Texas RRC- Railroad Commission Hires Seismologist Page 1 of2 Railroad Commission Hires Seismologist 03/2812014 AUSTIN - Railroad Commission Executive Director Milton Rister announced today the Commission has hired a seismologist. Dr. David Craig Pearson, a former team leader for a Los Alamos National Laboratory seismic, experimental field team, who also holds a doctorate in geophysics from Southern Methodist University. "My objective is to develop a broad understanding of the impact of oil and gas extraction activities on the dayto-day lives of Texas residents," Pearson said. "I believe the Railroad Commission must be able to quickly and factually determine the accurate location of all earthquakes in the state and be able to determine the cause of earthquakes, be they natural or man-made. "I plan to work to help clarify the root cause of earthquakes occurring by bringing all stakeholders' concerns, questions, ideas and insights together to identify the best possible solutions," Pearson said. Chairman Barry Smitherman said, "I look forward to Dr. Pearson's assistance in helping our agency become proactive in determining the exact cause of seismic events in Texas as very few and relatively minor seismic events have been documented over the past several decades compared to more than 144,000 disposal wells operating nationwide." Commissioner David Porter said, "After an extensive, nationwide search, I'm happy to announce that the Railroad Commission has added a highly qualified seismologist to our staff. This will allow our agency to further examine any possible correlation between seismic events and oil and gas activity and gain a more thorough understanding of the science and data available. Having a seismologist on staff will also enable the Commission to better coordinate with the academic community on future research." Commissioner Christi Craddick said, "The Commission has a long history of safely regulating disposal wells, with the Commission issuing its first disposal permit in 1936. With more than 34,000 of these wells currently operating in Texas, it is important that sound science be our guide in determining if there are any links to seismic activity. I welcome Dr. Pearson's expertise to help us work on these issues." Rister said Pearson will start effective April 1. His bio can be found at: http://www.rrc.state.tx. us/media/22997/seismologist_resume. pdf At SMU, while a PhD graduate student, Pearson was a graduate research assistant to Dr. Brian Stump, an SMU professor who is a member of the SMU earthquake team currently studying north Texas seismicity. Pearson has experience with many types of man-made earthquakes from nuclear explosions, coal mining blasts and controlled seismic sources for exploration geophysics. And he also has participated as an author on numerous academic papers on these subjects. Pearson is a member of the Seismological Society of America, American Geophysical Union and American Association of Petroleum Geologists. In his former position as a ranch manager in McCamey, Pearson has been a strong advocate of protecting and conserving Texas groundwater and surface water, and he currently serves as vice president of the Upton County Water District. http://www.rrc.state.tx.us/news/032814/ 8/19/2014 ·1·exas !{l{C- Railroad Commission Hires Seismologist Page 2 of2 As the Commission's in-house seismologist, Pearson will allow the Commission to strengthen its ability to follow new research, as well as coordinate an exchange of factual, scientific information with the research community. Pearson's duties will include coordinating with other academic experts studying seismic events in Texas; obtaining, studying and interpreting various forms of data to evaluate seismic activity associated with known faults and historic and/or ongoing oil and gas exploration and production activities; leading efforts to conduct research as well as internally integrate oil and gas science with seismic science; coordinating communications and information gathering with stakeholders; reviewing, analyzing, interpreting and commenting on technical data from seismic data sources, computer models and digital maps; and developing recommendations and action plans. ### About the Railroad Commission Established in 1891, the Railroad Commission of Texas is the oldest regulatory agency in the state. The Commission has a long and proud history of service to both Texas and to the nation, including more than 90 years regulating the oil and gas industry. Additionally, the Commission promotes research and education on the use of alternative fuels and has jurisdiction over gas utility, surface mining and pipeline industries. Our mission is to serve Texas by our stewardship of natural resources and the environment, our concern for personal and community safety, and our support of enhanced development and economic vitality for the benefit of Texans. To learn more, please visit http://www.rrc.state.tx.usl. Barry T. Smitherman Chairman Christi Craddick Commissioner David Porter Commissioner Contact Ramona Nye P; 512-463-4817 E: http://www.rrc.state.tx.us/news/032814/ 8/1 9/2014 NEWS RELEASE: Railroad Commission To Hire Seismologist- January 7, 2014 Contact Us Page 1 of 1 LcQ In FAQ's Links Home > Press Releases :. 20 14 .,. NEWS RELEASE : Railroad Commission To Hire Seismologist- January 7, 2014 RAILROAD COMMISSION OF TEXAS NEWS RELEASE ·JANUARY 7, 2014 BARRY T. SMITHERMAN Chairman DAVID PORTER Commissioner CHRISTl CRADOICK Commissioner Contact Ramona Nye 512- 2014 > NEWS RELEASE : Ra il road Commission Hires Seismolog ist· March 28, 20 14 RAILROAD COMMISSION OF TEXAS NEWS RELEASE -MARCH 28, 2014 BARRY T . SMITHERMAN DAVID PORTER CHRISTl CRADDICK Chairman Commissioner Commissioner Contact: Ramona Nye 512-463-4817 Railroad Commission Hires Seismologist AUSTIN -Railroad Commission Executive Director Milton Rister announced today the Commission has hired a seismologist, Or. David Craig Pearson, a former team leader for a Los Alamos National Laboratory seismic, experimental field team, who also holds a doctorate In geophysics from Southern Methodist University. "My objective is to develop a broad understanding of the impact of oil and gas extraction activities on the day-to-day lives of Texas residents," Pearson said. "I believe the Railroad Commission must be able to quickly and factually determine the accurate location of all earthquakes in the state and be able to determine the cause of earthquakes, be they natural or man-made. "I plan to work to help clarify the root cause of earthquakes occurring by bringing all stakeholders' concerns, questions, Ideas and Insights together to identify the best possible solutions," Pearson said. Chairman Barry Smitherman said, "I look forward to Dr. Pearson's assistance In helping our agency become proactive in determining the exact cause of seismic events in Texas as very few and relatively minor seismic events have been documented over the past several decades compared to more than 144,000 disposal wells operating nationwide." Commissioner David Porter said, "After an extensive, nationwide search, I'm happy to announce that the Railroad Commission has added a highly qualified seismologist to our staff. This will allow our agency to further examine any possible correlation between seismic events and oil and gas activity and gain a more thorough understanding of the science and data available. Having a seismologist on staff will also enable the Commission to better coordinate wijh the academic community on future research.w Commissioner Christi Craddick said, "The Commission has a long history of safely regulating disposal wells, with the Commission Issuing Its first disposal permit In 1936. With more than 34,000 of these wells currently operating In Texas, it Is important that sound science be our guide In determining if there are any links to seismic activity. I welcome Or. Pearson's expertise to help us work on these issues." Rister said Pearson will start affective April 1. His blo can be found at: www. rrc. state .tx.us/press releases/20 14/Seismologist Resume.pdf At SMU, while a PhD graduate student, Pearson was a graduate research assistant to Dr. Brian Stump, an SMU professor who Is a member of the SMU earthquake team currently studying north Texas seismicity. Pearson has experience with many types of man-made earthquakes from nuclear explosions, coal mining blasts and controlled seismic sources for exploration geophysics. And he also has participated as an author on numerous academic papers on these subjects. Pearson is a member of the Seismological Society of America, American Geophysical Union and American Association of Petroleum Geologists. In his former position as a ranch manager In McCamey, Pearson has been a strong advocate of protecting and conserving Texas groundwater and surface water, and he currently serves as vice president of the Upton County water District. As the Commission's in-house seismologist, Pearson will allow the Commission to strengthen its ability to follow new research, as well as coordinate an exchange of factual, scientific information with the research community. Pearson's duties will include coordinating with other academic experts studying seismic events In Texas; obtaining, http://www.rrc.state.tx.us/pressreleases/20 14/032814. php 4/24/2014 NEWS RELEASE: Railroad Commission Hires Seismologist- March 28, 2014 Page 2 of2 studying and Interpreting various forms of data to evaluate seismic activity associated with known faults and historic and/or ongoing oil and gas exploration and production activities; leading efforts to conduct research as well as Internally integrate oil and gas science with seismic science; coordinating communications and information gathering with stakeholders; reviewing, analyzing, Interpreting and commenting on technical data from seismic data sources, computer models and digital maps; and developing recommendations and action plans. ### About the Railroad Commission Established in 1891 , the Railroad Commission of Texas is the oldest regulatory agency In the state. The Commission has a long and proud history of service to both Texas and to the nation, Including more than 90 years regulating the oil and gas industry. Additionally, the Commission promotes research and education on the use of alternative fuels and has jurisdiction over gas utility, surface mining and pipeline Industries. Our mission is to serve Texas by our stewardship of natural resources and the environment, our concern for personal and community safety, and our support of enhanced development and economic vitality for the benefit of Texans. To learn more, please visit http://www.rrc.state.tx.us/. Advanced Search ICompact with Texans IOpen Records ITexas Homeland Secunty I Texas Veterans Portal ITRAIL I Search ITexas Online IReporting Fraud, Waste & Abuse IRRC Expenditures-Where the Money Goes ISite Policies I Site Map IJobs I http://www.rrc.state. tx.us/pressreleases/20 14/0328 14.php 4/24/20 14 Available Jobs at the Railroad Commission Page 1 of3 I Sun:h I• Contacr Us c::;J • Jo b o > Ava;lable Jobs ;>1 l.29.l.!! f!Q:I. .b..!n!! tho Railroad Comm•n•on Available RRC Jobs .~....;::....r, NEOGOVJob Code:201<1 ·002!58 Job l1Ue: Selamalogltt Clotlng Dati/Time:Conllnuoua Selery:$!5, 1!5!5.58 • $8,2<18.00 Monlhly Job ~ptt: FIAI-Tlme LoceUon: 1701 N Congreu Aw. AusUn. Texaa Print .Job ln!ormaUon 1!.R.IIb: Job Deecriptlon II Benefttl II Supplemental Questlone I The Railroad Commission of Texas seeks a seismologist capable of utilizing United States Geological Services (USGS), National Earthquake lnformalion Center (NEIC), and other sources of seismic information to evaluate the projected epicenters of seismic events In Texas relative to all potential sources of origin. JOB DUTIES: Under the direction of the Executive Director, the Seismologist will provide leadership, direction, technical expertise and oversight to Include: Coordinating with academic experts studying seismic events In Texas . Obtaining, studying, and Interpreting various forms of data to evaluate seismic activity associated with known and historic and/or ongoing oil and gas exploration and production activities. ra Leading efforts to conduct research as well as intemally integrate oil and gas sclance with seismic science. Coordinating communications and information gathering with various stakeholders. Reviewing, analyzing, interpreting, and commenting on technical data from seismic data sources, computer rrjo and digital maps. Developing calculations, estimates, recommendations and action plans. Presenting information In layman's terms to an audience through presentations and In writing to the public, to Industry, to Commissioners, Commission staff, and the Legislature. Maintaining relationships with internal and extemal customers, Attending meetings In the scientific community to leam and gather relevant Information. Planning, assigning and supervising the work of others in a work group. Performing additional duties as assigned. Performing work with considerable latitude for the use of Initiative and Independent judgment. State Cla11lflceUon II: 2366 Stale Cluelflc:atlon Title: GEOSCIENTIST V Selary Group: 825 Minimum Quellflcellona: http://agency .govemmentjobs.com/texasrrc/default.cfm?action=viewJob&jobJD=788706&. .. 3/18/2014 Available Jobs at the Railroad Commission - · ·~- Page 2 of3 ............ . A minimum of five (5) years of progressively responsible experience in positions Involving the study of selsm c activity. Demonstrated experience In the analysis and evaluation of technical data. Demonstrated experience In a supervisory or management role. Demonstrated experience with Input ground motions for seismic design preferred. Demonstrated experience In digital signal processing preferred. Knowledge, Skill, and Abilities: Knowledge of: (an understanding of facts or principles relating to a particular subject area) Ability to demonstrate knowledge of geosciences principles, techniques, and procedures; of mathematics an statistics; and of the practical application of geosciences and technology. Knowledge of seismic data acquisition methods and systems. Knowledge of the principles and contemporary practices of public administration and management. Knowledge of oil and gas regulatory functions is helpful. Skill in: (the application of knowledge resulting from a development of basic abilities through tonnal training and practical experience) Skill in data collection, analysis and management of geological, geohydrologlcal, and geophysical data. Pronclent skill In performing statistical analyses. SkiR In applying modeling and statistical procedures. SkiQ In exercising logic and soood reasoning to ldenUfy the strengths and weaknesses of alternative solutions conclusions, or approaches to problems. Skillin forming and maintaining positive and productive working relationships both Internally and external to th ~ Commission Including with other state/public agencies, Internal work teams, and state/federal oversight agenc e Skillin operating telephones, computers, and applicable computer software, Including PETRA, Microsoft Word, Excel, PowerPolnt, Access and Outlook. Skillin handling multiple prlorttles efficiently. Skillin oral and written communication with the ability to present material to an audience. Skillin demonstrating attention to detail and accuracy of Information. Ability to: (capacity In a general area that may be utilized to develop detailed, spectflc skills) Ability to Jearn, understand, and Interpret oil and gas policies and procedure. Ability to conduct Inspections. Ability to apply geological concepts. Ability to establish and maintain trust and credlbiUty with management, colleagues and direct reports. Ability to prepare reports, memos, calculations and other documentation. Ability to communicate effectively verbally and In writing. Ability to maintain confidentiality of Information. Ability to wortt as a member of a team. Ability to maintain a professional demeanor In any situation. Ability to reach high and low; open, close and retrieve flies from file cabinets; and lift and/or carry equipment l p 25 pounds. Btgjtlrltlon. Ctrt!Oc!llpn, pr LlctDIY!Ji Pro!elllonal reglalniUon 11 10 11ngln.,r, gealogl&l, g110phyllclat, or clo111ly 111illtd neld pwltmld. REMARKS (Application procedure, Special requirements): up to 20%. Salary Is up to $8248/month. Trc~vells One (1) vacancy exists for this position. THE RAILROAD COMMISSION OF TEXAS ONLY ACCEPTS ONLINE APPLICATIONS FOR THIS POSTING. Applications, for this posting. should be submitted on-Una by clicking on thi'Apply' link abova. Oue to the high volume of applications we do not accept telephone calls. Only candidates selected for Interview will be contacted. http://agency .govemmentjobs.com/texasrrc/default.cfm?action=viewJob&jobiD=788706&... 