean. annually averaged near-surface temperature. The solid line gives the magnitude of the observed temperature trend (in "Ciwa over the recent recordz+ over the last ll} years (1934 to 1993}. 20 years (1914 to [993], etc- to 100 years. Observed data are from Jones and Briffa {1992]. Model results from three AOGCM control integrations (see Figure 8. l] were used to evaluate whether the observed linear trends are unusually large relative to the trends likely to result from internally generated natural variability of the climate system Linear trends were ?tted to overlapping ?chunks? of the model temperahtre series. thus allowing sampling distributions of trends to be generated for the same It] to year time-scales for which observed temperature were estimated. The 95th perceiItiles of are plotted foreach model control run and teach me length; to underestimate the ?total? natural variability of the real? world climate system. They may also be affected by flux correction procedures. Furthermore, the short length of the three control runs used here precludes reliable estimates of the statistical properties of internally generated variability on time-scales longer than two to three decades. The conclusion that can be drawn from this body of work, and from earlier studies reported in Wigley and Barnett is that the warming trend to date is unlikely to have- owurred entirely by chance due to natural variability of the climate system. Implicit in such studies is a weak attribution statement some (unknown) fraction of the observed trend is being attributed to human in?uences. Any such attribution-related conclusions, however, rest heavily on the reliability of our estimates of both century time?scalc natural variability and the magnitude of the observed global mean warming trend. At best, therefore. trend signi?cance can only provide circumstantial support for the existence of an anthropogenic component to climate change. 423 3.4. i .2 Comparisons ofmodei results and observed data The basicstrategy underlying these studies is to force some form of climate model with anthropogenic audior natural forcing factors and then compare the resulting global mean temperature output with the observed record. The types of model used have varied from what amounts to a ?null? model assuming, unrealistically, that solar forcing and temperature response are proportional and in phase as in the work of Friis-Christensen and Lassen, 1991}, through simple regression models (as in the wad: of Schtinwiese ct ch. 1994, and earlierpapers cited therein] that are unrealistically constrained to have a constant lag between forcing and response, to using an upwelling diffusion- energy balance model (UDIEBM), as in (1990) (Wigleyand Barnett, 1990). In terms of imposed forcing, some studies have considered only a single forcing factor, such as solar variability alone in the work of Friis- Christensen and Lassen, while others have considered multiple forcing factors. both natural and anthropogenic. Since the primary area of uncertainty is in the respOnse to anthropogenic forcing, global mean studies. using only a single ?natural? forcing factor and ignoring any anthropogenic component are inadequate, as pointed out by Schlesinger and Rarnanltutty (1992) and Kelly and Wigley (1992). The main conclusion that can be drawn from these investigations is that the record of global. moan? temperature bewell simulated byfa range of combinations of forcing. Best fits are obtained when: anthropogenic forcing factors are included, and, when this is done, most of the observed trend is found to result from these factors. Within the range of forcing and model parameter uncertainties, there is no inconsistency between observations and the modelled global mean response to anthropogenic in?uences. 8.4.1.3 Empirical estimation ofthe sensitivity of climate to forcing A related univariate approach attempts to use observational data to deduce the so-called climate sensitivity pamneter a measure of how sensitive the climate'systern is to external forcing.1 This is done by running a series of EBM simulations with different sensitivity values and choosing the value that gives a best ?t with observed global mean changes. In such experiments there are a number of ?tuning knobs" other than the sensitivity, both The climate sensitivity is usually expressed'as the equilibrium warming for a doubling. Detection of Climate Change and Attribution of Causes with observed changes in order to assess where there is and where there is not qualitative consistency. This section does not attempt to provide an exhaustive analysis of qualitative consistency for a wide range of climate variables- Instead. we focus on areas that have received recent attention in the scienti?c literature ?ihe potential problems of comparisons between a C02- only signal and observed changes have been discussed above. Where possible. therefore, we also malte use of recent model projections based on forcing by both greenhouse gases and sulphate aerosols- 8-5.2 Global Mean Tempemntre and Surface Temperature Patterns The best available evidence suggests that observed near- stufaoe airtempcramre has increased by to in the last lMyeamfGIapter 3).This.resultis inaccord 1rtvithboth simple and more complex model predictions of the global mean temperature change in response to time-dependent changes in greenhouse gasesand aerosols. As noted in Chapter 6. the time-series of global mean temperature dungessimulated in two recent AOGCM experiments with Mitchell et al.. 1995a} are generally in better. agreement with observations . than the results from oonrpmable experiments with CO2 forcing alone. This agreement does not. however. constitute identi?cation of an anthropogenic effect on climate and may be serendipitous. The degree of consistency between modelled and observed global mean. annually averaged temperature changes depends on a variety of factors. These include the model?s climatesensitivity. the magnitude and sign of simulated multi?decadal variability (and or climate drift). oceanic thermal inertia. and the relative strength of the positive forcing due to greenhouse gases and the negative forcing due to aerosols. It is certainly feasible that qualitative agreement could be due to compensating errors. such as a climate sensitivity that is too high being partially offset by cooling due to a residual drift. or by an overestimated aerosol effect. Nevertheless. the recent modelling wor'lt does show: that it is possible to explain peat changes in a global-scale property of the climate system in a plausible way- More importantly, we have seen initial evidence of qualitative and quantitative consistency between observed patterns of near-surface temperature change and model predictions in response to combined greenhouse gas and aerosol forcing- Pattern correspondence is more dif?cult to achieve than global mean consistency. and hence is more meaningful in the context of attribution- Our best model? 43.7 based estimates of natural internal variability indicate that this correspondence is unlikely to be due to natural causes. Simultaneous modelaobserved agreement in terms of changes in both global means and panems. as in the recent study by Mitchell et al. {1995a}. is even less likely to be a chance occurrence or the result of compensating model errors. 8.5.3. Diurnal Temperature Range Past changes in the diurnal Emperanue range (DTR) also show some similarity in the real world and in model predictions. Several studies in Chapter? reported that. since the i950s. minimum temperatures have increased to three times faster than maximum temperatures over large areas of land in the Northern Hemisphere (Karl et al.. 1991b: Karl et aL. 1993; Hulda and Karl. 1993'. Horton. I995). The result is a reduction in DTR. especially during summer and aummn. - Early attempts to explain these changes were unsuccessful. Model simulations with increases in greenhouse gases alone or in combination with the direct effects of sulphate aerosols could not capmre the amplitude and pattern of the observed changes. New modelling and observational studies.