Estimation of Hypocentral Paremeters of Local/regional Earthquakes by Grid Search Method With a Fuzzy Logic Approach

Abstract

This study presents improvements to the hypocentral location method based on Grid Search technique with fuzzy logic approach for resolving regional/local earthquake locations. The method is tested on synthetic models to evaluate its effectiveness in various situations, and then, after gaining experience, it is tested on actual earthquake data. For this purpose, three different artificial seismic networks with different station distribution geometries are selected to investigate the effect of the network shape on the solution. Additionally, for all networks, for the three scenarios, i.e. the earthquake's proximity to the networks, whether inside, just outside, or further away, is taken into account. Four types of norms are examined in this study, that is, L₂, L₁ for P wave and L₂ and L₁ for S wave are used, and they are transformed into fuzzy logic space using a half trapezoidal membership function that constructed minimum and maximum RMS values for the all norms. The location determination process was performed in two ways; One is defuzzification of the output of the intersection process on four fuzzy logic output matrix, and other is defuzzification of only all grid points having maximum fuzzy output values of the highest values. The results show that this method for epicentral estimation work well if the depth of an earthquake is shallow enough, according to the distance between the event and seismic network and it has a special advantage in the location of an event far from the network. This method is only applicable to the hypocentral location of the earthquake that happened in the upper crust because it only uses Pg and Sg arrivals on the seismograms and a half-space velocity model so its validity is somehow limited. By using this method, we were able to estimate the hypocenter locations of 151 shallow earthquakes that occurred in the eastern Black Sea and found them to be highly comparable to those obtained by other seismic agencies. The average total difference between most earthquakes is approximately 5 kilometers. Believed that this method works well for local earthquakes taking place within network and for regional earthquakes taking place outside a network and hypocenter-station distances significantly larger than distance between stations i.e. having a great azimuthal gap.