The 1934 Bihar-Nepal Earthquake – Simulation of Broadband Ground Motions and Estimation of Site Amplification

Abstract

The 1934 Bihar-Nepal earthquake, one of the most catastrophic events ever to occur in the Himalayas, inflicted extensive devastation with reported MMI of IX-VI in the Kathmandu valley and the Indo-Gangetic (IG) basin. The earthquake triggered significant ground liquefaction and landslides as it occurred in the proximity of densely populated river basins causing a huge economic loss and over 15700 fatalities. However, it is unfortunate that there are no ground motion data available for the event, as it remained unrecorded due to a lack of instrumentation. Therefore, simulating ground motions for the 1934 Bihar-Nepal earthquake would provide new insights into the influence of regional characteristics on Himalayan earthquakes. However, incorporating the Himalayan topography and the IG basin in the ground motion simulation is very challenging. In contrast, proper validation of modeling of ground motions is difficult due to the unavailability of recorded data. To circumvent these challenges, we simulated broadband ground motions for the 2015 Nepal earthquake, another significant catastrophe that occurred in the same seismo-tectonic region in the Himalayas which provides a well-recorded database. For the 2015 Nepal earthquake, a thorough comparison of the recorded and simulated ground motion spectra reveals that the simulated ground motions are consistent with the recorded data in terms of amplitude, strong motion duration, and spectral ordinates. Therefore, we considered the same medium characteristics to simulate broadband seismograms for the 1934 Bihar-Nepal earthquake by combining deterministically generated low-frequency (LF) and stochastically simulated high-frequency (HF) ground motions. The HF accelerograms are generated by considering incident and azimuthal angles obtained from rays of P and S waves traced from the finite fault slip model to the station, passing through the regional layered stratified velocity model, free surface factors and energy partition factors (Otarola and Ruiz, 2016). For deterministic simulation, a 3D computational model (Sreejaya et al., 2022) for the study region of approximately 9°×7° (between 80°–89°E longitude and 23°-30°N latitude), incorporated with basin geometry, material properties, and topography of the region is embedded with the finite fault rupture model of the event to generate LF ground motions. For the finite fault source model, five samples with various spatial variability of the slip on the rupture plane are simulated as a random field (Mai and Beroza, 2000; 2002) using the seismic moment and fault dimensions provided by Pettanati et al. (2017). Ultimately, the broadband (0.01–25 Hz) ground motions are obtained at 6461 hypothetically gridded stations with a 0.1°×0.1° spacing by combining the suitably filtered LF and HF ground motions in the frequency domain with the target frequency of 0.3 Hz with a bandwidth up to 0.05 Hz. A systematic comparison of estimated MMI values (Iyengar and Raghukanth, 2003) and the observed MMI values at 459 sites revealed that the PGA between 0.25-0.6g is significant within 200 km of the epicentral distance. Thus, the results can be used for addressing the ground failure and liquefaction caused due to the earthquake and also find applications in seismic hazard assessment of the cities in the basin.