(Hansen et al. 1995b; Stenchikov and Roboclt. 1995) have provided insight into this apparent inconsistency between model projections and the observed decrease of DTR by noting the importance of cloud effects. Thus. Hansen et al. (1995b) found that no plausible combination of forcings other than C01. sulphate aerosols and increases in cloud cover could explain the observed reduction in DTR and the increase in mean temperaturein their model. Whether the required cloud changes. which are in accord with available observations. are also primarily an anthropogenic effect due to the indirect effects of sulphate aerosols; see Chapter 2} or are due to annual climate variability is currently unknown. 3.5.4 Changes in Phase afAttuuaI Temperature cycle A recent study by Thomson (1995} of the mold-century Central England time+series of mean temperatures (and of other long instrumental recordslemphyed a sophisticated spectral analysis technique to discern underlying periodicities in the data. After identifying and removing a signal due to the procession of the Earth?s axis. Thomson found that the phase of the annual cycle of surface air temperature was remarkably stable until roughly 1940. at which time it changed abruptly and at an - unprecedented rate (see Chapter 3). The occurrence of this phase shift at a time when atmospheric greenhouse gas concentrations and anthropogenic aerosol loadings were n? .. - n-mum-? A?s-u?a??n?n-m?n - ..- JIM: Climate Change 1995 Impacts, Adaptationsland Mitigation of Climate Change: Scienti?c-Technical Analyses Edited by Robert T. Watson Marufu C. Zinyowera Richard H- Moss 0133:": tJSrience and Technoiom' Polio}: Zimbabwe Meteomiogieai Service: Bette?: Pacific Northwest Examine 0135:: of the President Nationai Inbom MIT Project Administrator David J. Dokken Contribution of Working Group II to the Second Assessment Report of the Intergovernmental Panel on Ciimate Change Pubiishedfor the Intergovernmental! Panel on Clinmte Change CAMBRIDGE a; UNIVERSITY PRESS Technical Guidelines for Assessing Climate Change imports and Adaptations 26.6.2.3. Point in Time or Continuous Assessment A. distinction can be drawn between considering impacts at dis- crete points in time in the future and examining continuous or little-dependent impacts. The former are characteristic of many climate impact assessments based on doublet-CO2 equivalent scenarios, In contrast, transient climatic scenarios allow time-dependent phenomena and dynamic feedback mechanisms to be examined and socioeconomic adjustments to be considered- 26.6.3. Projecting Environmean Trends in the Absence of Climate Change development of a baseline describing conditions without climate change is crucial, for it is this baseline against which all projected impacts are nreasured. It is highly probable that future changes in other environmental factors will occur even in the absence of climate change. which may be of importance for an exposure unit. Examples, as appropriate, include changes in land use, changes in groundwater level, and changes in air, water, and soil pollution. Most factors are relat- ed to, and projections should be consistent with, trends in socioeconomic factors- Greenhouse gas concentratiOns may also change, but these would usually be linked to climate (which is assumed unchanged here)- Projecting Socioeconomic Trends in the Absence of arm Change Global climate change is projected to occur over time periods that are relatively long in socioeconomic terms- Over that peri- od it is certain that the economy and society will change, even in the absence of climate change. Official projections exist for sonic of these changes, as they are required for planning pur- poses. These vary in their time horizon from several years economic growth, unemployment}, through decades urbanization, industrial development, agricultural pro- duction), to a century or longer population}. 26.6.5- Projecting Future Climate In order to conduct experiments to assess the impacts of cli- mate changc, it is ?rst necessary to obtain a quantitative repre- sentation of the changes in climate themselves. No method yet exists of providing con?dent predictions of future climate. Instead, it is customary to specify a number of plausible future climates- These are referred to as ?climatic scenarios," and they are selected to provide climatic data that are spatially compatible, mutually consistent, freely available or easily derivable, and suitable as inputs to impact models. There are three basic types of scenario of future climate: syn? thetic scenarios, analog scenarios, and scenarios from general circulation models. 329 26.15.35. Scenarios A simple method of specifying a future climate is to adjust the baseline climate in a systematic, though essentially orbits-am, manner. Adjustments might include, .for example, changes in mean annual temperature of 2, etc, or changes in annu- al precipitation of :15, 10, 15%, etc, relative to the baseline eli- mate. Adjustments can he made independently or in combina~ tion. In this way information can be obtained on the following: - Thresholds or discontinuities of response that might occur under a given magnitude or rate of dunge. These may represcnt'lcvelsofchange above whichthe nanne of the response alters (cg, warming may promote plant growth, but very high temperahnes cause heat stress}. Tolerable climate change, which refers to the magnitude orrateofclimatechange thata modeled systemcantol- crate without major disruptive effects (sometimes termed the "critical load"). This type of measure is potentially of value for policy, as it can assist in delin- ing speci?c goals or targets for limiting future climate change. One of the main drawbacks of the approach is that adjustments to combinations of variables may not to be physically plausible or internally consistent 26-6-5-2- Analog Scenarios Analog scenarios are constructed by identifying recorded cli- matic regimes that may serve as analogs for the future climate of a given region. 'l?heserecords can be obtained either from the past (temper-a] analogs) or from another region at the pre- sent (spatial analogs). Temporal analogs are of one types: those based on past instru- mental observations (usually within the last century} and those based on proxy data, using paleoclimatic indicators such as plant or animal remains and sedirnentary- deposits (from the more dis- tartt past geological records}. The main problem with this tech- nique concerns the physical mechanism andboundary conditions mate in the past and a future greenhouse gas?induced warming. Spatial analogs require the identi?cation of regions today hav- ing a climate analogous to the study region in the future. This approach is severely restricted, however, by frequent lack of correspondence between other nonclimatic features of two regions that may be important for a given impact sector.(e.g., daylongth, terrain, soils, or economic development}. 26.6.5.3. Scenarios from General Circulation Models Three?dimensional numerical models of the global climate sys? tem (including atmoISphere, oceans, biosphere, and cryosphere} are the only credible tool currently available for simulating the 330 physical processes that determine global climate; Although simpler models also have been used to simulate the radiative effects of increasing greenhouse gas concentrations, only GCMs. possibly in conjunction with nested regional models. offer the possibility to provide estimates of regional climate change. which are required in impact analysis- GClh?ls produce estimates of climatic variables for a regularnet- wurit of grid points across the globe. Results from about 230 GEMS havebeen reported to date see 1990. 