3/18/2014 Avail able Jobs at the Railroad Commission AdVJnec p-o~s:. 'l..e... k ~~ ld-- 1 We.. ~sw.~~-et( 0 0 David Craig Pearson, Ph. D. P.O. Box 488 Crane, Texas 79731 Education Bachelor of Science in Geology, University of Texas of the Permian Basin, Odessa, Texas Master of Science in Exploration Geophysics, Southern Methodist University, Dallas, Texas Doctor of Philosophy in Geophysics, Southern Methodist University, Dallas, Texas Work Experience Ranch Manager Upton, Crockett and Reagan Counties, Texas · January 2006 to present Technical Staff, Department of Defense Program Office Los Alamos National Laboratory, Los Alamos, N.M. June 2004 to June 2006 Acting Deputy Division Leader, Earth and Environmental Sciences Division Los Alamos National Laboratory, Los Alamos, N.M. February 2002 to June 2002 and April2003 to June 2004 Deputy Group Leader, Geophysics Group, Earth and Environmental Sciences Division Los Alamos National Laboratory, Los Alamos, N.M. June 2000 to April 2003 Technical Staff Member, Geophysics Group Nuclear Explosion Monitoring Research and Engineering Program Earth and Environmental Sciences Division Los Alamos National Laboratory, Los Alamos, N.M. December 1999 to September 2001 Technical Staff Team Leader, Geophysics Group Earth and Environmental Sciences Division Los Alamos National Laboratory, Los Alamos, N.M. May 1993 to December 1999 Graduate Research Assistant, Geology Department Southern Methodist University, Dallas, Texas January 1989 to May 1993 1 0 0 Research Geophysicist, Energy Research and Service Department Teledyne Geotech, Garland, Texas September 1985 to January 1989 Summer Research Geophysicist ARCO Resource Technology, Plano, Texas May 1985 to August 1985 Special Operator, Hydraulic Frac Treater, Acid Treater, Cementer, Driver Halliburton Services, Big Spring, Texas May 1978 to June 1982 Professional Society Memberships American Geophysical Union Seismological Society of America American Association of Petroleum Geologists International Society of Explosives Engineers Organizations and Activities Vice President, Upton County Water District Published Works Stump, B.W. and D.C. Pearson, 1996. Design and Utilization of a Portable Seismic/Acoustic Calibration System, Conferences on Disarmament, Group of Scientific Experts, United States Delegation, GSE/US/114, May 1996. Stump, B.W., D.P. Anderson and D.C. Pearson, 1996. Physical Constraints on Mining Explosions, Synergy of Seismic and Video Data with Three Dimensional Models, Seismological Research Letters, 67, 9-24. Stump, B.W., D.C. Pearson and R. Reinke, 1999. Source Comparisons Between Nuclear and Chemical Explosions Detonated at Rainier Mesa, Nevada Test Site, Bulletin of the Seismological Society of America, 89, 409-422. Phillips, W.S., D.C. Pearson X. Yang and B.W. Stump, 1999. Aftershocks of an Explosively Induced Mine Collapse at White Pine, Michigan, Bulletin of the Seismological Society of America, 89, 1575-1590. Yang, X., B.W. Stump and D.C. Pearson, 1999. Moment Tensor Inversion of Single-hole Mining Cast Blast, Geophysical Journal International, 139, 679-690. Bonner, J.L., D.C. Person, W.S. Phillips and S.R. Taylor, 2001. Shallow Velocity Structure at the Shagan Test Site in Kazakhstan, Pure and Applied Geophysics, 158, 2017-2039. 2 . . 0 0 Hedlin, M.H., B.W. Stump, D.C. Pearson and X. Yang, 2002. Identification of Mining Blasts at Mid- to Far- Regional Distances using Low Frequency Seismic Signals, Pure and Applied Geophysics, v159, No.4, 831-864. Stump, B.W., M.H. Hedlin, D.C. Pearson and V. Hsu, 2002. Characteristics of Mining Explosions at Regional Distances, Reviews of Geophysics, 40(4), 1011, doi:10.1029/1998RG000048, 2002. Stump, B.W. and D.C. Pearson, 2002. Source Scaling of Single-Fired and Delay-Fired Explosions Constrained by In-Mine and Regional Seismograms, Proceedings of the TwentyEight Annual Conference on Explosives and Blasting Techniques, February 10-12, 2002, Las Vegas, NV, International Society of Explosives Engineers, Vol. 2, p. 243-252. Bonner, J.L., D.C. Pearson and S. Blomberg, 2003. Azimuthal Variation of Surface Wave Energy Form Cast Blasts in Northern Arizona, Bulletin of the Seismological Society of America, 93, 724-726. Stump, B.W., D.C. Pearson and V. Hsu, 2003. Source Scaling of Contained Chemical Explosions as Constrained by Regional Seismograms, Bulletin of the Seismological Society of America, 93, 1212-1225. Awards and Recognitions Salutatorian, McCamey High School, McCamey, Texas, 1976 Outstanding Geology Graduate, Southern Methodist University, 1983 Who's Who Among American Universities and Colleges, 1984 First Graduate of Exploration Geophysics degree program, Southern Methodist University, 1985 3 NEOGOV Insight - Applicatio?" Detail Page 1 of7 0 2014 -00258 - Seismologist Contact Information ~ ~n ID: 19026019 ~arson Name: Da 1d Craig Home Phone: Email: Former Last Name : (940) 765-4846 , nrnr t:att · n none Address: Alternate Phone: Notification Preference: Month and Day of Birth : P. 0 . Box 488 Crane, Texas 79731 (432) 693-2790 Email 07/26 US Personal Information Driver's License: Can you, after employment, submit proof of your legal right to work In the United States? What Is your highest level of education? Yes, Texas , 24491588 , Class C Yes Doctorate Preferences $52.88 per hour; $110,000.00 per year Yes None Regular Full Time , Part Time , Per Diem Day Preferred Salary: Are you willing to relocate? Types of positions you will accept: Types of work you wUI accept: Types of shifts you will accept: Objective My objective In seeking this job Is to develop a broad understanding of the Impact of oil and gas extraction activities on the day to day lives of the citizens of the State of Texas. I believe that the Railroad Commission must be able to quickly and factually be able to determine the accurate location of all earthquakes In the state and be able to ascribe the cause of the earthquake to likely processes, be they natural or man made. Education Graduate School Southern Methodist University 1/1989- 5/1993 Dallas, Texas Did you graduate: Yes College Major/Minor: Geophysics Units Completed: 30+ Semester Degree Received: Doctorate Graduate School Southern Methodist University 9/1984- 5/1985 Dallas, Texas Old you graduate: Yes College Major/Minor: Exploration Geophysics Units Completed: 30+ Semester Degree Received: Master's College University of Texas of the Permian Basin 9/1983- 12/1984 Odessa, Texas Did you graduate: Yes College Major/Minor: Geology Units Completed : 60+ Semester Degree Received: Bachelor's High School McCamey High School 9/1971- 5/1976 McCamey, Texas Old you graduate : Yes Degree Received: High School Diploma Work Experience Ranch Manager 1/2006 - Present Robert Eaves McCamey, Texas 79752 Hours worked per week: 60 Monthly Salary: $130,000.00 # of Employees Supervised : 3 Name of Supervisor: Robert Eaves - Owner May we contact this employer? No https://secure.neogov.com/OHC/view resume.cfrn?Print=Y&JobiD=788706&ResumeiD=... 1/ 16/2014 N~OGOV Insight - ApplicatioC)etail 0 Page 2 of7 Duties Responsible for all aspects of ranch management Including animal husbandry, Infrastructure maintenance and Improvement, purchasing and procurement of services and materials, recruiting and scheduling of regular and temporary employees, marketing of products, and negotiating contractual agreements with oil and gas exploration and development companies and their agents. Reason for Leaving Opportunity for a unique position at the Railroad Commission Deputy Division Leader, Technical Staff Member Hours worked per week: SO 5/1993 - 6/2006 Monthly Salary: $142,000.00 # of Employees Supervised: 300 Los Alamos National Laboratory Name of Supervisor: Paul Weber - Division Leader Los Alamos National laboratory, New Mexico May we contact this employer? No ( Duties May, 1993 to December, 1999 As team leader for the seismic, experimental field team In the Geophysics Group, I was responsible for planning, procurement, deployment and maintenance of state-of-the-art seismic data acquisition systems at, typically, harsh field locations throughout the world. Included In these experiments were 1) Nuclear and Chemical explosions at the Nevada Test Site, 2) enormous overburden removal explosions In coal mines In Wyoming and Indiana, 3) an underground copper mine engineered collapse In Michigan, 4) decommissioning of Inventory nuclear test holes at the Former Soviet Semlpalltinsk Test Site In Kazakhstan, and 5) military applications In the Korean Demilitarized Zone. I was also responsible for planning and scheduling these experiments and analysis and reporting of the resulting data . These experiments were conducted to support treaty verification and explosion source phenomenology research projects. The team I led to Kazakhstan was the first US team to record local seismic measurements from explosions at the Balapan and Degelan testing complexes and resulted In validation and refinement of the local velocity structure determined using historical teleseismic observations. December, 1999 to September, 2001 As lead project leader for the Ground-Based Nuclear Explosion Monitoring Research and Engineering Program, I was responsible for programmatic definition and development, project direction and coordination with Department of Energy Headquarters (DOE/HQ), Lawrence Livermore National Laboratory (LLNL), Sandia National Laboratory (SNL), Pacific Northwest National Laboratory (PNNL) and the Air Force Technical Applications Center (AFTAC). I was required to negotiate with numerous Los Alamos National Laboratory line managers for Internal LANL staffing and budget allocations to assure timely deliverables to the DOE/NNSA NA-21 Knowledge Base for use by AFTAC. I directed and participated In development of seismic data analysis techniques, as well as analysis and evaluation of technical results from numerous academic and private contractors for consideration of Inclusion In the DOE/NNSA NA-21 Knowledge Base. I actively participated In Inter-laboratory and Interagency evaluation of current and future research and analysis needs for the AFTAC United States Atomic Energy Detection System (USAEDS). June, 2000 to April, 2003 As Deputy Group Leader and Acting Deputy Group Leader for the Geophysics Group In the Earth and Environmental Sciences Division of Los Alamos National Laboratory, I was responsible for management of personnel and programs In a group composed of more than 35 Ph.D scientists, technical and administrative support staff. I was also responsible for providing technical leadership and publication of research results for data analysis techniques, experimental observations and computer simulations and modeling of earth structures. I was responsible for continuous oversight of Group Integrated Safety Management, Integrated Security and Safeguards Management, Environmental Safety and Health Management, salary management and performance appraisals for all Geophysics group personnel. February, 2002 to June, 2002 and April, 2003 to June, 2004 As Acting Deputy Division Leader for the Earth and Environmental Sciences Division of Los Alamos National Laboratory, I was responsible for management of personnel and programs In a 300+ person division composed of more than 130 Ph.D scientists, technical and administrative support staff. I managed division wide research and development programs In National Security and Defense, Computational Geosciences, and Nuclear Materials Repository Science with an operating budget In excess of $65,000,000 In FY03. I led Laboratory efforts In business development for C02 sequestration research leading to funds In of In excess of $2.5 million. I participated [n high-level Laboratory Strategic Planning exercises, program development activities and interfaced regularly with sponsors and customers to ensure programmatic excellence and timely delivery of research and development products. I was responsible for continuous oversight of Division Integrated Safety Management, Integrated Security and Safeguards Management, Environmental Safety and Health Management, salary management and performance appraisals for all EES Division personnel. June, 2004 to June, 2006 As a Department of Defense Program Development Team Member, I was responsible for Laboratory wide business development with Department of Defense customers Including Future Combat Systems, lED https://secure.neogov .corn/OHC/view_resume.cfm?Print=Y&JobiD=788706&ResumeiD=... 1116/2014 NEOGOV Insight- Applicatiop Oetail 0 Page 3 of7 Detection and Mitigation in the War on Terror, and Force Protection Technologies. I directed internal business development funding to strategically position Los Alamos capabilities to address emerging DOD requirements. I represented the DOD Program Office to upper Laboratory management strategic planning and Internal funds return on Investment. Reason for Leaving Return to family ranching business Graduate Research Assistant, Ph.D Graduate Student 1/1989- 5/1993 Southern Methodist University Dallas, Texas (214) 768-1223 Hours worked per week: 60 Monthly Salary: $2,000.00 # of Employees Supervised: 0 Name of Supervisor: Dr. Brian Stump- Professor May we contact this employer? Yes Duties My research focused on development of 1) data acquisition Instrumentation and techniques, 2) design and characterizatl.on of seismic sources for generation of the full seismic wave-field, and 3) utilization of the full seismic wave-field for multiple, constrained inversions of the observed data for determination of shallow (=filter analysis (Dziewonski et aL, 1969) and phase match filtering (Herrin and Goforth, 1977) to extract Rg. Both of these programs are packaged in software developed by HerrmaM (2001). The two cast blasts generated Rg dominated by 4-sec period energy as shown in seismograms (Fig. 7) recorded at the Pinedale, Wyoming array (PDAR). The fact that Rg is recorded at PDAR. which is approximately 700 km from the coal mine, suggests that the northern Colorado Plateau is IUl efficient propagation medium for SP seismic-wave energy. The Rg has larger signal-to-noise ratio (SNR) than both Pn and Lg rendering it important in discriminating this event. The 4-sec period corresponds to the approximate total duration of the blast plus the spalled material impact with the pit floor as measured from the videographic data. Examples of the amplitude analysis for the 4sec Rg waves at the seismic stations WUAZ (Arizona), KNB (UWt), PDAR (Wyoming), PV09 (Colorado), RW3 (Colorado), and TUC (Arizona) are shown in Figure 8. Figure 9 shows the amplitude ratio of the 24 March J999 cast to the 25 March 1999 event as a function of source-to-station az- Figure 5. Bandpass filtered (O.S to I Hz) local seismic recordings from lhe 24 M:lrch 1999 cast blast. imuth for all 14 stations. As an example, the Rg amplitude from the 24 March blast was a factor of 2 greater than the 25 March Rg ns recorded at KNB (Fig. 8). TheRg amplitudes recorded at PDAR are approximately the same for both blasts, while at RW3, the amplitudes for the 25 March event are larger thllll for the first blast. These differences suggest that the variation in the shape of radiation patterns for the two blasts are functions of the orientation of the explosions. If the Rg amplitude variations were solely caused by differences in the yields for the mining explosions, even though the blasts were designed with similar explosive tonnage. similar ratios would be expected at each station. Also, the radiation pattern observed for the Rayleigh waves was not noted for the regional P and Lg phases. We observed that the P-phase and Lg-phase amplitudes scaled similarly for both blasts and were on average 1.3 times larger for the 24 March blast than for the 25 March event. This is consistent with the USGS magnitudes (MJ of 2.7 (24 March 1999) and 2.6 (25 March 1999) as listed on the USGS mining seismicity web page. We note that there were not enough three-component data for thorough examination of Love-wave radiation phenomena. Modeling We completed a detailed modeling study of these cast blasts to investigate the source of the azimuthal variation of Rg. The modeling was accomplished using MineSeis (Yang, 1998), a Matlab program based upon the explosion model created by Anandakrishnan et aL ( 1997). This model assumes the principle of linear superposition (Stump and Reinke, 1988) of multiple subsurface explosions that are modeled as spherical isotropic point sources (Mueller and Murphy, 1971). Spall is modeled as a set of vertical and Azimuthal VarlaJ/on of Short-Period Rayleigh Waves from Cast Blasts in Northern ArirPnll 729 0 0 4.0 _, _ • Figure 6. Cast blast mosaic for the 1A MlltCh 1999 cast blast. In the upper left pbotograph, tbe small nrrow points 10 the location of the coal seam being uncovered by the cast blast. The spall begins to impact the ground between 2 and 2.5 sec after the lniti:ll explosion In tbe cast sequence. 