First arrival time picking for microseismic data based on shearlet transform

Abstract

Automatic identification and first arrival time picking of microseismic data play an important role in microseismic monitoring technology, and it is the precondition for real-time microseismic hypocenter location. This paper presents a novel first arrival time picking method based on shearlet transform (ST), which aims to get satisfactory results in low signal-to-noise ratio data. The ST is used to decompose noisy microseismic data. By the coefficient differences between the signal and noise at fine scales, the signal points can be preliminarily selected from the noise. To further improve the accuracy of the signal recognition, a correction to the selected signal points is made by utilizing the scale correlation between adjacent scales. The realization of the correction depends on the distances between the signal points at one scale and those at its adjacent scale. After the correction, the moment of the first identified signal point is the first arrival time. This proposed method can produce a superior performance in the accuracy of the first arrival time picking, compared with the other methods, as demonstrated using synthetic and field microseismic data. The actual picking performance of the method is further verified by receiver operating characteristic curves.