1992}. However. these estimates are highly uncertain because of some nesentationofcloud processes;acoarsespatial resolution(atbest anploying grid cells of some 250-lcm horizontal dimension); generalized topography, disregarding some locally important features; and a simpli?ed representation of land-aunospherc and MW interactions- As a result. GCMs are currently sent-day climate at a regional scale. Thus. GCM outpUIs repre- sent. at best. broad-scale sets of possible futun: climatic condi- tions and should not be regarded as predictions- GCMs have been used to conduct two types of experiments for estimating climate: equilibrium-response and transient- forcing experiments- The majority of experiments have been conducted to evaluate the equilibrium response of the global. climate to an abrupt increase (commonly. a doubling) of nouns pheric concentrations ofCOl. A measure that is widely used in the inter-comparison of various GCMs is the climate-sensitivh - ty parantetcr.; This is de?ned as the: global mean equilibrium . 5 anface air temperature change thatiuccuts in response to an increase in radiative forcing due to a doubling of atmospheric concentration (or equivalent increases in other greenhouse gases}. Values of the parameter obtained from climate model simulations generally fall in the range (IPCC. l992). Knowledge of the climate sensitivity can be useful in com structing climate change scenarios from GCMs. Recent work has focused on fashioning more realistic experiments with (Ellis?speci?cally. simulations of the response of climate to a transient forcing; 'l?hesc simulations offer several advantages over equilibriumresponse experiments. Fust. the speci?cations of the atmospheric perturbation are more realisdc. involving a continuous change over time in greenhouse gas con- centrations- Secondthe representation ismorereal? istic. the most recent simulations coupling annospheric models to dynamic ocean models. Finally. transient simulations provide drangc. which is of considerable value for impact studies. The following types of information are currently available from GCMs for constructing scenarios: - Outputs from a "control" simulation, which assumes ?xed greenhouse gas concentrations. and an ?experi- ment." which assumes future concentrations. In the case of equilibriumrresponse experiments. these are values from'multiple?year model simulations for :the Teclmr'cul Guidelines for Assessing Climate Change Impacts control and 2 (or equivalent increases in ether greenhouse gases) equilibrium conditions. Transient- rcspouse experiments provide values for the control equilibrium conditions and for each year of the tran- sient model run 1990 to 2100). - Values of surface or near-surface climatic variables. for model grid boxes characteristically spaced at intervals of several. km around the globe. - Values of air temperature. precipitation (mean daily rate). and cloud cover. which are commonly supplied for use in impact studies. Data on radiation. wind~ speed. vapor pressure. and other variables are also available from some models. 0 Data averaged over a time period. However. daily or hourly values of certain climatic variables. from which the statistics are derived. may also be stored for a number of years within the full simulation periods- 26.6.6. Rejecting Environmental litersz M??r Climate Change already should havebeenirmporated in thedevelcpment of the lytoclimatechangc. scena?osm?CQ 1990). climate (such as river flows. runoff. erosion) would probably requirefullimpactassessments of theirown. althoughsomernight be irmporatcd as ?automatic adjustments" in projections. 26. 6.1 Projecting Socioeconomic Trends with Climate Change The changes in environmental conditions that are attributable solely to climate change serve as inputs to economic models that project the changes in socioeconomic conditions due to climate change both within the study area and. where relevant and appropriate. outside it over the study period. All other changes in socioeconomic conditions over the period of analy-. sis are atuibutable to nonclimatic factors and should have been included in the estimation of socioeconomic changes in the absence of climate change. I 26.7. Step of Impacts Impacts are estimated as the differences over the study period between the environmental and socioeconomic conditions pron jected to exist without climate change and those that are pro- jected with climate change. Assessments may include the ele- ments described in the following subsections. Climate Change 1-995 Economic and Social Dimensions of Climate Change Edited by James P. Bruce Hoesung Lee Erik F4 Haites Canadian Climate Program Board Korea Energy Econondcs Institute Margaree Consultants Inc. Contribution of Working Group to the Second Assessment Report of the Intergovernmental Panel on Climate Change Published?u? the {mat-gouernmenml Panel on. Climate Change CAMBRIDGE ~11; UNIVERSITY PRESS 22 Climate Change 1' 995 Economic and Social Dimension-r Change 1.1 Introduction In recent decades, atmospheric emissions of greenhouse gases have risen signi?cantly. Concentrations are currently about 25% greater than at the beginning of the Industrial Revolution. if current trends continue, concentrations will double from preindustrial levels before the end of the next century and, if unchecked, continue to rise thereafter (IPCC, I990a). The scienti?c community has noted the potentially serious effects of increased concentrations- These climatic effects could, in turn, have further effects on the biOSphere, including an increaso in mean global temperature, an_increasc in sea level, changes in agricultural yields, forest cover, and water resources, and a possible increase in storm damage- increased concentrations of greenhouse gases are the result of fossil fuel burning, deforestation, livestock raising. and other human activities. Concerted action on the part of indi- viduals and governments will be required to slow the increase in concentrations. Changes in greenhouse gas concentrations and the analysis of the climatic and other physical conse~ quences of those changes lie within the purview of the phys- ical sciences. The role of human activity in generating greenhouse gases, the consequences of those changes for hu- mans, and possible responses lie within the purview of the so- cial sciences. Climate change impacts are likely to vary dramatically from country to country. A warmer climate could benefit sec- tors of the economies of some mid- and high-latimde coun- tries. It is possible that anthropogenic warming might heat the atmosphere enough to prevent or delay another ice age. 0n the other hand, even modest economic losses averaged over the globe could mask large regional losses: a rising sea level and the possibility of increased stonn'surges could threaten the survival of some small island states and coastal areas and could increase the risk of midcontincnt drought and deserti?? cation for inland areas on the periphery of deserts; Such changes could promote human migration and major conflicts as well as famine, disease, and increased mortality. Within the past decade, a consensus has emerged on some key issues in the economics of climate change. This report de- scribes areas of consensus as well as areas of disagreement, the sources of disagreeman and fu'rther research that could narrow the range of disagreement. This chapter frames the is- sue of climate change largely from the point of view of eco- nomics but also from that of other social sciences, introducing the more detailed discussions in the chapters to folio?!