0 horizont41 forces as described by Barker et a/. (1993) and modified by Yang (1998). The single shots in the multishot blast model are modeled as point sources with second· order source moment tensor representations for both the explosion and spall. For a more detailed explanation of MineSels, refer to Yang (1998). We incorporated P- and S-wave velocities measured from linear geophone profiles near the source together with videographic images to accurately represent the source and spall parameters (Table 2). The estimate of the spalled mass per single-point explosion is one of the largest sources of uncertainty for the model. We used the powder factor for the explosions, listed in Table 2 as 0.65 kg of explosives for every cubic meter of spaJied material to determine the spaJied mass. The powder factor is determined from blasting experience at the mine and the explosive performance and should be considered an average value over the entire scale of the blast. Localized perturbations in material properties can change this value by an order of magnitude or greater, often resulting in inefficient use of the explosives. For this modeling study, we assume full coupling of these spall sources into the radiated seismic wavefield, although the results presented in the subsequent section suggest that complicated nonlinear effects may be altering this coupling factor. Single-point seismograms were generated using these parameters and the MineSeis models for an explosion point source and spall. The seismograms were then convolved with canonical Green' s functions generated by the reflectivity method (Muller, 1985) for aID Colorado Plateau crustal structure (Keller et aL, 1976; Prodehl and Lipman, 1989). The single-shot seismograms and resulting spectra modeled for the 24 March 1999 cast blast as recorded at the epicentral 730 J. L Bonner, D. C. Pearson, and W. S. Blomberg 0 procedure discussed above to model the 25 March 1999 cast blast: however, we increased the number of holes in the shooting design from 440 to 450 to reflect the increase in the total yield of the second cast blast as a result of 10 ad· ditional blast boreholes. Finally, we used a face azimuth of 120° for the second cast blast; however, we note that the blast is also curvilinear about the southern pit wall (Fig. 3). Figure 13 illustrates the theoretical surface-wave amplitudes (nm/sec) as a function of azimuth for the two cast blasts. Discussion ~ -2 ~.~~ .~~~~~~~~~~~~~~~~~ I e<1 Figure 7. 11mc(.........U) Unfiltered (BHZ) and pluise-llUitcb fil- tered (PMF) seismogrnms (top • 24 Man:b 1999, bonom "" 25 March 1999) from tbe northern Arizona CllSI blasts as recorded at the Pinedale, Wyoming (I'DAR) IIJ'r.ly. Rayleigh waves with periods between 2.5 and S sec nre observed at PDAR with group velocities of approxillUitely 2.1 kmlsec. The surface waves have larger amplitudes tban both Lg and an emergent Pn. 0 0 distance and azimuth for WUAZ are shown in Figure 10. Linear superposition was then used to represent the 440 individual explosions for this event. and the impulse superposition time function and spectra are displayed in Figure 11. The overall effect of the linear superposition, as seen by the spectra of the impulse time signal. is to effectively lowpass filter the single-shot data as a result of the 2-sec duration of the blast. For the 24 March 1999 event, we used a face azimuth, the bearing from north that parallels the topographic bench being blasted, of 35°. As noted in Figure 3. the blast progresses along a face azimuth of approximately 35° for approximately 05 km and then bends to a face azimuth of 35SO for the remainder of the blast that is less than 0.3 km. As it exists now. MineSeis assumes that the shot has a simple, rectangular shape with parallel straight rows, equal burden for all shots, and equal spacing between shots. More complex shot patterns, such as chevron-shaped patterns similar to the 24 March 1999 design, are planned for future releases of the code. We then convolved the single-shot seismograms with comb functions that represent the superposition time function of the individual explosions to represent the seismogram at regional clistances. Figure 12 shows the results of using this model to synthesize seismograms from the 24 March 1999 cast blast as recorded at WUAZ. The data were filtered between a 2- and 5-sec period to accentuate the dominant Rg period. Because of the uncertainties associated with the spat! estimate, it is difficult to compare absolute amplitudes for the resulting synthetics: thus we have normalized to the maximum amplitude of the fundamental mode Rayleigh waves. The model effectively predicts a significant segment of the Rg-wave packet. We completed the same The radiation patterns for the modeled cast blasts are not isotropic as shown in Figure 13. The theoretical radiation patterns show larger amplitude lobes perpendicular to the bench, with a small amplitude enhancement noted for the clirection of the delay firing (the arrows shown in the center of the plot). The cause of the amplitude enhancement perpenclicular to the bench Is the horizontal spall force. When the same blasts were modeled with no horizontal spall (or an eject angle of 0° relative to vertical), the racliation patterns were almost isotropic with the exception of a I%-3% amplitude increase in the direction of the delay firing. Thus, the overall effect of the horizontal spall is to create amplitudes as much as approximately 25 times greater normal to the bench than paraJiel to the free face. We compare the theoretical radiation patterns for the cast blast models with the observed data in Figure 14. The peanut-shaped dipole is the ratio of the 24 March 1999 Rg amplitude radiation pattern to the 25 March 1999 pattern. both of which are presented in Figure 13. The small circles represent the 24 March/25 March Rg ratios calculated at 14 regional stations that recorded surface waves from the event. The observed data are plotted in Figure 9. In Figure 14 we plot the data in polar form to compare with the racliation patterns and the amplitude ratios for both observed and preclicted data are tabulated in Table 3 under the column for model BMl. For 12 of the 14 stations, there are similarities in both the pattern and magnitude of the observed and theoretical Rg ratios for the two cast blasts. To quantify the differences between the observed and preclicted data, we define the root mean square (rms) error, 4\m.: (1) where tPN """ Rpn: - Robs is the clifference between the ratio of the 24 March 1999 Rg amplitudes to the 25 March amplitudes as predicted from the MineSeis modeling (Rprc) and the ratio of the surface-wave amplitudes observed (Rob.) for the same blasts at the N = 14 seismic stations. The magnitude of tPN describes how far the observed amplitude ratio for Rg deviates from the predicted amplitude ratio. The sign provides information as to whether the observed data has a AzimuJhal Varimlon of Shon-Period Rayleigh Waves from Cast Blost.r In Northern Arlwna 731 0 WUAZ KNB AZ218° DIST 142 km 2 -Man:h24 ••• Mard\25 =t=u~ 3 3!.-----~..-------:!:-z----~,.5 . AZ285° DIST224km ______-71.5 3 3r-------~2~~------*2 GRiup Velocity (km/NC) GIDUp Velocly (kmiHC) PV09 PDAR ! 1' .. s.-~---.. f 0 -March24 ••• Man:h25 AZ5° DIST697km l =tt=~ AZ26° DIST247km 3 ~3--------~275--------~2-------~,.5 GRiup Velocly (lcnVuc:) TUC RW3 -March24 ••• March 25 3 ~-------~2.~-----~2-------~$ March24 ••• Marcll25 AZ 184° DIST468km GIDUp Velocity (kmiNC) Figure 8. Regional short-period surface waves (2.5- to 5-sec period) recorded at WUAZ, KNB, PDAR, PV08, RWJ, and TIJC for botb cnst blnsts (solid ..,. 24 March 1999 and dashed = 2S Mllrcb 1999) of similar yield (- 726,000 kg of explosives). The amplitudes were normalized to tbe maximum surfoce-wave amplitude of tbe 24 Mardi 1999 event as recorded at eocb station. Each tmce was pbase-oullch filtered and plotted as a function of group velocity. The source to station llZimutb and distance are also sbown. 0 732 0 J. L Bonner, D. C. Pe411'50n, and W. S. Blomberg 4 Single Shot Selamograma 3 ~ r J IITU 0 2 (\j WUAZ lJ 0 t 0.9 ~ r·8 0 .7 ~0.6 < :m ~0 AW.I 0 0.5 0 so TOO TSO 200 250 300 So=e to Slalian AzimU1h ( - - ) 350 Figure 9. Ratio of the Rg nmplitudes for the 24 March 1999 cast blast to the 25 Mllrch 1999 event as 11 function of the azimuth from the coal mine to nll stations for which surface-wave data could be o~ tained. Table 2 Source Parameters Used to Model the Northern Arizona Cast Blasts 10 0 Explosive depth 30m BunSen Face AZimuth P·w~~ove velocity Ill dcplh Sm Sp!Jl impact pulse width Spal.l inililllion pulse width Spall yield Yield per hole Eject angle Vo:rtial falling dislance Spacing between rows Receiver distance Receiver AZimuth Inlenhoc delay Intcrrow dday Naunber of rows Nwnbc:r of holes per row 35 (24 Mardi) 120 (2S M1111:h) 7.9 kmfsec F._ 111 (Itt) Figure 10. Single-shot seismograms (top) and spectra (bottom) produced from modeling of 24 March 1999 cast blast at a distnnce and azimuth for WUAZ. 2sec: I sec 2.84 ktJhole 1626 Jcs 7(1' 30m 10m 142km 0:15:360 17 mscc 100, ISO, 200, 200 msec; s 400 (24 March) 450 (2S M1111:h) larger amplitude ratio (negative) or smaller ratio (positive) than the predicted data. The value of q,N can range from zero, a perfect fit, or to ±infinity. which would correspond to no correlation between the data and the predicted models. The rms error, q,rms, was determined for the observed and predicted data in Figure 14 as 0.48. We tabulated the individual q,N and ,p,_ values in Table 3 (BMl). To compare this q,mu value with other possible radiation patterns, we chose to rotate the predicted radiation for the MineSeis model by 90°, which would correspond to the largest surface wave amplitudes from the cast blast being predicted to occur parallel to the free face and the smallest being in the direction normal to the face. The resulting rms error was 1.27. Given that this is not a physically realistic model. we would hope that the q,.,. value would be larger with respect to the value of 0.48 determined for the plausible model. If the radiation patterns predicted from MineSeis had been isottopic, the resulting q,rms value would have been 0.72, which is larger than the observed value of 0.48. However. it is not as large as the hypotheticnJ model discussed above where the radiation patterns were rotated opposite to what MineSels predicts for these two blasts. A large amount of the misfit between the observed and predicted data is related to the stations WUAZ and ANMO. The nns error excluding both of these two stations was reduced from 0.48 to 0.31. Our initial concern was that the measured surface waves from these stations were in error. We phase-match filtered both of these station's data two additional times and found the same amplitude ratios. We also obtained the same amplitudes through bandpass filtering and conclude that the amplitude ratios at both of these stations are not in error, but are in fact being caused by physical phenomena. These physical processes must be related to the cast blast source but are not modeled as part of the Anandakrishnan et aL (1996) model. We note that the best correlation of the amplitude ratios exist for stations between azimuths of (f' to 30° Azimuthal Variation of Short-Period Rayleigh Waves from Cast Blarrs in Northern Arlwna 0 lmpuiM Supetpcdlon 733 - - 24 Mardl1999 - - 25 MIU'dll999 N E 0 0.5 I nne (seq 1.5 3 SpedNm d lmpuiM Sllf*PO*iljon s Figure J3. Theoretical Rg·rndialion p111tems for the 24 M:lrcb 1999 (solid) and 25 March 1999 (dashed) cost blasts. - Ratio 24 Mardl/25 March N 0 Figure 11. Impulse superposition time fwtcdon (top) and spectr.l (bottom) for the 24 March 1999 cast blast os modeled for stndon WUAZ. s StatlonWUAZ Figure 14. Ratio of the 24 March and 25 Mnrch 1999 model generated Rg (solid line) together with observed ratios of the Rg phases lit 14 seismic stations at regional distances to the northern Arizona coal mine. Time (sec) from Cast Origin 0 Figure 12. Observed (top) vs. synthetic (bonom) cost blast seismogrnms os recorded and predicted at WUAZ for the 24 March 1999 event filtered to higb· light the 4-sec Rg. and from 300'" to 360°. These azimuths are generally behind the topographic bench of the pit, while WUAZ and ANMO were at azimuths in front of the bench for both blasts. We believe this may be the cause of large discrepancies between the observed and predicted Rg ratios. A similar effect was noted for SP Rayleigh waves from quarry blasts in central Texas. Bonner et a/. ( 1996) found that relative energy in Rg extracted from regional and local 734 J. L. Bonner, D. C. Pemon, and W. S. Blombesg 0 Table 3 Ratio of the Observed and Predicled Rg Ratios for the U-25 March 1999 Cllst Blasts• Slabua Ollocn-cd R1 Rallo Model R1 Rallo Rfn4-R• Moold Rt Rlllo Rpn:d-R.._ Modelllt Rlllo RpM-RoN Raf2S Man:llllJ BMI BMI BMl 8M2 DI.Cl ow 0.85 0.66 0.69 O.S3 1.2S 1.09 134 2.05 2.50 0.70 0.42 0.39 :UM&R:h PDAR PV09 PV08 RW3 ANMO rue WUAZ KNB EKU BHU 2.56 1.89 MMV 1.05 1.33 1.20 oro OWUT SRU o..ro 230 0.72 0.38 1.94 2.19 2.44 2.41 1.68 1.42 0.94 nns Error - - 0. 15 0.24 0.30 0.13 1.05 0.38 0.96 0.11 0.31 0. 11 0.51 0.63 0.09 0.26 0.48 0.60 0.43 0.36 0.37 1.17 0.77 0.64 3.78 436 4.87 3.45 1.80 I.S6 0.95 - 0.25 0.23 033 0.16 0.08 0.32 0.70 1.73 1.86 2.31 1.S5 0.75 0.23 - 0.25 1.06 0.76 0.46 0.39 0.47 1.2S 0.96 0.84 2.03 2.22 2.48 236 1.62 138 0.92 - 0.09 0.20 0.30 0.06 0.00 - 0.13 - o.so - 0.02 - 0.28 - 0.07 o.47 O.S1 0.06 - 0.29 0.28 • Also shown are &ru: di!Tueaces betwcca the obscm:d aad pn:dldcd data togelho:r wilh the rms error, L.TR-96-2112. KocaoaJu, A~ and L. T. Lons (1993). Tomogr:aphlc inversion ofRg wave srour velocities fot n:aiou!IIC•-surfACe SUUcturc, 1. ~op~. R~s. 91, 6579-6587. l..cilh. w~ A. Spivak, ead L. l'mlick (1996). J...:qc rnillias blasts from the Kunk miain; 11!gion. Rus;ia, Proccedinas of the 181h AnnuAl Seismic R~ Symposium. 