LAUSR

This work aims at developing a hybrid ground motion prediction equation (GMPE) for spectral acceleration in Western Himalayas and North-Eastern India. The GMPE is derived using an efficient nonparametric modelling based on neural network algorithm. In this study, owing to sparsity in the recorded ground motions (498 recordings) for the region, the available information is combined with 13,294 records from the well-tested NGA-West 2 database. For the methodology adopted in the study, regional flags are assigned to the records. Thus, given a magnitude, distance, shear wave velocity, fault type and region, the model is able to predict the possible spectral acceleration. The developed GMPE is observed to be unbiased with respect to region. Further, the inter- and intra-event standard deviations are also in acceptable ranges. It is observed that developed GMPE for Western Himalayas and North-Eastern India is able to capture all the known ground motion characteristics. Additionally, the GMPE is compared with the existing GMPE for rock-type soil condition available for the Western Himalayas and North-Eastern India. Furthermore, applicability of the developed GMPE model in estimating hazard is analysed by obtaining the uniform hazard response spectra for Delhi and Guwahati.