- At least two arguments have been offered to justify the commitment of resources to mitigate climate change. The ?rst arises from fundamental values, the second from decision analysis. They may be summarized as follows: (I) We have only one planet. Some changes are largely irre- versible and may occur rapidly. Prudence calls for avoiding a large?scale experiment with the planet- Thus. avoiding anthropogenic climate change lies beyond the scope of normal economic calculation. The potential exists for the occurrence of sudden, largely irreversible..nonlinear changes in the global (2) ecosystem. These would have major economic effects, which would be particularly severe in some countries or regions. Even if the ?rst view is adopted. economics has much to con- tribute to the discussion, for the question of cost-effective emission reductions must still be addressed. [f the second view is adopted, economics and cost-bene?t analysis will clearly be relevant, both-in deciding how much mitigation to undertake and in designing the measures. This chapter, and others in this Assessment Report, draw on the ?ndings of the IPCC's Working Groups I and II, and follow the guidelines provided by the Framework Convention on Climate Change The Convention [eaves open a number of important questions that must be addressed at the political level through future negotiations, including review- ing the adequacy of commitments. It is hoped that the ?ndings of this chapter, and the assessment report more broadly, will contribute to these future negotiations by providing an under- standing of the costs and consequences of alternative actions and their scienti?c basis; 1.2 Features of Climate Change Climate change could impose a variety of impacts on society. Volume 2 of this report analyzes these impacts in detail. They include effects on agriculture, forests, water resources, the costs of heating and cooling, .the impact of sea level rise on small island states andlow?lying coastal areas, and a possible. increase in extrem events storms). Although most at- tention to date has focussed on negative impacts, some im- pacts will be positive. Beyond these tangible impacts are a variety of intangible impacts,I including damages to existing ecosystems and the threat of species losses.1 Climate change presents the analyst with a set of formida- ble complications: large uncertainties, the potential for has versible damages or costs, a very long planning horizon, long time lags between emissions and effects, a global scope, wide regional variations, and multiple greenhouse gases of concern. Large uncertainties. Although natural scientists agree that greenhouse gas concentrations are rising, there remain major uncertainties about the impacts on temperature and climate- These are re?ected in a wide range of estimates of future global mean temperature increases and in uncertainties about regional climate'changes- Esrimates of net economic losses for the most likely range of warming over the next century, and the great uncertainties associated with such estimates, are discussed in Chapter 6. Social scientists do not agree on the size of the behavioural responses or economic effects that would follow or on the effect of these changes on human well- being (Mantle and Richels, 1992; Peck and Teisberg, 1993; Nordhaus, 1993).3 Nonlineorities and irreversibilities. Nonlinearities occur when changes in one variable cause a more than proportionate impact on another variable. Irreversibilities are changes that, once set in motion, cannot be reversed, at least on human timescales. For example, some have suggested that even a modest increase in atmospheric greenhouse gas concentra- 26 Climate Change l995 Errmmuic and Social Dimensions ofCliumre Change assets now. countries will have a richer economy to draw on should climate change damages occur later. A policyr of pre? cautionary investments means investing more than would oth~ envise have been invested. Second. investments can be made (including investment in research and development) that would enhance the economy's ability to adapt should climate change damages occur. Precautionary investments may also enhance the ability of future generations to react. An important reason that people establish savings accounts is to reduce the impact of un- favourable events in the future. Similarly. a society may elect to accumulate capital .against the possibility of a large loss from climate change. This is one thread of the debate over dis- count rates discussed in Chapter 4. Those who argue for a dis- count rate close to the opportunity cost of capital point out that society may choose between immediate greenhouse gas mitigation. at a cost. and delayed mitigation, with some of the money saved put aside as a savings account for our grandchil- dren in the event of large climate-induced damages. 1.3.2 Sequential decision melting As a policy question. global climate change is sometimes posed as a choice between doing nothing at all or com- mitting to all-out effort. Given the large current uncertainties about the costs and bene?ts of greenhouse mitigation. this is the wrong way to frame the issue. as it obscures the choices that should be evaluated. Moreover. in part because option may be perceived as too expensive to get political support. policy paralysis often results- A more useful forntulation is: ?Given current knowledge and concerns. what actions should we take over the-next one or two decades to position ourselves to act on new informa- tion that will become available?? (Lind. 1994). For example. decision makers would like to know if the possibility of irre- versible damages. such as might he suffered by low-lying states. justi?es undertaking an aggressive abatement pro- gramme irrunediately.?i Climate change demands. a decision process. that is Sequen-f rial and can {incorporate new information. Timing will be a; key element. and the-date of resolution of uncertainty an im- portant element of the analysis- Figure :2 shows schemati- cally the progression from'a simple decision to a sequence of linked decisions. In this example. the simple decision might be whether to take aggressive abatement actions now. Let us assume that the uncertainties are resolved in 2005. In the case of sequential decisions. Decisi0n (in 1998} could be whether to take aggressive abatement actions now; Uncertainty {re- solved in 2005} might be the cost of mitigation; Decision 2 (in 2016) might be whether to tighten abatement programmes al- ready in place; and Uncertainty 2 (resolved in 2020) might be the relation between greenhouse concentrations and tempera- ture increase. Since both climate change and new knowledge {learning} are continuous processes. actions to address cli? mate change should be adjusted continuously in the light of new information- A sequential decision?making strategy aims to identify short-term strategies in the face of long-term uncertainty. The next several decades will offer opportunities for learning and Simple decisions Sequential decision making .Dhnvedomenmes .Decisions Figure 1.2: Sequential decision making. making mid-cowse corrections- The relevant question is not ?What is the best course for the next 100 years?" but rather. ?What is the best course for the next few years?" because a pru- dent hedging strategy will allow time to learn and change course. . For example. the choices might be (I) immediate invest- ment in new plant and equipment. aggressive research'and development on greenhouse abatement technology. or (3) dc- ferring large investment for ten years. when the nature and size of the threat are better understood. when costs will pre? sumany have dropped owing to the availability of improved technologies. and thejob can be. done more ef?ciently. Inappropriate interim goals increase the total cost of ad- dressing the problem (Richels and Edmonds. 1993}. For ex- ample. a commitment to certain levels of emissions in cer- tain years fails to take into account the effects 'of temporary economic disturbances on GDP. and thus on emissions and. hence. on the cost of controls. Because of the high cost of being wrong in either direction. the value of infommtion about climate change is likely to be great. In particular. the value of information about the sensitiv- ity of temperature to increases. the temperature damage function.- the GDP growth rate. and the rate=of energy ef?- ciency improvement is likely mm high (Chan. 1992; Peck and Teisberg; 1992: Marine andRichels. 