12-15 Sc:ptcmbcr 1996, PlrlR·%-2153, Phil· Iii'$ I.Aboratoty, Hanscom AFB, Massachusetts. Mudler, R. A., ead J. R. Mwphy (1971 ). Seismic cluncleristica of underground nuclear dcl.oii.Uona, Bull. Seinn. Soc. Am. 61, 1675-1692. MuUCT, G. ( 1985). The reflectivity melhod: a tutorial./. Gtoplrys. sa, 153174. Pearson, D. C., D. W. Scump. and R. L. Martin (1997). Azimlltbal varia&ion of radiati011 of seismic c:ocrgy from cast bluts. Proc. Urrl AMUal Confermce on Explosive~ 1111d Bltuting T~chniques and I Jth A1111ual Symposium on Explosive and Blasring Research, 2-5 Feb 97, Las Vegas, Nevada, International Society of Explosive EnginCCTS. Prodeb1, C., and P. W. Lipman (1989). Crustal sttucture of lhc Rocky Mouawn Regi011, in Geopbysiclll fr.unework of the continentAl United Sta&es, L. C. Palliser and W. D. Mooaey Gallogic:a1 Society of AIIICIKa MCIIIOir 1'72, Boulder, Colotado. Richards, P. G., D. Anderson. ead D. Simpson(l992). A swveyofbiDStill; activity in the Ullili:tl Stales, Bull. S~ism. Soc. Ant Ill, 1416. Sobel, P. A. (1978). The effect of spll11 on mb llftd Ms, Tdcdync Geotech Repm SDAC-TR-77-12. DAllas, Texas. Smells, G. G., E. T. Heni11, llDd J. L. Boancr (1997). Consuuction of rcgiOIIIll ground tnllh da&abascs using seismic ADd infrasound da&a. Seisnt &s. Lett. 68, 743. Stu~. D. W~ 4lld C. llaywlllfd (2000). The role of ground lnllh ill improved identllication of mining e1plosion sipals: utilization of clllibration t:~plosions and acoustic sil!aals. Proc. ofthe 22nd Annual DnDIDOE Seismic &s~arc:h Symposium for PltJIIIIing IDid V~rification of and CompliiiiiC~ with the Compr~hmsiv~ Nuclear-Ten-Ban T~aly (CI'IJT), 12- IS SC(IIembcr2000. New Orleoms, Louisiana. Stump, B. W., aDd D. C. Pearson (1997). Comparison of sinsJe-fired and dday-flrcd cxplosiou At n:gioaal Alld local dislnnces, Proc. of the 19rh Annual Seismic R~searc:h Symposium on Morritoring a Comp~­ ltnlsiv~ Ten Ban Treaty, AllllllfOlis, Marylud, 23-25 Scptcmbct 1997. B. W., and S. K. Reamer (1988). Tempcnllllld spllial soun:e ef. feels from IICIU"-Siri~~ec explosions, in lOth A1111ual AFGI/DARPA Seismic Restarch Symposium, FDilbroolc. CalifonriL Scu~. D. W~ 111111 R. E. Reinke ( 19&8). Experimeatal confumalion of superposition from small-sale explosions, Bull. Seisn~ Soc. Am. 78, 1059-1073. SlUmp, B. W. (1985). Conslr.lints on e~ptosive sources wilh .cpllll from near·sourcc waveforms, Bull. ~ism. Soc. Am. 75, 361 - 3n. Yans. X. (1998). Min&~A MA TI.AB GUI J'rosnun to Calcullle Synlhctic Seismograms from :~ Linc:lr. Multi·shoc Blast Soura: Model. LAUR-98-1486, Los Alamos National Labcntory,l..os Alamos, New Mexico. Stu~. Weston Gcphysic:al C~on 917 Ellis Ave., Suite 222 Lulkill, Texas 75904 bonner@westong~physical.com (J.L.D~ D.C.P., W.S.D.) MMusaipt received 23 Apil 2002. Bulletin of !be SeiJmoloajcal Sociel)' ot Alllaica. 89, 2. pp. 409-422, Aprill999 0 Source Comparisons between Nuclear and Chemical Explosions Detonated at Rainier Mesa, Nevada Test Site by Brian W. Stump, D. Craig Pearson, and Robert E. Reinke Abstract A series of nuclear [MINERAL QUARRY (MQ), HUNTERS TROPHY (HT)J and chemical [NON-PROLIFERATION EXPERIMENT (NPE) and NPE CAL] ex- 0 plosions were detonated in the same geological material at Rainier Mesa, Nevada Test Site. These sources were extensively instrumented with the same near-source, free-surface instrumentation array. The data from these explosions allow the establishment of empirical source scaling relations as wen ao; investigation of possible chemical and nuclear source differences. Even in the near-source region at common receivers, the data display propagation path effects resulting from slight differences in source locations. These effects are effectively taken into account by smoothing the source comparisons across the different stations in the receiver array. As many as 30 individual waveforms from each source are used in this smoothing process. Comparison of HT and NPE at near-source distances indicates that within the bandwidth of the data (0.36 to I00 Hz), there is no apparent spectral difference between the nuclear and single-fired chemical source. The smoothed spectral ratio between the NPE and NPE CAL is consistent with the long-period source spectral difference (I 0 4), comer frequencies (2 to 3 HziNPE and 40 to 60 Hz/NPE CAL), and highfrequency decay (j'- 2) similar to the Mueller-Murphy source model for wet tuff after modifying the cavity radius to scale as rhe cube root of yield. Comparison of the two nuclear sources, HT and MQ, indicates that at long periods, the HT/MQ ratio is 0.4 to 0.6 with the spectra from the two explosions merging above 5 Hz, which is consistent with the 0.3 magnitude difference observed for the two sources. In all the source comparisons, the spectral ratios of the transverse components of motion are indistinguishable from those produced by either the vertical or radial components. This fact argues that the transverse component of motion from an explosion is generated at very close-in distances, in this case on the order of I to 2 km. These observations are in agreement with some type of linear scattering mechanism. Introduction and Motivation 0 Seismic source functions an: used to quantify the genemtion of body and surface waves from a wide variety of physical phenomena. Careful quantification of the relative excitation of different seismic phases can be used to identify tbe source type (Pomeroy ~t a/., 1982; Taylor eta/., 1988, 1989). The geomciiy of and the material properties around the source have a strong influence on the waves a particular source generates (Stevens and Day, 1985). The different sources of interest include earthquakes that are considered deviatoric in nature and explosions that are represented in their simplest fonn as isotropic or spherically symmetric. The deviatoric characteristic of earthquakes leads to efficient generation of shear waves, whereas the symmetry of the conrained spherical explosion results in initial P-wave excitation. Cylindrical explosive sources, typically found in the mining industry, result in reduced symmetry IUld somewhat enhanced shear-wave generation relative to spherical sources (HecliUl, 1953; GleM et al., 1986; Reamer et al., 1992). They are not. however, as efficient at generating shear waves as earthquakes. Identification of a source as a spherical explosion, cylindrical explosion, or earthquake is dependent on the combination or these different effects. Theory predicts that these different source types will have characteristically different time functions as well. A number of empirical discriminants for earthquakes and explosions are dependent upon the relative excitation of different frequency components of regional seismic phases. Combining time- or frequency-domain effects with the geometrical excitation, the most robust discriminants can be developed for monitoring the Comprehensive Nuclear Test Ban. The focus of this study is the investigation of the time function effects for single-fired chemical and nuclear explo409 410 0 0 B. W. Stump, D. C. Pcalllon, and R. E. Reinke sion sources detonated in a spherical geometry.lnfonnation developed here in combination with similar studies for earthquakes and mining explosions will be used to improve current discriminants, address the transportability of the discriminants to new regions, and suggest new discriminants utilizing current data sources. The quantification of the sei.o;mic source time function for nuclear and chemical explosions of different sizes provides the basis for identifying source differences that may develop as a function of yield as well as explosi vc type. The yield effects are useful in yield detenninalion as weU as assessing detection and identification capabilities at small magnitude. Source effects attributable to yield, configuration, and material properties can be used to establish new or verify existing scaling relations (Denny and Johnson, 1991). The characterization of the seismic source time function can be compleled in the time or frequency domain. Timedomain characterization quantifies not only the total source strength but also how energy is distributed in time. Frequency-domain representations provide the ability to identify static offsets of the source, characteristic frequency associated with the physical size of the source, and the smoothness of the energy deposition or high-frequency decay. In the case of frequency-domain representations, the phase of the signal can either be included or discarded in the analysis. Ignoring the phase infonnation in the source precludes the ability to track the distribution of source information as a function of time and as such is more restrictive. The data that are utilized in this chemical and nuclear explosion source function study bave been recovered from the free surface in the near-source region (<5 km). As in all seismological studies, the observed data are linearly dependent upon both the propagation and source contributions. This study focuses on near-source data to maximize the bandwidth over which the source comparison can be made and to make any propagation path corrections as simple as possible. It is well known that local receiver effects on observed waveforms can be quite strong (Hutchings and Wu, 1990). In order to mitigate these effects in a comparative way, all tbe explosion sources were observed with a consistent set of receivers (Fig. 1). Problem Definition Experimental Configuration 0 An experimental program was begun in 1988 to define the equivalent elastic source function for nuclear explosions. The location of these CJtperiments was Rainier Mesa at the Nevada Test Site. This region was chosen because the tunnels used to emplace the explosive devices provided the opportunity to make free-field measurements at shot depth as well as free-surface measurements. 1lJc NON-PROLIFERATION EXPERIMENT conducted by the DOE supplied a large chemical CJtplosion source that could be compared to the nuclear explosions (Denny, 1994; Denny et at., 1995). As part of these integmted investigations, free-field (Olsen and Peratt, 1994) and free-surface (Reinke et al.• 1994) observations were made on the nuclear explosions MISTY ECHO (ME), MINERAL QUARRY (MQ), and HUNTERS TROPHY (Hl) and on the chemical explosions NON-PROUPERATION EXPERIMENT (NPE) and NON-PROLIFERATION EXPERIMENT CAUBRATION (NPE CAL) (Table 1). The combination of free·field, free-surface near-source network and freesurface tight array (8 to 80 m) provides a unique opportunity for separation of propagation path effects from source processes (Fig. I). The range of different types of sources (chemical and nuclear) and different yields (-1()2 to lzys lb equivalent TNT) provides dam for characterizing source· dominated processes that may be important in monitoring and identifying explosions in other environments. The range of different source types and sizes in the data set allows constraint of the following aspects of the CJtplosion source model: (1) chemicaUnuclear source similarities and differences (NPE to HT comparison section); (2) utility of small-scale calibration explosions for source quantifies· tion (NPE CAL to NPE comparison section}; and (3) explosion source comparison of small-yield nuclear explosions (HT to MQ comparison section) detonated in similar materials, The observations from this study were made in and on Rainier Mesa above theN-tunnel complex (Fig. 1). The surface instrumentation consisted of Terra Teclmology forcebalance accelerometers (fc > 50 Hz) and Sprcngnetber S6000, 2-Hz velocity sensors recorded by Refraction Technology 16-bit and Terra Technology 12-bil digital acquisition systems. Timing and location were provided by a combination of WWVB, GOES, and GPS receivers. It is important to emphasize that for the MQ, HT, NPE CAL, and NPE sources an identical set of receiver sites were used, thus eliminating apparent source differences that are actually attributable to local receiver structure. The ME data are not analyzed in this comparative study because the overlap of its initial receiver array with the subsequent experiments was small. Data Characterization The characterization of the free-surface data and quan~ titication of propagation path effects is given in an article by Reinke et al. (1994). Local receiver effects, even in the nearsource region, are found to have a significant effect on the observations. The results of the free surface, ncar-source network and free surface, tight array analysis indicate that at relatively low frequencies (~10Hz), stochastic propagation effects are important contributors to the observed wavefonns (0.5 to 3.0 km range). The large scatter in the peak, nearsoun:e velocity data (Fig. 2) is the simplest expression of this characteristic. Over the dimension of the tight array (8 to 80 m), signal coherence above 10 H1. is lost. Similar effects have been aUributed to local receiver effects (McLaughlin et al., 1983; Menke et al., 1990; Vernon, 1985). These data characteristics are indicative of wave propagation Source Comparisons between Nt~clear and Chemical Explosions Detonated at Rainier Mesa, Nevada Test Site 0 411 RAINIER MESA FREE SURFACE NETWORK 0 Figure 1. Rainier Mesa near-source free-surface array for the experiments MISTY ECHO (ME), MINERAL QUARRY (MQ), HUNTERS TROPHY (HT), NON-PROUFERATION EXPERIMENT (NPE), and NPB CAL (same location aa NPE). 0 effects attributable to a complex, lhree-dimensional sbUctun:, possibly at the receiver, and suggests that deterministic modeling of the observations for purposes of constraining the source may be difficult. Relative comparisons between sources recorded at the same receiver can take into account these complex receiver effects but do not take into account differences in source locations (Fig. I). Questions associated with the effect of differences in locations of the sources on wavefonns recorded at common receivers will be addressed empirically. Typically. the differences between the locations of the sources in Ibis experiment are on the order of 10 to 30% of lhe tolaJ propagation path distance, about 2 km. Hutchings and Wu (1990) have argued that coherence for closely spaced sources is higher than observutions from closely space receivers of the same source. They argue for coherence in signals for sources separated by several wavelengths but observed at the same receiver based upon experimental work. The lack of coherence at subwavclengtb receiver spacing (Menke et aL, 1990) is explained in terms of near-receiver structure at the free surface. Similar analy- 412 0 B. W. Stump, D. C. Pearson, and R. E. Reinke Table 1 Soun:c Properties E..,. NPE CAL PREDICTIONS AND DATA DIMe Lloll1.aDa Misty Echo 10 Dec 88 37.19903 116.20944 400 s.o Mineral Quany 25July90 37.20687 116.21426 400 4.7 HuniCIS 18 Sept 92 Dcph(lll) POI!m• 37.20693 I 16.20998 385 2l Sept93 37.20141 Non Proliferation 116.20993 Experiment 390 Trophy 10 r-------------------------~ ComD:ula ~1 E .e >C3 .... 4.4 I RadaiVel 1 ' 0 ...J Vertical Vel Transverse Vel IJJ 4.1 285 X IO'Ib of blasting > .01 a&ent; 1.11 b of TNT 001 equivalent• Non 30 Oct92 Prolifcralioo Experiment 37.20141 116.20993 390 +-----------~----------~ 1000 100 10000 RANGE (m) 300 lbofC4: 390lb1Nf equivalcott HT DATA C3libtation 100 • Alsumes a TNT/bla5ting qent equiv:almce of0.70 (Soucn aod Lanoa, 1994). tAISwnes a TNf/C4 equivllleoc:e of 1.3 (Penson ~r 11L, 1994). ses and conclusions have been put forward by Hanis (1991), using quarry explosions recorded at near-regional distances. e"' .to. Vertical Vel •Radcai Vel e Transverse Vel .e >.... C3 10 0 ...J w Analysis Approach 0 > The wavelengths at 10Hz observed in lhe near-source region are on the order of 200 m, and as a result, lhe analysis approach taken is stochastic in nature seeking to smooth lhe data over a number of receivers in order to minimize source location differences 1111d emphasize the more dctenninistic source function effects. Several researchers (Moeller, 1985; Chael, 1987; Reamer and Stump, 1992; Goldstein et aL, 1994) have made relative source comparisons, in some cases including bolh the phase nnd modulus of the Fourier representation of lhe source and in other, more restrictive cases only the modulus. Due to the stoChastic character of the observations used in this srudy and in an attempt to quantify the source.