Abstract: A microseismic event often contains some abnormal channel signals, strong noise interference signals and weak signals. The picking accuracy of these channel signals’ P-wave phase arrivals is often low, and even those P-picks are false. Currently, the invalid P-picks are not screened out before automatic source location in microseismic monitoring system, and the manual intervention need to be carried out. To solve this difficult problem, a modified Akaike information criterion (AIC) picker was proposed firstly. Secondly, the quantitative relationship between the P-picks accuracy and the source location accuracy was analyzed based on a microseismic monitoring case of a deeply buried tunnel, and then a new concept of P-pick admissible error was introduced. Furthermore, all kinds of microseismic signals’ P-wave phases were picked up by the modified AIC picker, and the corresponding waveform parameters were calculated. A great amount of statistic work has revealed the relationship between P-pick accuracy and waveform parameters. Finally, the P-picks were classified into valid group (tagged 1) and invalid group (tagged -1) based on the admissible error. And support vector machine (SVM) classifier was applied to build a forecasting model for identifying valid and invalid P-picks. Practical application shows that the invalid P-picks are identified correctly and effectively by this quality control method. The accuracy of microseismic source location is improved greatly after eliminating the invalid pickups.

Key words: P-wave phase identification, AIC picker, waveform parameters, quality control, source location