1992;.Nordhaus. The presence of uncertainty along a dynamic path creates an option value. the value of preserving choices for the future. In climate change. the term has been used in two different ways. One stresses the .irreversibilities ofclimate change: mit- igation expenditures now preserve the option of avoiding adaptation expenditure later. The other stresses irreversibili- ties in investment and the cost of premature turnover of capi- tal. Any action taken today changes the options available later or. more precisely. changes the consequences ofany future ac- tion- Sequential decision making focusses on how those con- sequences are affected if one action is taken today rather than another. and on how these consequences are affected if act tions are taken in parallel rather than serially. I -13 Dynamics The problem of greenhouse gas warming involves additions to concentrations resulting from net emissions over extended Introduction: Scope of the Assessment and agricultural activity. Establishing property rights over emissions does not provide guidance for the consideration of these other in?uencing factors. Thus. establishing property rights for the atmosphere is only one of the mechanisms'nec- cssary to regulate climate change. 1.3.4.2 Payingfor on international public good Who should pay for a global public good? Every country faces this question internally in determining who should pay for the public goods it provides. Who should. for instance. pay for pollution control within a country? Economists generally agree on the following principles: First. for the purposes of analysis. it is useful to separate equity. The implication of this principle is that - hocause pollution is a social cost of production (and consump- tion). everyone should be made to pay the full social coats of the pollution they generate- Thus. if there is a social cost to a unit of greenhouse gas emissions. that cost is the same no matter who produces the emissions- All should pay the full so- cial costs of their actions. whether rich or poor. In this perspective. corrective (Pigouvian) taxes should be imposed uniformly. Second. it is inappropriate to redress oil equity issues through climate change initiatives. although climate'change should not aggravate disparities between one region and an- other. No scienti?c consensus exists on the framework for decid- ing the burden of ?nancing mitigation and adaptation. At least four approaches have been proposed to determine how the burdens of taxation should be shared. One approach looks at bene?ts: Just as those who bene?t from private goods must pay for them. those who bene?t from a public good should be made to pay for it. The principle has some force when large di?erences in preferences exiSt within any income class. Pro- viding a particular public good bene?ts some of those individ- uals more than others. creating inequalities in the absence of bene?t taxes. A major problem frequently encountered. par- ticularly for pure public goods. is that it may be dif?cult to de- termine who bene?ts. It is. in general. possible to ascertain the economic benefits of mitigation. and these are likely to be quite unequally distributed- But this principle. by itself. does not fully determine who should bear the costs-wAppropriately designed mitigation strategies will produce a surplus of bene- ?ts over costs. a surplus that must somehow be divided. A second approach looks at ability to pay. It is often held that richer countries (or individuals] should pay more than poorer ones. This approach sometimes rests on the claim-that all people are entitled to a certain minimum consumption {Dasgupta. 1982}. But this principle does not answer the ques- tion of how much extra the richer countries should pay. A third approach is based on contribution to the problem. Because the industrialized countries have cuntributed more than two-thirds of the stock of anthropogenic greenhouse gases in the atmosphere today. this approach seems to suggest that they have a larger responsibility for hearing the costs. On the other hand. by the time greenhouse gas concentrations double from preindustrial levels. the developing countries are projected to be contributing more than half of annual emis? 29 sions. and roughly half of the total stock in the atmosphere (IPCC. 1990a; Cline. 1992}. Thus. under this criterion. the de? veloping countries might eventually pay far more of the miti- gation costs than under the other principles described earlier. Economists have turned to a fourth approach the social welfare function - to answer the question of how much extra different parties should pay. as well as the question of how to distribute the surplus. The discussion of equity below differ~ entiates between the Rawlsian and utilitarian approaches. In this case. both approaches yield similar results: In the absence of incentive problems. both imply that all of the surplus should be allocated to the poorer countries. or that all of the burden of effort should be borne by the richer countries? Yet a different approach holds that social scientists as such have nothing to say about these ethical issues- Coase (1960). for instance. approaches the problem of externalities by em- phasizing that in the absence of bargaining costs. an ef?- cient solution can be obtained by assigning property rights? and this solution is independent of how property rights are assigned-35 Coase also emphasizes the importance of transac- tion costs. which will often influence the choice of policies. A simple approach that yields ef?ciency but does not re- - quire redistribution (and is thus consistent with the two prin- ciples enunciated above) requires coordinated tax rates so that all countries face the same energy prices. This approach makes the cost of emitting an extra tonne of carbon equal across all countries. with each country retaining the revenues thus The net cost of such a tax (ignoring the ben- e?ts from reduced greenhouse gas emissions) will. in general. be smaller for poorer countries. as a percentage of their na- tional output. The burden of the tax is progressive in its distri- bution across countries. even though the tax is levied at the same rate in all countries?"Jr Accounting for post emissions. Article.3.l of the Frame- work Convention on Climate Change directs the Annex I developed) countries to take the lead in responding to the threat of climate change ?on the basis of equity and in accor- dance with their common but differentiated responsibilities and respective capabilities.? Some have argued in addition that because the industrialized countries have been the major contributors to current levels of greenhouse gases. they should bear most of the costs of mitigation. This view says that costs should be borne. not in proportion to bene?ts ex- pected. but in proportion to contribution to pollution. This ar- gument. however. is not based on the principle of economic ef?ciency. Ef?ciency requires that incentives be prospective (forward looking]. not retrospective.m No incentive effects re? sult from imposing charges based on post actions. Whether to charge nations that contributed Ct?!2 to the atmosphere is an is- sue of ethics. not ef?ciency-39 The controversial issues of population growth and con- sumption patterns. although central to economic develop- ment. bear on climate change largely through their effects on emissions. Population growth in developing countries may also exacerbate the ecological and socioeconomic impacts of climate change. At the same time. high per capita consump~ tion in industrialized countries. where populations have nearly stabilized. will also affect mitigation and adaptation Applicability of Techniques of Cost-Bene?t Analysis to Climate Change Table 5.2. Comparisons of cost estimates for C02 abatement (in of C02) Study type Wind No coal 511' Lilith Systemic. single ti'l- I 3 16-25 303 MW wind farm GEF Generic 1 16423 45-39 Supply cost l50~600 Absolute Cost {lift of carbon} SeanceEFor Sri Lanka, Meier et ul. (1993]; for India- Hussein and Sinha (1993}. The generic estimate is from London Economics India, which