-; in as broadband a manner as possible, the focus here is on the modulus of the Fourier transform and more specifically the power spectrum. The general problem of interest is represented as (where the summation convention has been suspended unless explicitly noted) 1 100 Figure 2. 1000 RANGE {m) 10000 Peak velocity datn for the radial (R), vertical (Z), and transverse componc:niS of surface velocity observed at the smtion.s diagrammed in Figure 1 from the explosions NPE CAL (top} and HT (bot· tom). Predictions plotted for the NPE CAL (solid and dashed lines) are from the models of Perrel and Bass {1974). This decay model developed for f~field data at close ranges is used only for reference to the longer range ftec-surface dam. m sources will be made. In what follows, the explicit inclusion of frequency, f. is dropped. Consider the comparison of MQ(l) and ln"(2) sources on the vertical component (1) of station 1: (1) (2) 0 where U~(}) is the power spectrum from the /cth component (!-vertical, 2-radial, 3-transverse) at lhe ith receiver resulting from lhe fth source (1-MQ, 2-IIT, 3-NPE, 4-NPE CAL). Rlri(f) is lhe receiver function at the ith receiver (assumed to be identical for all sources). PkJJ(f) is the transfer function or Green's function from thejth source to the ith receiver. The jtb source is represented as S1(f). Comparison of power spectral estimates from common receivers of ncar-by In the case where the propagation paths are identical, such as comparing NPE to NPE CAL, lhen the ratios of the transfer functions are 1: pld3 = L PU4 (3} Source Comparisons bt!tween Nuclear anJ Chemical Explosions Detonaled at Rainier Meso. Nevada Test Site 0 Hutchings and Wu (1990) argue that equation (3) can hold for sources that are separated by several wavelengths. 'Iltis assumption will be tested empirically by investigating the effect of source separation at specific receivers on there~ suiting source dctennination. Bias introduced by source location differences will also be quantified by comparing source estimates made from neighboring receivers and searching for systematic effects in these estimates. In cases where such bias exists, smoothing over a number of different receivers (NR) can be used to minimize these effects: (4) If the sources are close to one noother, possibly as far apart as sever.tl wavelengths, then (5) The estimate of the source ratios becomes _ {s,) == \S2 0 NR uil, ul/2. ~ i '* ' NR This procedure applied frequency by frequency provides a mechanism for computing both a mean source spec~ tral estimate and associated variance. Because the spectral estimates are assumed to be log nonnally distributed, these calculations ore completed in log space. Even in the case such as the NPE and NPE CAL comparison where the pro~ agarion paths are identical, utilization of equation (6) pnr duces more stable results. If the source signature is the same on nil components of the observations, additional smoothing can be completed over these components: _ NR 3 L L Uu1 ., ,. ,3•-• u1:12 . (s•) S2 · NR 0 (6) (7) In reality, Uld/(f) is a spectral estimate made from an observational waveform. The foregoing equations involve a spectntl division that can lead to theoretical difficulties in variance estimatioo (Benda! and Piersol, 1971) and practical problems in stability when the denominator approaches zero. To minimize these difficulties, careful consideration was given to smoothing windows applied to both the numerator and denominator in these equations. The first step in this process was the identification of the bandwidlll with ac~ ceptable signal-to-noise mtio. Comparison of lhe signal spectra to prc~vent noise estimates and careful enmioation of tile spectral shape led to a conservatively determined bandwidth of0.36to 1.00 to 100Hz for the data used in this srudy. A characteristic set of data is displayed in Figure 3. 413 The second step was the detennination of an appropriate smoothing window in order to reduce variances in the individual spectral estimates and ultimately improve source difference estimates. Various numerical tests of different smoothing windows were perfonned, and a window criteria that minimized the variance and bias of the speCtral estimate was selected. A single window was chosen with a maximum width of 3.48 Hz (B), close to lhe comer frequency of the data. At a given frequency, the smoothing involves frequencies that an: within ± half the window width. The lowest frequency included in the smoothing is identified a.~ the frequency where the signal is above the noise (0.36 to 1.00 Hz). The uniform and symmetric window is not allowed to extend below this frequency, ~nd so low~ frequency values are smoothed with windows that are smaller in width than the maximum window. This results in higher variances at low frequencies but reduced bias. Com~ parisons given later in the article compare avClllge spectral ratios for 3.48~ and 0.43~Hz windows. The resulting spectra] ratios for these two different windows are quite similar ex· cept that tbe ratios using the 0.43-Hz window have larger variances. The bandwidth (B) time I ol ~~- •~ ~ • e~ ·4000 'I tit ~~~t S tO =l t5 20 Time(s) 200r """' J(, tO' 10 _: :i Time(s\ 15 0 ~ (.) (.) 100 <( · 200 0 10 :: 10 ·g 150 Qj 100 > Ul tO ~~ 10 (.) (.) <( 10 IL._~--;;;---""j"S~to 0 tOO e ~ \ ~,;~ r···. . I Frequency (Hz) Ul 'en 10 < .2()() I ·: " l 200 10 ;;;200 1 10 • 10. 400,-----~-~--, 'l'~ .. HT 1AR (1 77km rango, 90 baz) HT/NPE 1AR COMPARISON HT 1AZ (1 77km range 90 baz) tAZ COMPARISON -~ ! .... (b) ~ ---;--;----' tO 10 0 . 50 15 20 ~~~ 100 5 Froquoncy (Hr) to • Tl.,\l?(Bl !.O ISO 0 . . 5 to Timo{s) IS 20 HT 1AT ( 1 77km rnngo , 90 baz) HT/NPE 1AT COMPARISON 200 20 Ul 'en to " e ~ 100 0 (.) tO u <( ~ E'o 100 200 0 ~ tO 5 15 20 Timo( g 10 > Ul ~ t O ... ~ (.) 10 . 10 2 3 ';' tOO ~ 0 0 f ~ · 100 ~ 10 Froquoncy {Hzl Figure 3. Ul 200 iii 10 - 200 0 p 5 10 Timo(sl (a) Comparison of the vertical accclcrntions from HT and NPil at station JA (fig. 1) in lhe time PJld frequeacy domain. (b) Comparison of the radial (R) and transverse (1) accelentions fOf HT and NPE at station lA (Fig. 1) in the time aod frequc:ocy domain. Jll ~ ::: 15 20 !;:o rn ::<1 l n 0 0 ~ ~ (a) .. (b) HTI , lAZRATIO 10 l 10 ~ !~ 10 1 I .-----------------, 0 to• ••• 10 1 ~ 10 ~ 10 L-----~~------------~ 10'' 10 ~ Frequency (Hl) 1 i E I ~tO I u I ~ 10 u ~ 10 · , IO ~ c·--------~------------~~ I t0° ~ (/)i 10° 1Q I 10 l .!! ~ 10 ,-·-- 10 t E I ~ I 10 3 10° Frequency (Hz) I 10 . 1 ,j • 10'31 n • Hf ----· 10 g ra ~ ~ ,, I I .., j E I i .Ii I r l I I 8. I ~ ~ €1 II s· 10 ' """ ' to " I ~ 10.~---\-·-·· ~ I "" 10 I .Jr., I Froquoncy (ttz) 10 I ~ }l ~ ~ ~ I ! ~ ~ 10 ., 10\f .,"' ~ >10 ,, ~ .s:2' !a ~ .o 10. t [ ~ 1 , •..,,r·t - ' (/) 10' " I ~ ~· Froquoncy (Hz) ! iI 10' I ll u ~ I ~ I HT 1AT (0 36Hz, 348Hz ~ 10 ~- - - - - - ·- 'I I ~ Q l. Ftequoncy [H z) ,. 1AT RATIO ' \ r --,! ·o• r roquoncy (Hz) HT• I i ~ 1 ,I \ 10' Frequoncy (Hz) 10 1a~ 10 0 I 1 . ~ 1 ~ > -'O J .----., . ~ a: Frequency (Hz) .. ~ HT 1 AR (0 36Hz 3 48Hz SM) 1AR RATIO ~ ex: ~...t ··t· 10 ' Cij ~ HT 10c 10 1 1 Q , HT 1AZ (0.36Hz. 3.48Hz SM) > 10 1 .,~ tO ·.l ~·- - - - - - : - ,0::-;;-.------------;:...~10 > Froquoncy (Hz) Figure 4. (a) Smoothed vertical (Z) spcclnl from HT and NPE at station lA (Fig. I) using a 3.48-Hz smoolhing window. The speclnll ratio (HT/NPE) of the smoothed speclnl is also given. (b) Smoothed radial (R) and IOlnsverse (T) spectra from JIT and NPE at station lA (Fig. 1) using a 3.48-Hz smoothing window. The spectral ratios (HT/NPE!) of the smoothed spectra are also given. ~ VI 416 0 B. W. Stump, D. C. Pearson, and R. E. Reinke Table 2 HT and NPB Slanl Ranges Slodoa lP HT SlaM Raqc (m) .. Dillaaace iD Pllll u; ~ ;; ;:: 1990 1810 lA 2P 1560 4A 4P 7A 1270 1380 2340 7P 2430 9A 2380 940 12 .NPil st. Raqc (Ill) Hf·t• l lA Alc.,.,_,.,ls so' 2150 8'1> 1920 1530 61{. 9SO 1120 1770 1880 1780 990 2'1> 28'1> 21'1> 28.... 26.... 29'1. 5.... Hso' r1. • i so' 10 10 10 to' so so' tO' 0 0 near-receiver or near-source structure. Significant differences in the time series are observed. The spectral comparisons, in contraSt, show almost identical shapes and amplitudes over the bandwidth of the signal (0.36 to I00 Hz). The presence of a strong transverse component of motion is not consistent with an isotropic model of the soun:e. The geological structure must be represented by a complex threedimensiooal structure where scattering into the ttansverse component can occur. Smoothed spectral estimates from two different explosions observed at a common receiver were formed into a ratio (equation 2) in order to eliminate the common receiver and propagation effects and emphasize the differences between the two sources. The resulting Z, R, and T spcclr.tl ratios (IITINPE, Fig. 4) display similar gross characteristics: (1) nearly fiat from 0.36 to 8 to 10 Hz; (2) low-frequency ratio near 1; (3) high-frequency ratio that increases with fre.. quency; and (4) significant high-frequency variability. These ratios suffer from the common estimation problems in ratios associated with the large variances in both the denominator and numerator of the ratio. Slight differences in propagation paths from the two sources results in incomplete separation of propagation path effects from the data. The transverse components are found to give results that are similar to those from the R and Z component analysis, suggesting that the transverse data retains a source signature similar to the other components. Taking advantage of the source signature in each component of motion and attempting to improve the variance estimates, the ZRT ratios from each station were combined into a single-station dependent estimate, Figure 5 (inner summation, equation 7). The mean (solid line) and log normal variance from the mean (dashed lines) illustrate the small scatter in the low-frequency estimates and the increased scatter with increasing frequency. The increase in variance occurs at 5 to lO Hz. Taking a phase velocity of 2 lan/sec, the wavelength at which the scatter increases is 200 to 400 m, approximately the scale of the source separation. This wavelength argument is in agreement with the findings of Hutchings and Wu (1990). If the high-frequency variation in the spectral ratios is reftective of the different locations of liT and NPE, as seen at each receiver, then one might -10' so ~ HT/NPETPAIC_...,. 10' 10' so Figure 5. Combined (mdial, vertical, and llllnsvcrsc) mean spectral ratios at &tlllions lA, 2P, and 7P (Fig. 1) ror the explosions HT and NPB plotted as a solid line. The plus and minus one standard deviation (log normal) from the mean spectralmtio estimate is also plotted liS a dashed tine. expect to average these differences by smoothing over a number of observations at different azimuths and distances, equations 6 or 7. Figure 6 displays the smoothed, all-component spectral ratio using all station pairs with propagation path differences of no more than 30% (equation 7). Thiny individual spectra from each of the HT (numerator) and NPE (denominator) experiments (each wilh the time-bandwidth product of 57) went into this source difference estimate. 1be high-frequency variations observed in the single-component and single-station estimates are minimized, although lhe variances in the spectral ratios increase with frequency as expected from the stochastic model. Also included in this plot is the spectral ratio estimate using a smoothing window with width 0.43 Hz. One can see lhat there is little substantive change in the ratio, although the variances are increased as a result of decreased smoothing. There is some indication of a rise in the long-period ratio although it is within the variance estimate using the longer window. This analysis eliminates propagation path differences Source Comparisons between Nuclear and CMmit:al Explosions Detonated al Rainier Mesa, 0 due to different source locations for the HT to NPE. These variations have been averaged out to emphasize source processes. Comparisons between single-component, single·sta· lion, and whole-array source estimates support the nssump. lions needed to use equation (7). Tile resulting spectral comparison between NPE and HT is flat from 0.36 to 100 Hz with the ratio of the source siJ'engths equal to I. There appear to be no significant spec· tral differences between the chemical (NPE) and nuclear sources (HT) in the near.source region from 0.36 to 100Hz after stochastic propagation path effects ore taken into ac· count. 0 417 10:1.:-_ _ _ _ _..,__ _ _ _ _'":-----.......1 10 0 Test Site , HTINPE 14. 1P.2P. ~ AAP,7A 7P.BP.9A. 12P 3 48 and 0 43Hz Smooth•ng 10 Utility of SmaU-ScaJe CaJibration Explosions for Source Quantification (NPE CAL to NPE Comparison) In order to quantify propagation path effects expected from the NPE and exercise the data acquisition systeiDS, a small, JOO.Ib charge of C-4 was emplaced and detonated ar the center of the planned NPE source cavity prior to its excavation. This small source provided the opportunity to test the empirical Green's function approach to source scaling in the near-source region (equation 3). Since the NPECAL was nearly four orders of magnitude sma11er than the NPE and the dynamic range of the accelerometers (used for recording the NPE, HT, and MQ) was limited, a set of Sprengnether S6000, 2-Hz seismometers was used to record the NPE CAL at the same locations where the accelerometers were fielded for the other explosions. The instrument comer of the seismometer is in the band of interest for source comparisons; therefore, this well-known insiJ'Ument response was decon· volved prior to any spectral comparisons. The smaller NPE CAL waveforms have a higher comer frequency and more complexity than those observed from the NPH. Spectral comparisons in velocity illustrate the four orders of magnitude difference in spectral level at low fre. qucncies, an order of magnitude difference in source comer frequency, and the l.O to tOO Hz bandwidth of the data. Because the centroids of the NPE and NPE CAL sources are identicaJ, there should be no differences in propagation path effects for the two sources as long as the point source rep. rcsentation is appropriate. H secondary source processes such as spall are important contributors to the NPE waveforms, this assumption may not be vaJid. In order to improve the statistical significance of the source comparisons, the observed spectral ratios for all the Z, R, T, and combined single-station estimates were averaged as done in the previous anaJysis (equation 7). The averaged ratio for NPHINPE CAL is displayed in Figure 7 (lP, 2P, 4P, 6P, 7P, BP, 9A, lOP, I lP, 12 for a total of 30 individual components). Again, ratios calculated with a 3.48-Hz window (thick Jines) and 0.43-Hz window (thin lines) are compared. Within the band· width of good signal to noise (1 to 100 Hz), there is little difference between the results except that the variances for the larger smoothing window are reduced. Unlike the HT/ NPE comparison, there is only a small illcrease in variance N~ada 1 lOr. frequency (HZ) t0 1 Figure 6. Mean spectrlll ratio (HTINPE) determined by averaging all station pairs (propagation path differences 8P 9A 2130 2120 0.3~ 2380 2410 1~ lOP 1360 1440 6~ IIA 1010 1230 20'>1> 2A 2P 20~ 24~ 25~ 17~ 19~ 3~ Source Models Source Comparisons between Nuclear and Chemical Explosioru Detonated 01 Rainier Mesa, Nevada Test Site 0 vided the ability to replicate such a large r.mge of sources has been its depth dependencies. All the shots in this data set are below the water table. Additionally, the scaled depths range from 377 mlkt 1113 (NPE) to 6725 mlkt113 (NPE CAL), so the scaling of the model cnn be compared to the data. Denny and Johnson (1991) in a comprehensive study of source models in a wide range of materials suggest improved scaling relations for long-period level and comer frequency that can also be used in interpreting the empirical spectral ratios. In order to interpret the yield (W) and depth (h) effects quantified by our empirical data, a set of Mueller-Murphy source functiorLc; for wet tuff was calculated using the following material properties for the source region: P velocity (a) 2.20 km/see S Velocity (/J) 1.27 lonlsee Density (p) 1.85 gm/cc The Mueller-Murphy source function in the frequency domain is represented as U,. = P(w) · re1 4pr • w& + iwa iWrPJ - l:tiJ2' 419 tunity to investigate the effects of overburden pressure predicted by the models. Figure 9 plots the spectral ratios pre· dieted by the Mueller-Murphy model for the NPE CAL at depths of 7 m (122 mlkt113 ) nod its actual depth of 390m. A similar comparison is given for the NPE as well. The higher corner frequency for the deeper source in each comparison is in general agreement with the empirical comer frequency detennined from the data (Fig. 7). There are similor large effeets on long-period levels, over an order of magnitude for the NPE CAL at the two depths of burial. Depth of burial source time function effects are often difficult to separate from the relative excitation function of body waves and surface waves as a function of depth implicit in near-source Green's functions (Flynn and Stump, 1988). Placing the NPE and NPE CAL at the same depth eliminates differences in Green's functions and thus allows the data to be used to investigate the effect of depth on the two sources. It is interesting to note with the wide range of yields and explosive type.