Simulation-based PSHA for the Kathmandu Valley: sensitivity to hypocentre randomisation

Raffaele De Risi, Shaoqing Wang, Max Werner, Flavia De Luca, Paul J Vardanega, Rama Pokhrel, Prem Maskey, Anastasios Sextos

Research output: Contribution to conferenceConference Paper

150 Downloads (Pure)


This paper presents a sensitivity study of simulation-based Probabilistic Seismic Hazard Analysis (PSHA) for Kathmandu, Nepal. Two aspects are investigated in detail: (i) the technique for simulating fault ruptures compatible with scaling laws and fitting into the Main Himalayan Trust (MHT) and (ii) the choice of different Ground Motion Prediction Equations (GMPEs). Since the 2015 Gorkha earthquake, a number of new studies have provided new approaches to model the MHT as a single seismic source. This more realistic characterization of the MHT has resulted in higher peak ground acceleration (PGA) for all of Nepal and in particular for the Kathmandu Valley. Here, the results of a new simulation-based code are compared with those by the software OpenQuake. It is specifically assessed how different source simulation methods influence the hazard calculations. Furthermore, the influence of different GMPEs assumed for subduction earthquakes is assessed for the specific case of Kathmandu, which is only 11 km above the MHT. Results show that because of the proximity to the megathrust, the estimated hazard is very sensitive to different choices.
The results of this study may be useful for informing ongoing efforts in Nepal to update the building code with a new seismic hazard map. Understanding the sensitivity and robustness of PSHA is critical from a policy and preparedness perspective.
Original languageEnglish
Number of pages9
Publication statusPublished - 18 Sept 2020
EventThe 17th World Conference on Earthquake Engineering - Sendai, Japan
Duration: 27 Sept 20212 Oct 2021


ConferenceThe 17th World Conference on Earthquake Engineering


Dive into the research topics of 'Simulation-based PSHA for the Kathmandu Valley: sensitivity to hypocentre randomisation'. Together they form a unique fingerprint.

Cite this