Fast one-dimensional velocity model determination using station-pair differential times based on the differential evolution method in microseismic monitoring

Jiewen Zhang*,Haijiang Zhang,Jianyuan Yu,Qiang Liu

DOI: https://doi.org/10.1190/segam2014-1137.1

2014-01-01

Abstract:Summary For accurately locating microseismic events, it is crucial to have an accurate velocity model. A common way to determine the velocity model is to use the grid search method to find the optimal model to fit the arrival times from perforation shots. However, the accuracy of the velocity model depends on the model search range and the search interval as well as the number of layers in the case of the one-dimensional model. It will be extremely timeconsuming if there are many layers and the search interval is small for a reasonable search range. In our study, a fast and robust global optimization method, known as the differential evolution (DE), is used to speed up the search process. The objective function for finding the optimal velocity model is related to the minimization of station-pair travel time residuals, which has the advantage of removing the effect of uncertain origin times of perforation shots. Considering that the method evolves from the Monte Carlo method, the randomness it brings to the inversion results cannot be neglected. Thus the final velocity model is given as the weighted mean of the 50 models with weighting parameters inverse proportionally to the corresponding objective function values. We tested this method on a synthetic dataset and showed its effectiveness.