has available relatively large quantities of low- cost domestic coal. Similarly, in the generic cost estimates nude by the GEF for no-coal options, the presumed substitute fuel isdomestie oil or gas, not imported oil (as is the case for Sri Lenka). The point is that there likely exist very large, country-speci?c variations in estimates of the cost of green- house gas emission abatement through given technologies. Joint costs and bene?ts that are local in nature will very likely vary even-more from place to place. Another way of making the same point is by noting that single-point cost estimates for many technologies can be quite meaningless, because the supply curve for even an individual technology is not ?at. For example, as Hussein and Sinha (I993) have shown in a recent study of wind and hydro for In- dia, the supply curves have the expected classical. upward sloping shape as illustrated-in Figure 5.9 for the supply cost of wind farms in India. Such static cost curves neglect the coun- terargurnent that. in a more dynamic analysis, cost may dc- cline due to economies of scale and technological advances- 5.5 Issues 5.5.1 Risk, uncertainty, and irreversibility Our knowledge about how anthropogenic emissions of green- house gases affect global temperature, what kind of effects a change in global temperature may have, and how efforts to mitigate climate change may wont is clearly restricted. How different greenhouse gases react in the atmosphere is not fully understood, and even if exact predictions of the average in- crease in global temperature could be made, the different re- gional effects of these increases will be exceedingly difficult to foresee. There is also considerable uncertainty about the economic and social effects of abatement measures, which are decisive fur determining their associated costs and bene?ts. One cannot, therefore, evaluate climate measures without taking these uncertainties into account. On the other hand, ac- knowledgment of vast uncertainties should not lead to an inert attitude but rather to the development of rational strategies for titling under uncertainty. Economic analysis under uncer? tainty aims at developing strategies for decision makers who face uncertainties in future costs and bene?ts. The uncertainty 159 soon 5500 sun Cumulative Offset (Mt of carbon) Figure 5.9: Supply cost curve for wind farms in India. in the outcome of a variable is often described by a probabil- ity attached to each possible . outcome; In some cases, the probability distribution is objectively presented; in such cases one normally talks about risk. More often, subjective proba- bility distributions are assumed, in which case one talks about uncertainty. In Section 5.4 we noted that the level of uncertainty is greater in the benefits curve than in the cost curve (recall Fig? use Figure 5.10 illustrates the practical consequence of uncutaintyin a different way. Figure 5.10s depicts the cost- bene?t analysis of the optimum level of emission reduction. The optimal reduction, is given by the point at which MB MC. Ifthe marginal mitigation cost and marginal dam- age cost functions were known with certainty or if, in the ab? seoce of risk aversion, one were to use expected values of the uncertain mitigation cost and damage functions, then the opti~ mum degree of emission reduction would be as shown. However, given uncertainty in the marginal mitigation cost and damage functions, the optimal emission reduction cannot be determined precisely. It could lie anywhere within a rela? tively wide range. Uncertainty in the damage function and risk aversion lead one to a ?precautionary approach,? which requires more stringent emission reductions, lying to the right of the expected value out and roughly determined by the in- tersection of the cost curve and sculenoticnal upper envelope estimate of the damage function, as indicated by in Figure 5.5. Figure 5.10!) illustrates-the case in which the damages are suf?ciently uncertain that a marginal damage function cannot be de?ned- The risks associated with various emission levels are considered, using the best available evidence. This infor- mation, together with the associated costs, is used to select an emission reduction, is?, that censtitutes an a?'orduble safe minimum standard. Analytically, such a standard. would have to be based on a multicriteria analysis. Finally, in Figure 5.100 the emission reduction RM is based solely on a scienti?c assessment- This corresponds to the ?rst of the two views of the sustainability approach dis? cussed in Chapter ti. Since the obligation to avoid harm is abw solute, the cost of avoiding harm is irrelevant. The bene?ts of 179 EUMMARY This chapter is concerned with the socioeconomic assessment . of climate change Estimates of damage related to these impacts can make an important contribution to decision mining about climate change respOnses. Monetary values reflecting human preferences can provide useful information for decision making. The cost-bene?t ap- proach. in particular. requires that the damages from climate. change be represented. as far as possible. in termsof money units. To the extent that this is possible. the chapter expresses impacts in these terms: that is. human preferencesare ex- preswd by people's willingness .to to secure a ben- e?t or their willingness to accept compensation for a cost- Such monetary estimates only measure the impact on in-. dividual welfare. Aggregating individual damages to obtain to- tal social welfare impaCts requires dif?cult ethical decisions. Many of the impacts of climate change will not be revealed- directly in the marketplace. These are the so?called nonmarket impacts- In these cases andfor WTA are measured through ?surrogate markets" or ?1typotltetical markets.? Surro- gate markets are real markets in which environmental change has an in?uence: A house priceor value may be higher of an environmental amen for example. Hypothet-. ical' markets reflect people?s restrooms to questions put It): them about their willingness to pay. Monetary estimates are thus able to cover both market and nonmarltet impacts, al- though estimates of the latter are more controversial and less con?dence is placed in them. The level of sophistication in socioeconomic assessments- of climate change impacts is still rather modest. Damage esti- mates are tentative and based on a number of simplifying and often controversial assumptions- Most estimates are for equi- librium climate change associated with a doubling of the preindustrial (SDI-equivalent concentration of all greenhouse gases. Best-guess central estimates of global damage, includ- ing nonmarket impacts. are in the order of of world GNP for 2x013z concentrations and equilibrium climate change. This means that if a doubling of C0: occurred now. . it would impose this much damage on the world economy now. This chapter stresses the uncertain chameter'of these- estimates- The ?gures are best-guess results. and several im- pact categories could not be assessed for lack of data- More- over. the range reflects variations in the best~guess estimates and cannot be interpreted as a con?dence interval- Particu- larly vulnerable sectors include agriculture. the coastal zones. human mortality, and natural ecosystems. The possibility of catastrophes (low probability-thigh impact events] and sur- prises cannot be ignored- The regional variation in damage is substantial. The avail- able studies estimate damages for developed countries at between climate. Central es- timates of the damage in different developing regions range from a minimum of 2% of a maximum of For in- dividual nations. or if alternative assumptions are used about the value of a statistical life (see Box 6.1), the ?gure could be even higher. Small island states and low-lying coastal areas are especially vulnerable. Most impact work is con?ned to de- veloped nations. however. The con?dence in estimates for de? veloping countries .is much lower; The chapter emphasizes the