The Structure-tensor Analysis for Optimal Microseismic Data Partial Stack

Microseismic monitoring of hydrofrac is an actively developing technology utilizing various acquisition arrays. In this paper we consider processing of microseismic data recorded by specific surface network geometry - patch arrays (far separated local receiver groups). The project aim is to produce an optimal partial stacking of the data within patches for improving SNR for microseismic events detection and location. We propose to use a structure-tensor analysis for estimating directions of coherency in the data, which can be used for data stacking for each patch. Unlike to standard slant-stacking method, we do not scan all possible directions, but receive them as eigenvectors of the structure tensor. The corresponding eigenvalues become the attributes that can be used for checking if there are useful waves in the data. We applied the proposed method to synthetic and real data. Within the synthetic data, we considered the presence of random, coherent noise and signals interfering. The testing showed that the structure-tensor partial stacking results are comparable with the standard method and robust to the different noise types. In the paper, we also discussed the usefulness of the structure-tensor attributes for detecting presence of arriving wave in the data and its arrival direction.