On Computational Costs for Monte Carlo Regional-Scale Simulation for Reliable Estimate of City and Urban Area Earthquake Disaster

Abstract

Regional-scale simulation in high spatial and temporal resolutions is being developed for a more reliable estimate of earthquake disasters in city and urban areas. Monte Carlo simulation is used to account for various uncertainties in analysis models. Since one regional-scale simulation needs a large amount of computation, it is important to estimate computational costs for the Monte Carlo simulation. In this paper, we first clarify uncertainties involved in the simulation of ground motion, structural seismic response, traffic flow, and economic recoveries. We then evaluate computational costs that are required for the reliable estimate using Tokyo Metropolis Earthquake as an example. The reliability of the Monte Carlo regional-scale simulation is discussed, considering the future trends of digital information about city and urban areas and monetized computational cost as well as the uncertainty of an earthquake scenario which is input to the regional-scale earthquake simulation.