need for a long-tenn perspec- tive reaching beyond a 2x002 scenario. even though socio- economic forecasts over more than a century are highly uncertain. Most models assume a nonlinear (convex) damage- temperature relationship. resulting in damages of 6% or higher for warming. These ?gures are illusuativc only. Doubled-CO'zdamage estimates usually form the basis for: the calculation of marginal damage the extra damage done by one extra tonne of carbon emitted. Marginal damage is esti- mated at 335-3125 per tonne of carbon emitted now. The wide range re?ects variations in model assumptions. as well as the high sensitivity of ?gures to the choice of discount rate. Al- though estimates'based on a social .rate of time preference (discount rate) of the order of 5% tend to be about $35-$12. ures assuming a rate of 2% or less can be almost an order of magnitude higher. Current models are simplistic and provide poor representations of dynamic processes. The effect of oil: mate change adaptation in particular is poorly understood. Marginal climate change damage is equal to the marginal climate change bene?ts of emissiOn control. However, the bene?ts of greenhouse gas abatement will not be limited to re? duced climate change costs alone. A reduction in emis- sions will often also reduce other environmental problems relatedto the combustion of fossil fuels- The size of these so- called secondary bene?ts is strongly site dependent. Studies for Norway. the UK and some other countries indicate that the bene?ts of reduced air pollution could offset between 30% and 100% of abatement costs. lnregrnted Assessment of Climate Change: Art and Comparison and Results greenhouse-related. gases. Agriculture. livestock. and forestry represent the most extensive anthropogenic uses of land. In addition. agriculture and livestock are important determinants of CH. and N10 releases. ?nally. full-scale IAMs must consider the array of other greenhouse-related emissions to the atmosphere. Most promi- nent among these are the chloro?uorocarbons and their sub- stitutes. although there are others. From the perspective of the consequences of climate change. an overlapping but somewhat different list of issues must .also be dealt with by LAMS. The problem of climate change impacts is more dif?cult to deal with in Ms because impacts are anticipated to affect a wide array of human?activi- ties. with no single activity thought to be substantially more vulnerable than others. [Alvis thus frequently confront the im* pads issue abstractly. using ?damge functions." rather than explicitly. Nevertheless. underlying any treatment of impacts within an 1AM are. at a minimum. the following human activ- - agriculture. livestock. and forest systems - energy systems - coastal zones . water systems a human health - the value of local air quality - the values of unmanaged ccosystems' The second information set that a full-scale must gener- ate is the concentrations of greenhouse gases. which the model must translate from both natural emissions and the emission flows generated by human antivitics. Greenhouse gas concentrations also depend on natural sources and sinks. In general. greenhouse gases can be segregated into and other gases. The non-(Pitt1 greenhouse-related gases are con- trolled by atmospheric processes. Their sinks are predomi- nantly in the atmosphere. C02. on the other hand. is governed by the proceSSes of the carbon cycle- The concentration of (201 in the atmosphere is determined predominantly by inter- actions between atmospheric concentrations and the oceans and terrestrial systems. Models deal with in a variety of ways. ranging from simple airborne fraction models. which use a proportional approximation method to determine atmospheric concentra- tions. to interactive process models of the atmosphere and biosphere- The present understanding of both the carbon cycle and atmospheric chemisz have been surveyed in Volume '1 of the present report and in previous IPCC scienti?c reports {see 1990. 1992. 1995}. Full-scale [Alvis should ultimately also consider the prob- lem of local air quality. as the removal rates for local air polluv rants depend on weather conditions. and greenhouse gas abatement in?uences local air quality. These factors. in turn. interact with the economic value of changes in health condi- tions. The inclusion of local air quality is not yet possible. however. because of the totally different spatial and temporal scales :and aggregation levels of the climate change and local 379 air pollution problems. At the moment. such analyses can only be done through case studies, Such as those being done by RWM {the Dutch National Institute of Public Health and En- vironmental Protection} for 25 megacities around the world. Chapter 6 assesses these so?called ?secondary bene?ts" of greenhouse gas abatement. The third information set that a full-scale must gener- ate is the state of climate and sea level- Climate cannot be de? rived without dealing in one way or another with oceans. Oceans are an imponant determinant of the timing of climate change. as they represent an enormous heat sink. Thus. ocean- feedbacks also in?uence the rate of sea level rise- In addition. interactions between the atmosphere and cryo- sphere affect climate change and sea level. Sea level calcula- tions. for example. must include changes in the volume of mettwater from the major land~based ice sheets. Furthermore. the ocean that interacts with atmospheric processes in de- termining climate and sea level change also absorbs carbon that has been accounted for in the atmospheric composition model. In Figure 10.1. the fourth category of information is ecosystems. This category includes information associated with natural emissions of greenhouse+relatcd gases. the terres- trial carbon cycle. and the effect of climate change. sea level rise. and (34131 on crops. pastures. grazing lands. forests. hy+ drology. and unmanaged ecosystems. These systems are strongly interactive. Some models han- dle them in a holistic manner. explicitly considering the inter- actions of natural system emissions. the status of unmanaged ecosystems. hydrology. ground cover. crop. and foresr pro- ductivity. Other models treat them as if they were indepen- dent. The managed biosphere interacts strongly with human systems. which determine the selection of crop and managed forest species and the allocation of water resources among competing ends. Interactions between ecosystems and the cli- mate and sea level functions are presently thought to be of second-order importance and are not dealt with in a majority of IAMs. In addition to the degree of complexity {including disag? gregation) considered within and between modules. another major design consideration in an integrated assessment model is the treatment of the considerable uncertainties about virtu? ally every major relationship in the climate change assess- ment system. Future population and economic growth are uncertain: future greenhouse gas emissions. given pupulation and economic activity. are uncertain; future greenhouse gas concentrations. given emissions. are uncertain; future climate. - given of greenhouse gases.gis u?n- . certain; future physical impacts-of climate change are un- cenain; and the future valuation of the physical impacts attributable to climate change is uncertain. Uncertainty can be handled in a number of ways in inte- grated assessrnent modelling. Extensive sensitivity analyses can be performed on key model inputs and parameters. or ex- plicit subjective probabilities can be assessed for these inputs and parameters and fed into a- formal risk or decision analysis framework. if a formal risk or decision analysis approach is pursued. it is generally possible to calculate the value of infor? Enclosure N0. - .