P-Wave Back-Azimuth and Slowness Anomalies Observed by an IMS Seismic Array LZDM

Abstract

The seismic array LZDM in China is a primary station of the International Monitoring System (IMS) for verifying compliance with the Comprehensive Nuclear-Test-Ban Treaty. P -wave back-azimuth and slowness anomalies were found at LZDM during research on teleseismic events; that is, the back-azimuth and slowness errors of the P phases were found to be systematic and varied with theoretical back azimuth in a sinusoidal-like and cosinusoidal-like manner, respectively. The maximum values of back-azimuth and slowness errors reached 87.1° and 8.68 sec/°, respectively. Analysis showed that these errors could not be expressed as measurement errors because they were mostly caused by the dipping (Moho) below LZDM. These large systematic errors were well compensated by introducing several dipping Moho models. A velocity contrast of 0.84 km/sec across the Moho was chosen for these models. The best-fit dipping Moho model for all the data was found to have a 115° strike and a 9° dip angle. Moreover, the events in different slowness domains were adapted to specific submodels, which resulted in smaller average errors. The strikes of all the models were all around the point where the slowness errors change polarity from either positive to negative or from negative to positive. The corrected results for 151 teleseismic events were clearly improved by the introduction of the single dipping Moho model. Error analysis after structural correction showed that the average errors of back-azimuth and slowness residuals drop from 31.5° to 11.6° and from 3.2 sec/° to 1.2 sec/°, respectively; that is, the error reductions of back-azimuth and slowness residuals are 63.3% and 62.5%, respectively. To validate the model, we used another group of 166 events and did the structural corrections. Error analysis after structural correction showed that the average errors of back-azimuth and slowness residuals drop from 30.0° to 13.2° and from 3.5 sec/° to 1.3 sec/°, respectively; that is, the error reductions of back-azimuth and slowness residuals are 55.9% and 63.3%, respectively. Several submodels for different slowness domains have been introduced for the first 151 events group and 21 events from southeast with low slowness values, the error reduction ratios were higher (about 10% in back azimuth and slowness, respectively) than the application of the single Moho model. Error analysis after structural correction showed that the average errors of back-azimuth and slowness residuals drop from 30.8° to 8.9° and from 2.8 sec/° to 0.7 sec/°, respectively. The structural correction results of the second group of 166 events by the using of submodels gave a 9% improvement in the back-azimuth and slowness determination than the single Moho model, which implied the complexity that exists below the array. Error analysis after structural correction showed that the average errors of back-azimuth and slowness residuals drop from 30.0° to 10.6° and from 3.5 sec/° to 1.0 sec/°, respectively.