Empirical Analysis Of A 3-D Pakistan Regionalized Velocity Model For Improved Seismic Event Locations

摘要

The objective of this project is to improve regional event location in Pakistan and the surrounding regions by developing a 3-D velocity model for the crust and upper mantle. The CTBT goal of 1000 km2 location accuracy for small events is dependent upon the availability of such models, which can be used to compute accurate travel times of regional seismic phases. Our approach is to refine an a priori 3-D velocity model through joint nonlinear tomography and multiple event location applied to regional earthquake arrival time data. Previously, we presented our a priori model, which is constructed on a 1 deg-by- 1 deg grid to a depth of approximately 100 km. The preliminary model exhibits complex structural variation in the Pakistan region with crustal thickness ranging from 25 km to 75 km, a clear deviation from 1-D global models such as the IASP91 model (Kennett and Engdahl, 1991). We have generated 3-D P, Pg and Pn travel time tables at regional distances by applying the finite-difference raytracing method of Podvin and Lecomte (1991) to our a priori model. To test our model empirically, we have relocated a suite of 44 Calibration Event Bulletin events using P and Pn arrival times from 14 stations in the Pakistan region, reported to the International Seismological Centre (ISC). Locations obtained with the IASP91 travel time tables were compared to locations obtained with our 3-D travel time tables. While our 3-D model results in smaller rms residuals for 29 of 44 events, the tests are not definitive since ground-truth locations are not available and the depths of the events are poorly constrained by the regional data. We are currently expanding our database to include 55,000 phase arrivals from over 4000 events recorded at 77 stations in the study region, which we are expanding to include additional parts of southwest India.