<; included in the Rainier Mesa data set that a small number of modest-sized chemical explosions detonated at several additional depths would provide a definitive data set for quantifying depth effects and overburden on the source. (8) •o•.-----S-'-pec1Tal--A....al_lo_of_t_ .1_1kl_at__t28_m_and_380_m_ _~-. 0 where P(m) is the Fourier transform of the pressure time function at the elastic radius: (9) r~1 is the clastic radius, r is the source to receiver range, Wo = air.,,£= (l + 2p)14p, "/ == l.5£0o, nod ,l,p are Lame constants. The depth dependencies that enter into the source representation for saturated tuff-rhyolite (Murphy, 1977) include the elastic radius, .... I ·~~~~.-------~....-------------~.0~.----------~.~ (1 O) J Speetra/ Ratio ot 390 lb at 7 m and 390m 10·~-....-.-;o.--_,......-----....-----. the cavity r.ldius, W0-29 r.: = 31.4 ""j'/ITT'; (11) and the static pressure that affeets the long-period spectral level, (12) •o·• 0 As indicated in the Mueller-Murphy source model, overburden stress has a significant effect on comer frequency and long-period level of the explosion source. The scaled depths for the events in this study provide an oppor- Fnoqulncy (Hz) Figure 9. Spectral ratios of a 1.1 1-kt MucUerMurpby sources (top) at depths of 126and 390 ~ (122 Dlld 3n mlkt1") and 390-lb soW"te (bottom) at depths of 7 Dlld 390m (122 and 6710 mlkl 1"). 420 0 0 B. W. Stump, D. C. Pearson, and R. E. Reinke Denny and Johnson (1991) summarize scaling relations for cavity radius by a number of workers and proceed to find an empirical scaling relation based on a comprehensive daaa set. They find strong evidence for cube-root scaling for cavity radius as well as a dependence on material properties such as gas-filled porosity and shear velocity. One might expect that for explosions detonated in the same material and at the same depth that cavity radius would scale as wm. This approach bas been recently aaken by Murphy and Barker (1994). The nearly four orders of magnitude in source scaling between the NPE and NPE CAL accentuates this effect. Figure 10 compares spectral ratios between MuellerMurphy source functions with yields of 1.11 let and 390 lb. In the first ratio (solid line in Fig. 10), the cavity radius is scaled as W0-2!1 (equation 11). The second ratio (dotted line in Fig. 10) is the same spectral ratio as before, except that the cavity radius is scaled according to W113 based on the similarities of emplacement media for tbe two explosions. The primary effect of cavity radius scaling on the theoretical spectral ratios is the offset between the two curves with the cube-root scaled curve approximately a factor of 2.8 larger. The modified Mueller-Murphy source model speettal ratios that include cavity radii scaled according to wt13 are compared to the empirical speetral ratios in Figure 11. This theoretical model follows the empirical data through its entire bandwidth. If the standard Mueller-Murphy scaling for cavity radius had been included in the models, then the shape of the spectral ratio would have been modeled, but the absolute amplitudes would have been decreased by a foetor of 2.8, falling outside the mean ± 1 standard deviation estimates displayed as black dashed lines in Figure 11. Thus. not only the shape but the absolute amplitude of the Mueller-Murphy theoretical spectral ratio (with cavity radius proportional to the cube root of yield) arc consistent with the empirical spectral mtios between the NPE and the NPE CAL It appears that this model is appropriate for wet tuff over the four orders of magnitude of yield represented by the NPE CAL and NPE explosions. The Mueller-Murphy model was developed for nuclear sources and is here applied to data from two chemical sources, one of which is significantly smaller than the yield range norma11y thought to be applicable for the model. SpedrW Rallo 1.11kt/3901b will rc - W0-211 (dOlled) and rc ~ Yfi!J (101111) i "'~ F~(tlzl Figure 10. Comparison of speccral ratios of a 1.11-lct and 390-lb sources using the Mueller-MUI)!h_~ source model in wbicll cavity radius scales as ~ (solid line) and a modified Mueller-Murphr source model in which c:avity radius scales as wo- 3 (dotted line). •c'r----N-PEIN_P~E_c_AL_&_t_ao_nd_m_10Cie_lc,....OI11j)a~-"son---s_ __, Conclusions c In the ncar-source region there are no apparent spectral differences betweeo HUNTERS TROPHY and 11iE NONPROLIFERAnoN EXPERIMENT in dte bandwidth of 0.36 to 100Hz. Figure 6 demonstrates the flat spectral ratio for these two events with a mean value of 0.96. Although the variance of the spectral mtio estimate increases as frequency increases, the mean value sbows little departure from one across the entire data bandwidth. These results indicate that there is littJe information in the ncar-source wave field that can be w;ed to distinguish single-dlarge chemical from noclear explosions. Although this result suggests that seismic 10 ~ ~~ 0 --------~~ .,~-------,~0.~--------J,, froquoncy IHl) Figure 11. Comparison of the predicted spcccral ratios from the Mueller-Mwphy wet tuff model (light solid line) with cavity radius liCaled by W113 for explosion of yield J.II Itt aud the NPEINPE CAL empirical data (dark solid line) with dashed lines rcpre· sentinr;: the plus or minus one standard deviation bounds of the empirical estimates. Source Comparisons between Nuclear and Chmrical Explosions lktonated at Rainier MeSQ. Nevadn Te.st SiJe 0 0 0 waves cannot be used to discrimiruue nuclear explosions from chemkal explosions, it also supports the use of large chemical explosions for calibration to l'l!:plicate nuclear explosion effects. One could imagine a series of chemical explosion experiments to quantify source phenomenology. near-source material propeny effects, and regional explosion effects in areas whet\!: only earthquakes have been obseiVed in the pasL Glenn and Goldstein (1994) repon on free field data from the NPE and a nearby nuclear explosion. They scaled the resulting specltll for each event by yield and then proceeded to use numerical simulations including nonlinear material models and explosive characterization to satisfactorily replicate the observations. Goldstein and Jarpe {1994) repon on a comparison of chemical and nuclear source spectra ot close-in, local, and regional distances reaching a conclusion consistent with those of this article. Chemical and nuclear explosions of similar yield produce indistinguishable waveforms. They conclude, as do we, that this result suggests that conlained chemical explosions couJd be used for calibration in l'l!:gions where no nuclear explosions have been detonated (Denny and Johnson, 1991). Comparison of the empirical specltlll ratios for the NPE and the NPE CAL suggest that the NPE is best l'l!:plicated by 11 Mueller-Murphy model with a yield near 1.11 1cL The empirical data supports a Mueller-Murphy source time function for wet tuff with cavity radius scaling as the cube root of yield at constant depth. Despite the fact that the modified source model was developed for nuclear explosions. extension to chemical explosions is supported by this study. These comparisons also suggest that this modified model for wet ruff is extendible to quite small-yield explosions. The fit to the empirical data is also consistent with the deplh corrections suggested by the theoretical model. The Mueller-Murphy model includes strong depth corrections. The data analyzed in this study, although from sources at nearly identical depths, are consistent wilh these corrections. The wealth of explosion ground-motion data from different sowce types and yields suggests that these corrections could be further tested with the detonation of several additional moderate-sized chemical explosions at different depths in Rainier Mesa. The raw data display strong first-order propagation path effects at these dose distances that must be taken into account prior to making any source comparisons. lt was only because different types and sizes of sources were recorded by the same receiver array that this relative source comparison study couJd be under1aken. Even with the suite of constant receiver sites for the different sources, significant spectral averaging across different receiver sites was necessory to emphasize source-dominated processes (equation 7). These empirical results support the contention of Hutchings and Wu ( 1990) that sources separated by up to several wavelengths but %\!:Corded by the same receiver array can be used in relative source comparisons. The subwavelength variations observed by a number of investigators are attributable 421 to near-m:eiver structure. Additional frequency-domain smoothing was employed to stabilize the spectral division. In all the comparisons (HTINPE, NPE/NPE CAL, nnd HT/MQ), the analysis of transverse components of motion produced results that are identical to those from the radial and vertical components of motion. The mecbon.ism for SH generation must be linked to the source function in the same way that the P and Sv waves (radial and vertical motions) are linked. A linear scattering mechanism for the generation of transverse motions is consistent with these observotions. These results suggest that the transverse component of motion receives a strong source imprint within a region approximately one-wavelenglh distant from the explosion. Acknowledgments This wort was perl'onned under the auspica of the U.S. Department of EncrJy by Los Alamos National Laboratory IIOdc:r Contract W-7405· ENG-36, Defense Nuclear Agency, Air Fon:c Phillips Labora!Or)', and the AdvllllCCd R.eseuch Projects Agency by Soutbcm Methodist University Wider Contrast F29601·91·0.DB20. Special thanks &O 10 AI Leverette for dala acquisition. Ked Koch helped with the 50IIfa: modeling. The IIJticJe was impro~d by disc:1151iaas wid! Sieve Taylor, G.·S. Min, Laoe Jolm&oo. and Tom Weuvcr. Carcflll review. and co~ by Marvin Denny and Lewis Glenn arc uaud. References Bendal, 1. S. and A. G. Pienot (1971). RANDOM DATA: ANJiysis ond MtasurrtMnl, Wiley·lntersc:imc:e, New Yorlc, 4iT7 pp. Chlel, E. P. (1987). Spectral sc:alins ofearthquakes in the Mirunichi ~gion of New Brunswick, BulL Seism. SIX. Am. 77, 347- 365. Denny, M.D. (Editor) (1994). Proc. oftlrt Symposi11111 on rlrt: Non·PtOlif· eration E:q¥rimenl (NPE}: Rtsults ond lmpllcarionsfor rhe Ttn &rn TI'UIIin, 19-21 Apil 1994, Roclcvillc. Mluylaod. CONF 9404100, La~nce Uvennore NalioDall.abonnory. Denay, M. D. and L R. JohnsoD (1991). The explosi011 seismic soun:e llmction: models and scaling laws revisited, in bplcnlon Source Pile· nCitfiDIOiqgy, Gtoplrysiasl MOIWgrap/1 6S. Amcrlcan Geophysical Union, 1-24. Denny, M., P. Goldstein. K. Mayeda. and W. Walter (199.S). Seismic I'C!Iults from DOS'• Nou·Prolifcratiou Expcrbncoc a compari5011 of chcmicaJ Uld nuclear explosions, in MDIIilorlng u Comprtherui~ Ttsl .8Dn Trtary, NATO ASI Scri~. Kluwer AQdemic, Hingbam, Massachu. seiU. Der, Z. A.. M. R. Hinmo. and R. H. Shumway (1990). Cobcn:ut processing of regional signals 111 small seismic IUTIIY5. BulL ~ism. Soc. Am. 80, 2161-2176. Flynn. E. C. and B. W. Sftunp (1988). Effecu of 60\lrCC dcplh on ncar· soun:e seislllOglliiDS, J. Gtophy3. Res. 93, 482()..4834. Glenu, L. A. and P. GoldslciD (1994). The inllu= of material models on chemical or DDCiear ~plosioa soun:e-lime fuactions, in Proc. of rhe Symposium Oft lilt NOtt·PtOliftraziOft Ezptrimtlll (NPE): Results ond Jmplicatiomfor the Ttst B011 Trtaliu, 19-21 Aprill994, RoclcviHc. Marylalld. CONF 9404100, lAt>Jm~Ce Uvcnn01e National Labo111• IOQ', 4-68-4-77. Glenn, L.A., B. Mphys. I. 1111. 110. 435-450. Reinke, R•.E., A. Leverett.e. B. W. Stump, D. C. Peanon, and C. L. Edwards (1994). Near-source surface Jcismic 111CD$umnents of 111e NPE, in Proc. of the SympotiUift 011 the Nort-Prolijerallort ~n-111 (NPI!}: hsultsotullmplicatiottSf()rfM Ten&m Tnatiu. 19-21 Aprill994, RDcltviUc, Maryland, CONF 9404100, Lawrence Uvennore National l..abonitory. 6-121-6-135. Souers, P. C. and D. B. Larson (1994). TIGER calculations aad the NPE energy yield, Arm.r Colllrolarul NcmproliftratiOII Technologies, Fint QUDrttr 1994. Stevens, J. L IIDd S. M. Day (1985). The pbysical buls of m..:Ms and variable fmluency magnitude mctbods for carthquakrlexploslon discrimination, J. Gtopfry1. Ru. 90, 3009-3020. Taylor, S. R., N. W. Sbcrman. and M. D. Denny (1988). Spccual di5Crimi· nation between NTS cxplosious and western United Sta!Q earth· quakes at regional di5luc:a. Bldl. ~i.rm. Soc. Am. 78, 1563-1579. Taylor, S. R., M.D. Deony, E. S. Vergino, and R. E. Glaser (1989). Regional disaiminadoo betweea NTS explosi0115 and western U.S. ~ 8111f. &iMtt. Soc. Am. 79, 1142-1176. Vernon, P., J. Fletdlcr, L Harr, T. Bolswlck, B. St:mbera, and J. Brune (1985). Spatiul coherence of bodywaves from local eanhquakes recorded on • small aperture amy, Arms C0111rol turd NonprolifertJtiort Ttchnologles, Flm QUDrttr 1994, Trrms. A'". Gtophys. Uflion 66, 954. voo Scgaem, D. and R. Blandford ( 1972). Source lime functions lllld specIn for ltollerground nuc:leDr expiations, ~- J. R. Attr. Soc. 31, 83-97. Depll1mc:Dt of GeoioeQ1 Sciences Soutllem Mdhodist University P.O. Box 7S039S Dallas. Texas 75275-0395 (B.W.S.} Geopbysics Group, EES-3, MS-C33S Los Alamos National l.abontory Los Alamos, New Mexi<:o 87545 (D.C.P.) Defense Thn:al Reduction Agcnc:y SWP-1 Xirtlaad AFB, New Mexicu 87117 (R.E.R.) M110usc:ript received ll! Januazy 1995. .. 81111eWI ollbe Sdiii!OioPeal Sacidy ol "-'ic:a. 89, 6, pp. 1$7$-1~90. Dec:aDIItr 1999 0 Aftershocks of an Explosively Induced Mine Collapse at White Pine, Michigan by W. Scott Phillips, D. Craig Pearson, Xiaoning Yang, and Brian W. Stump 0 Abstract We recorded an explosively induced. 