Lg-Wave Cross Correlation and Epicentral Double-Difference Relative Locations in China

Thursday, 18 December 2014
David Paul Schaff1, Paul G Richards2, Megan Slinkard3, Stephen Heck3 and Christopher J Young3, (1)Lamont-Doherty Earth Observatory, Columbia University, Palisades, NY, United States, (2)Columbia University, Palisades, NY, United States, (3)Sandia National Laboratories, Albuquerque, NM, United States
In prior work we presented high-resolution locations of 28 events in the 1999 Xiuyan sequence in China using only cross correlation measurements of Lg-waves and a double-difference technique solving for the epicenter. Only five regional stations 500 to 1000 km away were used. The resulting locations revealed a 4 km stretch of fault with 95% location errors ~150 m and 7 ms residuals from internal consistency. Based on this success, we now attempt to extend this work on a broader scale to all of China to see if similar results can be obtained for a significant fraction of the seismicity. We first examine in detail three other clusters. The first has 9 events, and has relocations in a 1 km box with location errors on the order of tens of m. The second has 10 events, and relocates in a 10 km box with location errors on the order of hundreds of m. The third cluster, relocates in a 25 km box, and has ~1 km location errors. From this we see that the location errors increase with increasing event separation. The Annual Bulletin of Chinese Earthquakes we have determined from a repeating event catalog has average errors of 16 km. Therefore we are able to demonstrate one to two orders of magnitude improvement in the location errors as compared to the bulletin. Then we apply a pair-wise location procedure for the repeating event catalog we have identified for China in earlier work. 13% of the events are classified as repeating events (2,379 out of 17,898). Of these there are 1,123 events (1,710 pairs and 6% of the catalog) that have two or more stations from which we can estimate the locations by the same procedure. 81% of these events are demonstrated to have locations separated by less than 1 km with another event. There are 677 events in the repeating event catalog that have observations at four or more stations which enables an estimate of the location errors for these high quality events to be about 200 m.