Bias in the centroid moment tensor for central Asian earthquakes: Evidence from regional surface wave data

Surface waves recorded on the Chinese Digital Seismic Network (CDSN) station Urumqi (WMQ) are analyzed for 26 centroid moment tensor (CMT) earthquakes located within an epicentral distance of 1300 km. Applying a simple grid search method for source parameter retrieval, I obtain revised estimates of the focal depth, fault strike, and seismic moment (M-0) from amplitude spectra of Love and Rayleigh waves and compare these estimates with the CMT best double-couple solutions. The comparisons show that residuals are unbiased for focal depth and fault strike, where population spreads are +/-13 km and +/-17 degrees, respectively. On the other hand, M-0 estimates are found to be biased, with CMT M-0 larger than regional estimates by an average of 0.27 +/- 0.04 log units. The results for focal depth and strike are consistent with previous global and regional studies comparing independent estimates with CMT results. Recent full-waveform modeling studies for central Asian earthquakes support the findings of bias in M-0 estimates. I suggest that causes for M-0 bias may be related to great thicknesses of continental crust in Hindu Kush and Tien Shan regions and to data censoring practices at small magnitudes. M-0 : m(b) scaling relationships for central Asian earthquakes show better agreement with western U.S. scaling when M-0 estimates determined in this study are used.