320-m-deep mine collapse and subsequent aftershocks at White Pine. Michigan, using an array of 12 seismic stations, sited within 1 km of swface ground zero. The collapse, wbicb followed the rubblizing of a 2 X HY" m1 panel of a room-and-pillar copper mine, was induced to facilitate leaching operations. The explosions produced little seismic energy; bowever, fracturing and collapse stages produced large signals that were observed at distances up to 900 km. yielding a magnitude (mbz.,) of 2.8. Previous work showed the initial collapse to be an expanding seismic source, interpreted as an opening tensile crack, opposite to the implosional character most often observed for natural mine collapses (Yang el al., 1998). We COWlted over 4000 aftershocks; their occurrence rate followed the modified Omori law: rate = 560 (time - 0.01) - 13, with time in hours. Based on P-wave polarities, we identified events of shear-slip, implosional, and tensile character in the aftershock sequence. For shear-slip events, we found stress drops of 1 bar or less, seismic moments of 1015-1017 dyne em, (M._ - 0.8-0.5) and source radii of 10-50 m. Comer frequencies for implosional events were relatively low, an indication that the collapsed cavity played a role in the source process. This caused irnplosional events to separate from other events in source parameter plots, providing a technique for classifying events of unknown type. We obtained locations of 135 aftershocks using P- and S-wave data. The aftershock zone was less than 100-m thick. situated just above and along the western, mined edge of the coUapsed mine panel. Implosional events occumd at the bottom of the active volume, while shear-slip events were distributed throughout. Shear-slip focal mechanisms indicated thrusting along north·slriking planes, consistent with the high, eastwest regional compressive stress, coupled with a local decrease in vertical stress. The inferred deficit of vertical stress above the western panel edge following collapse indicated that overburden load shifted preferentially to the surrounding, urunined areas, consistent with lower-than-predicted stresses measured in lhe first row of intact pillars. Introduction 0 The 1995 controlled mine collapse aJ White Pine. Michigan, gave us the opportunity to study a collapse-aftershock sequence in detail, by allowing advance deployment of a close-in seismic aany. Our primary purpose was to study the seismic source. aiding our effort to discriminate between mine collapse, nuclear test. and other man-made and natural seismic events, which will be impoctant under the new Compreben&ive Test Ban (U.N. General Assembly, 1996). However, ground motions generated by the collapse and associated aftershocks yielded a wealth of information pertinent to mine engineering and environmental issues as well. In particular, the distribution and failure modes of the aftershocks should be related to the induced sbesS changes (Hasegawa et aL, 1989; McGarr, 1992a,b; Young and Max· well, 1992; Urbancic et al., 1993; Baker and Young, 1997). Collapse-induced stn:ss changes must be understood in order to predict the impact on swrounding mine structures, whicb will aid future design efforts to confine collapse to the planned area. Additionally, the distribution of aftershocks with depth may help evaluate any effect of the collapse on sballow layers where the Jocal, potable aquifer resides. In the following, we will describe lite coUapse and aftersbock data. analysis methods, and results with emphasis on event classification. aftershock locations, and so~ mechanisms and discuss implications for stress redistribution and effect on mine infrasuucture. 1575 W. S. Phillips, D. C. Pcan;on, X. Yang, and B. W. Stump 1576 Setting 0 The White Pine Mine is located near Lake Superior on the Upper Peninsula of Michigan (Fig. 1). The primary mineral mined is copper, which was hydrotbennally emplaced into low-grade, metamorphosed, sandstones and shales of pre-Cambrian age (Mauk et al., 1992). The underground workings at the mine, shown in map view in Figure 2. are extensive, with area of roughly SO Jan2• Historically, portions of the mine bave collapsed naturally. North-central portions of the mine have collapsed slowly over a period of many years. An area to the southwest of the White Pioe fault failed catastrophically 14 January 1988, producing locally rMinne,ot ~ J; II 0 COVCI)' method at White Pine. Recently, a nwnber of eco- nomic factors led to discontinuing the room-and-pillar operation and to investigating the effectiveness of pillar rubblization and in situ leaching of the ore body remaining in the pillars. The controlled collapse documented here occum:d in September 1995 and is the first of two collapse experiments performed at White Pine. A layer of glacial till. 10- to 20-m thick. covers the surface at White Pine. The top of the water table is shallow, 1-2 m beneath the surface. Pre-Cambrian bedrock. consisting of Freda sandstone, Noosuch shale, and Copper Harix>r conglomerate, underties the glacial till. The mine follows the sbal~onglomeratc interface at a depth of 320 m in our study area. The depth to this interface varies laterally. The White Pine fault strikes northwest-southeast (Fig. 2) and dips steeply to the southwest The fault bisects an anticlinal structure that plunges 10° to the southeast. These geological slrUCtures will be important to consider when calibrating the subsurface for microeartbquake location purposes and for understanding the seismic n:spoose of different layers to collapse-induced stresses. Seismic Data Collection W"SCOI'\I In 4.: :t! felt ground motions (ML 3.6) and extensive damage to underground mine structures. Room-and-pillar mining has been the primary ore re- 27C" 272 274' Figure 1. Location of the Wbite Pine Mine, Upper Peninsula, Michigllll. Regional stations used in the mw., analysis are also shown (lrillllg}e:!). Prior to the collapse, we fielded a thtee-component, surface seismic network above the mine panel that was to be collapsed (Fig. 3). Each station was instrumented with a sixchannel, Refraction Technology Model 72A-08 data logger that was continuously locked to OPS-broadcast timing sig- A 5 e ~ At ~ Collapsed :. a581 Panel z 94 e 95 Easting (km) Plan view of underpouod w~np. at White Pine. Failed areas of the room.·and·piUar mme arc indicated by black pau:hes, including the 1995 Figure 2. 0 induced collapse. figure 3. Map of lhn:e-c:omponent seismic sta· tiona (lrianglca) used to monitor the induced collapse at White Pine. Mined areas are indicated by light grny fill; the collapsed panel, by dark gray rm. Axes give eastina IUid nonhing in Michigan sliiiC coonlinates. 96 Afttnhoclcs ofan Explosively lndMced Mint Col/Qpst IJI Whitt Pille, Michigan c 1571 nals. Three-component, 1-Hz Mark Products Model ~3C geophones were fielded at all stations, and an additional three-component Terra Tech SSA-302 force-balance accelerometer was fielded at station 2. Sensors were deployed with horizontal components aligned to true north and east. Stations were programmed to record event-triggered data independently with the exception of station J3 (surface ground zero), which recorded in continuous mode. Stations 2 and 13 were digitized at 250 samples/sec, other stations. at 500 samples/sec. Antialias filters were set with corners 959& of the Nyquist Precise station locations were oblained using handheld GPS receivers (Table 1). The 1995 Induced Collapse 0 1be pillar-removal operation was conducted on 3 September 1995 at 5:39p.m. local time (246:21 :39:38 UTM). Ground zero was at 46.7297" Nand 89.5012° W, the location of station 13. Seventy-two pillm with average dimensions of 6. I m by 12.2 m were loaded with an average of 800 kg (1800 lb) of explosive per pillar for a total explosive source of 58,000 kg (130,000 lb). A delayed firing pattern, 325 milliseconds in leugtb, was used to minimize vibration effects at the surface and propagate the collapse toward the unmined faces (Fig. 4). The area of the collapsed panel was roughly 2 X 10" m1. All seismic srations triggered on the induced collapse event and continued to trigger during the aftershock sequence. Because there was no notkable expression of the collapse at the assembly point, 5 km from ground zero, a video camera deployed near surface ground zero was recovered within I hr of collapse to verify that the explosives had detonated and the ccUapse had occum:d. During recovery of the camera, aftershocks could be felt and heard at surface ground zero. Figure 5 shows the ground motion recorded at surface ground zero (station 13) during the collapse. At this amplification. the individual explosive soun:cs in the pillars are 50T Figure 4. Map view of delay-fire patlem used to rubblizc pillars. Intact rock and pillars are shown in gray. Only pillars that bad been planned to be demolished arc shown. Annotated contours indicate delay times (mscc:). ·ID,.---.---.----.-----.---.---...---, Slallon 13, Vertical Component Table 1 Slalion Locations (Michigan State Coordinaccs) aod Arrival-Time Com:c:tions for the 1995 White Pine Expc:rimeot SlldQo I 2 3 4 s 0 6 7 8 9 10 II 13 NanlllaJ l!aWIJ (till) (till) 581.346 581.125 580.368 580.346 582.138 581.010 581.186 581.688 581.631 581.612 581.83S 581..311 95.202 95.010 94.746 95.166 95.693 .. s.w-. 1'-W.ve ~( _) CumaiGD(.. _) 10 4 15 -4 1:1 -2 4 I 4 -23 -41 95.466 94.512 94.215 94.427 94.516 94.455 -18 94.959 20 2 7 9 11 10 -9 - 7 -13 -IS -17 29 ·tD'-~o--...,o'":!.s--~--,,'::-.s--..,z~--:!2.5':---'3~-.J Time (seconds) Figure 5. Vertical component of grouud-motion velocity at surface ground zero (station 13) during collapse. The lower trace has been amplified a factor of JOO to sbow effects of the pillar blasts. Time is measured relative to the initiation time of !be firiug seqaenc:e. W. S. Phillips, D. C. Pearson, X. Yang, and B. W. Swmp 1578 0 0 not visible, but failure of the pillars and the mine back (roof) is indicated by the cady high-frequency arrivals on the top trace. These failure signals ride on top of a long-period signal indicating an initially upward motion interpreted as the fonnation of a large, horizontally oriented tensile crack (Yang~~ al., 1998), opposite to implosional source mechanisms commonly observed for large, natural collapses (e.g., Pecbmann et al., 1995; Boler et al., 1997). This is followed by strong downward motion associated with the impact of the released material. Ground motion associated with pillar blasting can be seen if we amplify the signal immediately preceding the collapse, as shown by the lower trace in Figure 5. During collapse, peak accelemtion reached 300 cm/sec2 at surface ground zero and feU to 20 em/sec'- at station 5, at a distance of 1.1 km. Peak velocities were 7 and 0.5 em/sec, respectively. The collapse was observed at regional seismic stations, the most distant just over 900 Ian away. Coda lengths of 150 sec measured at stations EYMN and TBO, 202 and 213 Jon distant, respectively (Fig. 1), yielded a body-wave magnitude (mz,z.) of 3.1 for the collapse event. using a scale developed for New England (Chaplin et ol., 1980). We consider this value to be an upper bound because L1-coda attenuation is slightly higher in New England than in the north-central United States (Singh and Herrmann, 1983). Using L, data from EYMN, a value of 2.8 was obtained (H. Patton, personal comm., 1997), consistent with the upperbound mb/4 from the coda durations. It is instructive to estimate the mass of falling material that would produce an m61.4 2.8 collapse event From moment-tensor studies, Yang et al. (I 998) note a free--fall time, or intervnl between fracture and impact phases of the cotlapse, of 0.6 sec. This implies the free-fall distance is 1.8 m. Room heights are 3 m; the shorter fall results from the extra volume occupied by pillar rubble and bulking of fractured material from the mine back (roof). Using an m,4 of 2.8 and the free-fall distance of 1.8 m, we estimated the mass of falling material to be 5.7 X 107 kg, following the method of Taylor (1994). Taking the full panel area (2 X tot m1 ) as an estimate of the collapsed area, a density of 2.6 gmlcm3 and assuming a uniform thickness of the displaced materinl gave a thickness of I m. Clearly, confinement of collapse to a smaller IIJ'C8 would result in a proportional increase in the thickness. These estimates must be considered approximate. Aftershocks 0 We observed intense aftershock activity following the collapse. The first hour of data from surface ground zero (station 13) is shown in Figure 6. We counted just over 4000 events in 15 hr at this station. Event rate approached 1701 min at 6 min and fell to under 4/min in 2 hr. The rapid decay in event mtc points out the difficulty of pcrfonning an effective aftershock study if mobilization were to have occurred in response to the collapse. The aftershock ntte fit a modified Omori law (Utsu et aL, 1995): ntte = 560 (time - 0.0 l) - •·3, with time in hours (Fig. 7). The fit was applied to data between 6 min and 15 hr after the coUapse. For times less than 6 min, counting was incomplete, and for times greater than 15 hr. an instrument malfunction increased noise levels and biased the counting. The decay factor of 1.3 falls within the range obtained for eanhquakes. The subtractive, time-offset term of O.ot is Wtusual but may not be significantly different from zero. Aftershock Soun:e Characteristics The aftershock sequence produced a variety of signals. Most events generated high frequencies (up to 100Hz) and both compressional and dilatational P-wave urrivals. A second class of events generated only dilatational P waves, often of much lower frequency (10 Hz); a third class, only compressional P waves. We classified events based on these polarity patterns as shear slip, implosional, ood tensile, respectively. Jmplosional events may include closing crack mechanisms. Combined mechanisms may also be important (Wong and McGarr, 1990); bowever, some studies have shown mine events tend to be dominated by one style of defonnation, be it volwne change or shear slip (McGarr, 1992b). To classify an event as implosional or tensile, we required that many stations trigger and collect data (most often eight stations or more, the weakest triggered as few as five) and that aU unambiguously determined polarities be the same. To classify an event as shear slip, fewer stations were required, as long as both compressional and dilatational polarities could be determined. Only the largest and highest signal-to-noise events qualified for classification. We identified 41 events as shear slip based on P-wave polarity patterns. Data from a large shear-slip event is shown in Figure 8. This event will be referred to as event A 1 in later text and figures. For events classified as shear slip, compressional arrivals were most often observed at stations closest to ground zero and mixed polaritie.c; at intermediate distances. 1bcse observations indicate thrust-type, shear-slip motion. The shear-slip events could be further subdivided based on polarity patterns. The largest group generated upward motions at stations 1, 2, and 13, downward at station 7. Event Al is a member of this group. A second group of events produced the same pattern except for downward motions at station I and an occasional upward motion at station 7. We choose a large event (2A6:22:58:28.316 UI'M) to represent the second group (not shown); this event is referred to as event A2. Events AI and A2 are used later to determine focal mechanisms for the two groups. A total of 18 event.