Estimation of Earthquake Source Parameters Using Machine Learning Techniques

The earthquake source parameters are generally grouped into two categories namely source location and source dimension. For estimating location of an earthquake, generally world average slowness versus distance curves are used to find the distance of the source from the seismic station. These curves are inherently non-linear and a complex function of epicentral distance, earth structure and the path travelled by the seismic waves. For estimating the source dimension, Brunes source model is fitted to the observed displacement spectrum of seismic wave by optimizing spectral parameters. This article proposes employment of machine learning techniques, namely artificial neural network (ANN) and genetic algorithm (GA) to estimate source location and source dimensions respectively. ANN is used as function approximator to learn the relationship between the slowness and distance by using training data whereas GA is used to search the spectral parameters to fit Brunes model with observed displacement spectrum using theoretical constraints in the model space.