NH23A-1864
High-Resolution Tsunami Inundation Simulations Based on Accurate Estimations of Coastal Waveforms

Tuesday, 15 December 2015
Poster Hall (Moscone South)
Yusuke Oishi1, Fumihiko Imamura2, Daisuke Sugawara2 and Takashi Furumura3, (1)Fujitsu Laboratories LTD., Kawasaki, Japan, (2)International Research Institute of Disaster Science, Tohoku University, Sendai, Japan, (3)ERI, Univ. Tokyo, Bunkyo-Ku, Japan
Abstract:
We evaluate the accuracy of high-resolution tsunami inundation simulations in detail using the actual observational data of the 2011 Tohoku-Oki earthquake (Mw9.0) and investigate the methodologies to improve the simulation accuracy.

Due to the recent development of parallel computing technologies, high-resolution tsunami inundation simulations are conducted more commonly than before. To evaluate how accurately these simulations can reproduce inundation processes, we test several types of simulation configurations on a parallel computer, where we can utilize the observational data (e.g., offshore and coastal waveforms and inundation properties) that are recorded during the Tohoku-Oki earthquake.

Before discussing the accuracy of inundation processes on land, the incident waves at coastal sites must be accurately estimated. However, for megathrust earthquakes, it is difficult to find the tsunami source that can provide accurate estimations of tsunami waveforms at every coastal site because of the complex spatiotemporal distribution of the source and the limitation of observation. To overcome this issue, we employ a site-specific source inversion approach that increases the estimation accuracy within a specific coastal site by applying appropriate weighting to the observational data in the inversion process.

We applied our source inversion technique to the Tohoku tsunami and conducted inundation simulations using 5-m resolution digital elevation model data (DEM) for the coastal area around Miyako Bay and Sendai Bay. The estimated waveforms at the coastal wave gauges of these bays successfully agree with the observed waveforms. However, the simulations overestimate the inundation extent indicating the necessity to improve the inundation model. We find that the value of Manning's roughness coefficient should be modified from the often-used value of n = 0.025 to n = 0.033 to obtain proper results at both cities.

In this presentation, the simulation results with several different model configurations will be introduced including the results from the models that incorporate building shape data. This will be followed by a discussion on how to improve the accuracy of the high-resolution inundation simulations.