The surface monitoring microseismic data is characterized by low S/N，inaccurate first-break picking-up，difficult
to establish accurate velocity model.Meanwhile，surface monitoring microseismic data is measured by multi survey
lines and its shallow surface layers have complex strata velocity.In order to solve the above problems，the Bayesian
inversion method for the source positioning of surface monitoring microseismic data was probed.We assume the
inversion results of all lines as a total probability and the inversion of every line as a classification，and then discuss
the Bayesian maximum posterior method for source position inversion.During inversion，lateral velocity-variable
model is adopted in shallow layers and horizon lateral even velocity model in middle-deep layers.The theoretical
models are fitted on posterior probability density function of objective function，weighted function and velocity
parameter variance.The fitting results are regarded as estimation probability density.Taking the hybrid algorithm of
extremely fast simulated annealing grid method as searching method，and grid algorithm as guide to make
searching fall into the convex interval of the optimal solution.Then，extremely fast simulated annealing algorithm
is used to search the optimal solution.This is to prevent the algorithm converges to a local extreme point，and
greatly improve the speed of convergence of the algorithm.The application results on theoretical model and actual
data indicate the validation of the method that is to guarantee the accuracy of first-break inversion with large
random error.
论文详情