ON THE USE OF KRIGED P-WAVE TRAVEL-TIME CORRECTION SURFACES FOR SEISMIC LOCATION Lee K. Steck, Michael L. Begnaud, and Aaron A. Velasco Los Alamos National Laboratory Sponsored by National Nuclear Security Administration Office of Nonproliferation Research and Engineering Office of Defense Nuclear Nonproliferation Contract No. W-7405-ENG-36 ABSTRACT Accurate location of seismic events remains a critical issue for global nuclear explosion monitoring. Herein we present some observations on the nature of kriged P-wave travel-time correction surfaces and their application to improving seismic event location in China. We have adopted the correction surface approach due to ease of implementation and the fact that empirical correction surfaces can be constructed without detailed knowledge of crustal structure. However, correction surfaces can be built on any velocity model, no matter how detailed, so that when such models do become available, surfaces can be recalculated for that model. This, in turn, enables prediction of corrections in regions lacking seismicity. We use the modified Bayesian kriging method to construct surfaces for 76 stations around Asia, analyzing travel-time data from seismic events in the United States Geological Survey Earthquake Data Reports (EDR) and the International Seismic Center (ISC) catalog. Due to limited data, we gather residuals for events throughout the crust (as defined by the 1-D global model employed) whose location accuracies range from 2-25 km. The correction surfaces are used with the EvLoc algorithm to perform regional relocations of several thousand events in the region around China. Correction surfaces dramatically improve the clustering and linearity of regional seismicity and increase the stability of EvLoc. About 50% more events are successfully relocated when surfaces are used. Comparing regional relocations to high-quality ground truth also reveals a significant quantitative improvement in location accuracy. In an effort to further improve our location ability, we are creating a comprehensive merged database for the China region, comprised of EDR, ISC, Reviewed Event Bulletin, and several regional catalogs. This database will provide the most complete record of arrivals for events in eastern Asia, and its location performance will be validated against current databases. One measure of the robustness of the kriged surfaces is their correlation; nearby stations should have similar P- wave correction surfaces. We find that surface correlation is high for nearby stations but drops off beyond about 250 km, implying that, on average, crustal structure varies rapidly across Asia. This length scale may be useful for assessing whether or not surrogates should be used in developing correction surfaces for new stations. Moreover, this correlation length can also be used to constrain the model correlation length parameter in the kriging procedure. We have performed a suite of sensitivity tests to examine the effect of depth and epicentral mislocations on travel-time residuals. These tests were performed using four closely spaced high-quality ground truth events as observed by the 76 stations for which we calculate correction surfaces. Fixing the latitude and longitude of the events and letting the origin time and depth vary, we find that there is about 0.13-sec deviation in residuals for every 10 km of depth error. Epicentral mislocations result in P-wave residual errors of about 0.75 sec per 0.1 o . When compared to the root-mean-square residual value of about 1.9 sec, effects due to depth errors and depth averaging are minimal. KEY WORDS: location, validation, calibration, regional 282 410