DEVELOPMENT OF REGRESSION MODEL BASED ON STRONG MOTION DATA

Abstract

One of the crucial requirements for the estimation of the seismic hazard of an area is the attenuation relation and these relations are derived generally from the regression analysis. The coefficients of regression relation are derived from regional strong motion data. Selection of regression model stands as the primary step in regression I analysis and the selection of the model is done on the basis of how the peak ground acceleration depends on the hypocentral distance and the magnitude. Checking for many such functional dependencies are done and the regression model that gives minimum error and the best correlation is selected as the final regression model. The area of study forms the Garhwal and Kumaon regions of the Uttarakhand Himalayas within latitude 29°-33°N and longitude 78°-81°E, which are considered as important regions seismotectonically. Bulk of the earthquakes originate generally at depths approximating 10-15 km. Strong motion data is collected for these regions and formulated in the regression model by keeping the value of IV' as R+x where x varies from 10 to 14 corresponding to the depths. On successful execution of the program in FORTRAN, we get the regression coefficients for various models and the model R+14 gave the best correlation and minimum error. Hence, this is the new regression model which is selected and an attenuation relation is formulated for both Kumaon and Garhwal l-Iimalayan regions.