A moment tensor inversion approach based on the correlation between defined functions and waveforms

Abstract The moment tensor inversion is a commonly-used method to interpret source mechanisms of microseismicity. In the inversion for real data (e.g. microseismicity recorded during hydraulic fracturing), the waveforms recorded by sensors can be the mixtures of signals and noise, or the superposition of signals generated by multiple sources. Then the traditional approach may result in inaccurate solutions. In this article, we developed a new inversion approach based on the correlation between waveforms and correlation functions, which are defined based on the characteristics of signals. The correlation function determined by specific parameters is more sensitive to the signal generated by a specific source, and less sensitive to noise or signals generated by other sources. Then the correlation coefficient calculated by multiplying the waveform and correlation function are mainly determined by the signal. The moment-tensor solutions calculated by the correlation coefficients are more accurate. The new inversion approach was evaluated by synthetic tests. For noise filtering, compared with tradition inversion approaches, the new approach can improve the inversion accuracy by more than 50% at various noise levels. For multiple sources discrimination, the new approach can discriminate signals generated by multiple sources and provide more accurate inversion results for the sources simultaneously, but the application of the method is limited. This new inversion approach aims to provide accurate solutions in a very simple way, when the waveforms are distorted.