- . Greenhouse Forecasting Cloudy An international panel has suggest a decade before computer to ed that global wanning has arrived. but many scientists say it will be odels can con?dently link the wanning to human activities Mahmoud will entrance: 'Global justly (Jim. Experts 011-. . BWde'sLinlttoWaming?' all: Grunge report Surdies (GEES) in New-York City. ?but! mom??madam ?prep-reme I a time-afasueryn res gas-I that cherenydurtomld- maybe we'll have In 5 Minions to admit ym'hndinrludiugdse ?Mealtime of mmto?mto" ma ?50 was. my we Merrie-rt: airman er: 'lanaller. Malamute-warmth: a??ect climate. from the howbadgeen- house warning will be movement ofooean ed- nhenltmives. . dies. Because they can't Windward-the Hangs! tiaml?odelsean'lrarxnmoedords. Mohawk-mind media excitement was owNmmAmkamod- the [Part conclusion tempera- tlnt'dle half-degree rise in global tempera- nrrerlneethelste maybeara of human activity. The patchy disuihudorl of the warming around the globe bohmuchliltc the distinctive pattern if the heat-trapping gases being painted into the aorrosphere were beginning to warm-the planet. the report said. But IP03 scientists now say that neither the not many scientists appreciate how many il's, ends. and but's peppered the re port. ?It?s unfornmate that many people read the media hype before they read the drapter' on the detection of greenhouse winning. says climate modeler Benjamin Santer of Lawrence liver-more National laboratory in Livermore. California. the lead authorof the chapter. "1 think the cave- at: are dtere. We say quite clearly that few scientists would say the attribution issue was a done deal-" Santer and his colleagues' overt id ing team for stressing the caveats is their undentanding of the uncertainty inherent in 1040 hymdeuec?ngthe urivlld'globalwarm- tug. measurement-- to ?nancing levels ofgreenhot?e'gu?. And while predicting climate has always definitely: increase.? my: climate modelerDavidRind of Goddard for Space myself am not convinced that we have [gained] greater confidant" in recent years in our predictions of greenhame urarming. Says one senior climate modeler who pre- fers not to enter the fray publicly: ?The? more you learn. the more you understand that yOu don't understand very much." indeed. most modelers now agree that the climate models will not be able to link greenhouse warming unambiguously to human aetions for a decade or more. The effort to simulate climate in a com? puter 1faces two kinds of obsracles: lack of computer power and a still very incomplete picture of how real?world climate works. The climate forecasters' basic strategy is to build a mathematical model that recreates glo? bal climate processes as closely as possible. let the model run. and then test it by com- paring the results to the historical climate record- But even with today's powerful super- computers. that is a daunting challenge. says modeler Michael Schlesinger ol' the University Urbana?Champaign: dieteare Headers ofmagninldc ofsalc. Earn the nb?emidlhdom?limmenen?downm ouuhichwaternporcmel'meeplmema Chinese 14 orders ofmagnlusdquaues Schlesinger. researchetsate ableto include in their models only the two largest. the tures and humidities that will spawn di?lu- 5 Ifthcoeoonditionshold within a single grid box?the horizontal square that represents the model's ?nest level of detail?the computer cormts the entire area as cloudy. But as modelers point out. the grid used in today's models?typi- eally a 300-ltilometer-square?is still very coarse. One over the slate of Oregon. for instance, would take in the coastal ocean, the low coast ranges. the Willamette Valley. the high Cascades. and the desert of the Great Basin. Having the computer power to incorpo? rate into the models a more detailed picture of clouds wouldn't eliminate uncertainties. l?l?l?tvtl." because researchers are still hotly debating the overall impale: ofcl0uds on fu- ture climate. In today's climate, the net efr fect of clouds is to cool the pin ne t?although they trap some hear. they block even more by re?ecring sunlight back into space. How that balance. would change under greenhouse warming. no one knows. A few years ago. 3 SCIENCE VOL 21?6 - [6 MAY 199? - Model Gets-It Right?Without Fudge Factors Climate modelers have been ?dieatuig? for so long it's almost become respectable The problem has been that no computer model could reliably simulate the presentclimate. Even the best simulations ofthebehayiorofthe atmosphere. ooearLaca ice. and land ma?a: drift oil into a climate quite unlike today's as they nmforeenmries. Sodimate modelershaye gotten in diehabitof ?ddling widi?idgefacmts; atljustitierrtit.ll until the modelgetsitrighr. - . . No one liked this practice 9 September 1994.; p. 1528). 'Ifyou can't simulate the present adjust: the model?s sensitivity to greenhouse gases near the low end of current estimates. Based on an array ofdifferent models and other considerations. the Intergovernmental Panel on Climate Change estimated in 1995 that a carbon dioxide dou- bling could raise global ternperarines by as much as their .- pro- duced slow. in global temperature of about if- the real the same way. ?two-dihdsto l30years meats. have to worry.? says .meteoru modeler David Randall of Colorado State University (CSUJ in Hart Collins. But now there's'a alternative-Thirty researchers at the National Center- fOtAunmpheric Research in Boulder. Colo- rado. have developed the first complete model that tit I I I lntaeasirtg'Co {136m .. p. WE: HEAR eansirnulate the presentclimateas weilasodietmodelt do. but without flux adjustments. The new NEAR. model. says Randall. "is an important step mward're' movingsomeoi'die uneasiness people in i All .. chi? a in: these models to make predictions ofl'uture iclirrtaifc'I (see main text). . -- . -- modelers builtahostdre?nements into their new Climate Systemllilodel (CBM). But the developlnennsaysz-chairByron Nomi-rafter nirlingpoolsofuauertmmaoouple eddies. Iikeatmoqthetic more. help shapedimate byrriovit?tgheatamimdthe Surlaea Temperature ?Ni 3. it l; .?mm eh.th 286': 0 10 2030405060 703090100110120130 Model?t'enrs planer. Bet modelers havehad a tough time incorporating Drift-free. The NOAH model. which suggests that Earth mil wann moderately them into their simulations because they are too small to (red). can reliably simitlato presentclimate {bate}. slmuponthe doesn?t have a ?ner int-sh1 but irdoes includeanew 'parameterization" that passes the effects of these imseen eddies onto scales. usinga more realistic means ofrniring heat dinigh the ocean than any earlier model did. says Boyille. Even when run for 300 model ?years.? the CSM doesn?t drift any from 'a reasonably realistic climate. says Climate and Global Dynamics director Maurice Blackmon. I?Being able to do this without ?ux corrections gives you more credibility." he says. ?For better or worse. we're not biasing the results as was meessary before-" The ?rst results from this still vastly simplified model imply that future greenhouse warming may be milder than some other models have suggested?and could take decades to reveal itself. Doubling atmospheric carbon dioxide concentrations in the model raised the global temperature 2 degrees Celsius. which puts can be explained as natural Variation.? says Blackmon. That would make the detection of a modest-size greenhouse warming all the more difficult. The CSM is available on the Internet. but Blackmon warns that if you want to check out ?ltul'c climate scenarios. you'll I"titted the biggest supercomputer you (an get.? indeed. even NCAR researchers haven't been able to experiment with the model on as large a computer as they would like. "While their purchase of an NBC 3X4 computer is tied up in a trade dispute with Japan (Science. 30 Augusr 1996. p. 11TH. they are making do with a leased Cray (3-90 with perhaps 10% of the speed of the 5X4. That worries some modelers. Americans have ?been among the lead? ers of the ?eld from the beginning." says Randall. but ?if we can?t get acute to the most: powct?Jl machines. we are going to be left behind.? :ading climate model?developed at the -ritish Meteorological Office's Hadley Cam :1 for Climate Prediction and Research. in -racl