~ were large enough to be observed at most stations and exhibited all dilatational motions, consistent with an implosional source mechanism. These events also produced significant S-wavc anivnls. Eight of these events gcncntted distinctive, low-frequency (10Hz) waveforms; an example is shown in Figure 9. This event is referred to as event B in later text and figures. Only two large events were found that generated all compressional arrivals, indicating expanding source mechanisms, most likely tensile Ajurshocks of an Explosively Induced Mine Collapse at lVhite Pine. Michigan 1579 Figure 6. Station 13, vertical-component seismograms covering the first hour following the collapse. The amplitude scale is fixed, causing Jarger even IS to be off scale. ~ ~~=I 11o4 c: ::0 2':' ~ 103 ~ ~ 0 - ,..,.c:.. ~ -e i I ~ lii1o2 1 "' 10 t ~ II) ~ !i1o0 a: .= c Time from Collapse (hoUrs) Figure 7. Aftcrsboclc: occurrence rate (circles) and the modified Omori law fit (line). The rms amplitude of each aftershock is also plmted {sma11 symbols) on an arbiltllly scale to show the time nmge over which aftershock co\lllting i5 complete. 0 failure. Clearly, counting may be less complete for events classified as implosional or rensile because of the more stringent requirements we bad to impose. Differences between events Al (shear-slip) and B (implosionaJ) are apparent in their swfacc-ground-zero, P-wave spectra (fig. 10). Spectra were taken from 0.5-sec segments ofstation 13, instrument-corrected, vertical-component data, after applying a Hanning tapec centered on the P-wave pulse. We corrected for the free surface by dividing by 2. No Q eonection was applied because of tbe short travel distauces and relatively competent material. The implosiona1 event generated a comer frequency of 8 Hz and a high-frequency, c.o- 2 decay. The comer frequency for the shear-slip event wD.S more difficult to estimate. We adopted a comer of 60 Hz; however, a comer between I0 and 20 Hz could also be argued. We choose lbe higher comer because it is more representative of the frequency content of the initial ground motion (Fig. 8). Such double-comer events are not uncommon in the White Pine data set and may represent a sum of high- and low-comer or shear and implosional source types. We also see many high-comer events that follow the more classical, ro - 2 shape. Finally, the two tensile events generated high-frequency signals (not shown), similar to the shear-slip events. To further investigate aftershock source types, we assigned corner frequencies (/0 ) and low-frequency asymptotes (flo) by inspection for station 13 P waves from 201 large events, in the manner previously described. The 201event data set contains all events reconled by a minimum of three stations as specified by the requirements of the location studies. Spectra were analyzed in a blind manner, wilbout knowing event classification, to avoid biasing lbe results. Assuming a shear-slip soun:e, moment (M0 ) and source mdii (R) were then calculated using standard techniques (e.g., Hanks and Wyss, 1972): M0 R = 4 npCJ r flo I h, = 1.97 c I 2nfo. where p is density (2.6 glcm3), c is P-wave velocity (3.8 kml sec), r is source-receiver distance (300 m), and his an average radiation coefficient of 0.39 for P waves (Spottiswoode and McGarr, 1975; Boorc and Boatwright, 1984). Results sbow moments ranging from 1015 to 1017 dyne-em for shear-slip events (Fig. II), or moment magnitudes (Mw) of - 0.8 to 0.5 (Hanks and Kanamori, 1970). Event A I, the large shear-slip event shown in Figure 8, has an Mw of 0.3, over two orders smaller than the magnitude (m6LI) of the collapse itself. Source radii estimates range from 10 to SO m 1580 W. S. Pbillips, D. C Pemon, X. Yang, and B. W. Stump 0 Figure 8. Vertical, radial, llld tangential Oeft to right) components of ground velocity for a large shear-slip lftcnhoc:k (M. 0.3), plourd vemas cpicentral distance. Ea:b li8CI: is scaled to its maximum amplitude. Ttme is relative to Ulc ori&in lime determined from the event locaaon (246:22:00:13.414). Dubed linea reprcaent predicted P· and S-wave arrival limes. Stations are indicated next to each trace. This event is refemd to as event AI in lelt aDd later figures. 0 Vertical I i E1·~~~~~~ -jo. ~~~~ i 4 ~ • i . i 0 g o. v~:::!lfl-~·F'Q. ..,.,~""" lwo. 0 Figure 9. Seismograms a:cmerated by a 1arJe implosional aftcnbock, origin time 246:23:45:59.983, as in Figure 8. Tbia event is referred to as evCDt B In te.xt and later figures. Aft~nhock.r ofan 0 Explosively lntlucetl Mine CAJI/apse at White PiM, Michigan 1581 Comer Frequency (Hz) I ·~:....------ Ill (:'--- tOO II l j• 17 1! 10 100 Fllquan:y (HZ) I te :;? j •1.0 100 Source Radius (m) 10 100 ~(HZ) Figure 1 J. Momc:at and M., versus source radius and comer frequcocy from P·wave spectra of after. sbocb. Symbols represent event types as shown in the legeud. Shear-slip events AI and A2 and implosiollal event B lltC labeled. Lines of coostaot stress drop an: included. Figure 10. Diaplacemcnt spectra from slalion 13, vertical-component P waves for sbcat·slip event AI 0 (top) and implosional event B (bottom). Ughter lines iudicue noise spectra obcaioed wilb the slrbe window length and shape, as described in 1be text. Dasbed lines indicate cv- z slope. for shear-slip events, with most radii falling below 30 m. Source radii calculaled using this so~ model can be overestimated, based on studies comparing with independent measurements in mines (Gibowicz et ol., 1990). Sb'esS drops (Aa) were calculated based on the circular crack model of Brune (1970), Aa = 7/16 M0 I !e. 0 Lines of constant stress drop are included in Figure J 1. Clearly, source parameters for nonshear events could be reevaluated using more appropriate models; however, the stress drops as calculated will be useful for classification pwposes as they reOect the relationship between moment and source dimension in one parameter. Stress drops are low, 0.1 to 1 bar for the sbear-slip events, typical of many mine· related events (Gibowicz et oL, 1990), but an order of magnitude larger tban tbe implosional events, revealing a nice separation between event types as classified. Past studies have shown that stress drops can separate sbear-sJip and isotropic (explosion) events (Cong et al., 1996), although the sense of separation is opposite to that observed here. In this study, the small stress drops for implosional events partly result from their large soun:e dimensions (roughly 100m), reflecting possible involvement of the collapsed cavity in the source process. In Figure 11. a number of unclassified events fall in the low stress-drop range, suggesting larger numbers of implo· sional events may have occurred than were originally classified. One implosionaJ event falls with the shear-slip events. This event generated a prominent double-comer spectrum, implying a combined sbcac-implosional mechanism as described earlier; the higher corner was chosen during processing. For the smaller shear-slip events, we see a hint of departure from constant-stress-drop scaling at high frequencies. This is an artifact caused by the anti-aliasing filter limiting comer frequencies for small events. The sequence of aftershocks shows temporal clustering to be common in the White Pine data set (Fig. 12). Clusters often contain only one event type. The first cluster at 1 min contains five shear-slip events and a number of unclassified events of similar stress drop. At 5 min, an isolated implosional cluster of four events occurs. Later in the sequence, we see more clusters of shear-slip events. However, a cluster of implosiooal and shear events initiates at 6 min, while an implosionaJ event accompanied by five unclassified events of similar stress drop, as well as a number of shear events, occurs at 2 br. The tendency toward temporal grouping of simi)ar stress-drop events, including the unclassified ones, supports extending the source-type classification based on stress drop, as discussed earlier. 1582 W. S. Pbillips, D. C. Pearson, X. Yang, and B. W. Stump 10 0 ..:- ! ~ 0 .. 0.1 I tJi aAt • d ••• •• •• ....•• 0 • • •• 00 • • 0 • t .. . • 0 OhplcMional •Sh..,....lp OTentlle II I 0 • . •• 0 :0 \ .... .. 8 0.01 ......••·... . •• • • ,.~·-· ~ .. • : tt. . oa•. 0 0 .. 8 l 0 • UnclaMifted 0.001 -+-----~----~----""'T"-----l 0.01 0. 1 10 100 Time from Collapse (hr) Aftershock Location Procedures We next prepared to locate the White Pine aftershocks by collecting P- and S-wave arrival times. For all events that were detected by three or more stations, arrival times were determined manually from a display of lbe vertical, radial, and transverse components of motion. Radial and ttansVerse 0 0 componenL~ were obtained by rotating to the din:ction of ground zero. Five arrival-time quality levels {0-4) were also assigned at this stage: exceUent, acceptable, fair, poor, and guess; only excellent and acceptable arrivals were later used to locate events. Anival-time data quality decayed rapidly wi1h distana: from ground zero. For the innermost three stations (withio 300 m of ground zero), the chance of obtaining a useable arrival time was 70% or so, depending on the specific station and phase. Data rates dropped to 30% at intermediate distance stations (300-600 m) and to 10% or so at the outermost stations (over 1000 m). 'These percentages were calculated with respect to the number of lbrcc-station events, as pre-viously defined. On average, such events yielded four P- and four S-wave arrivals, giving a total of eight arrival times available for location. To calibrate the White Pine site, we developed P- and S-wave velocity models of the subsurface and calculated Pand S-wave station corrections. 1bc station corrections accounted for lateral variations in the layer depths, which are known to be significant, as well as variations in velocities and in the thickness of tbe glacial till layer at the surface. F'trst, we set up a 320-m-thick layer-over-a-half-space model, obtaining tbe layer P-wave velocity from a nearby refraction survey (Geosphere, 199S). The layer contained Nonsuch and Freda layers, the half-space contained the Copper Harbor conglomerate. Our intention was to locate aftershocks relative to the known position of the first pillar shol As pointed out, recordings of lhe pillar shots were of poor signal to noise, and we could determine reliable first breaks at only the closest four stations (1, 2, 7, and 13). In addition. a reliable zero time was not available due to the use of a 50 Figure 12. Inferred stress drop versus time of oa:urrcnc:e in the aftmbock sequence. Symbols represent event types as shown in the legend. Sbear-s)jp CVCIIt& At and A2 and implosional event 8 arc labeled. msec electronic delay cap ahead of the first shot, which could not be precisely accounted for. To overcome these problems, we first determined station corrections needed to correctly locate the initial pillar shot, using P-wave arrival times from the four closest stations. We then located two well-recorded aftershocks using P-wave anivals and corrections at the same stations. Arrivals from the located aftershocks allowed us to determine the remaining P-and S-wave velocities by inspection of travel-time plots. Following this, station corrections were determined by averaging travel-time residuals for the two events. The final velocities were 3.80 km/scc (P) and 1.60 kmlscc (S) for lbe Freda-Nonsucb layer, and 5.46 km/sec (P) and 3.07 km/5CC (S) for the Copper Harbor haJfspace. The contrast between Freda-Nonsucb and Copper Harbor is dramatic. reflecting an increase in rock competency and strength across the interface. The station corrections ranged from -18 to 20 mscc for P waves and from -41 to 29 msec for S waves (Table 1). The station corrections are consistent with known structure; for eXJIJilple, the early arrivals at station 6 result from a shallowing interface in that direction. This structural information was taken from a three-dimensional model of the mine on display at the White Pine headquarters. We obtained microearthquake locations using an iterative, damped-least-squares (Geiger's) method, employing the velocity model and station corrections described earlier. Arrival-time data were weighted by lfl', where T represents data error estimates of 1.5 msec for P waves and S msec for S waves. These values matched nns arrival-time residuals reasonably weD. Because initial results occasionally yielded large residuals, we added a reweighting scheme (Scales et al., 1988), which approximates the minimum L1 nonn solution. The location calculation also included an estimate of tbe standard location-error ellipsoid, using T as estimates of the data error. For selected groups of events that generate similar waveforms, we can obtain more precise relative locations. 'Ibis is done by choosing arrival times at the same point in A?ershockr of an Explosiveb' Induced Mine Collapse a: White Pine. Michigan the waveform for all similar events. then adjusting station corrections to hold the position of a well-recorded. master event ?xed to its original location. Based on residuals, we estimate that data errors improve to and 2 roses for and waves, respectively. resulting in precise locations relative to the master event. Relative location studies have proven useful for understanding joint structures along which geo- thermally induced microearthquakes occurred (Phillips et aL, 1997) and have been used successfully to identify failure surfaces in mining-induced seismic data {Spottiswoode and Milev, I998). Aftershock Location Results Map and cross-section views of 135 aftershock loca- tions are shown in Figtn'e 13. These aftershocks were re- quired to have six or more arrival times and magnitude of the major error-ellipsoid axis less than 20 m. The map view shows a distribution of aftershocks con- centrated along the western edge of the collapsed panel open to the mine. Activity is absent within 50 of most unmined faces. The cross-section views show an active zone just on- 1583 der thick, bottoming at mine level. A few events lo- cate 20 or so below mine level and may be mislocated. In the east?west cross section. events de?ne a triangular prism with upper edges dipping 45? to the east and nearly vertical to the west. Over 90% of the locatable events oc- curred within 2 hr of the collapse. The latter 10% occurred during isolated swarms of activity through the remaining 36 hr of network operation. The major axes of the standard (one-o) location-error ellipsoids pointed in northerly. near- horizontal directions, with a scatter of 3:35? and an average length of 10 m. The predotninanoe of north-trending major axes results from an east-west distribution of the closer-in stations (Fig. 3). Error-ellipsoid aspect ratios averaged 1.6 (majorlintermediate) and 2.6 (major/minor). Event type and stress drop strongly correlate with lo? cation depth (Fig. l4). lmplosional events are found over a narrow depd: range just above mine level, while shear-slip events occur throughout the active depth range and dominate at shallow depths. Unclassi?ed events follow the same, over- all pattern. In particular, only high stress-drop events, cor- responding to shear slip, are found at shallow depths. This 0.1 0.2 0.3 I 95.1 I I 948 949 950 Easting (km) Depth (km) I - 561.4 - as: 3 at .530.1 0.2 0.3 Depth (km) Figure 13. Map and cross-section views of alter-shock locations. Events Al. A2. and are plotted using largercireles as indicated in the map view. The annotated The mine and collapsed panel are panel is shown to scale in the moss-section views. Close-in stations 2 and 13 are indicated triangle in the cross-section view looking north used in modeling post-collapse stress redistri by triangles in map view. The dotted the assumed destressed zone button. 1584 0 W. S. Phillips, D. C. Pearson, X. Yang, DDd B. W. Stump ~~~----------------------------~ • ... • ... :. .C : .. ~·· ·a · • ~·· • a J/1 • .., •• • )II ... llo AI • • •.., r•· • ao •o• a.o ~ ;,• • 0 ... • cP·o: .. o•:.•~' • ·" ••• ·o • •. • • o;~~-o