[4K3-J-13] AI application: sea and rivers

[4K3-J-13-04] Maximum tsunami height prediction using Ocean-Bottom Pressure Values based on Gaussian Process Regression

〇Kotaro Takahashi¹, Kenjiro Kashiwabara¹, Yasuhiko Igarashi^1,2, Toshitaka Baba³, Hori Takane⁴, Masato Okada¹ (1. Graduate School of Frontier Sciences, The University of Tokyo, 2. Japan Science and Technology Agency PRESTO, 3. Institute of Technology and Science, The University of Tokushima, 4. Research and Development Center for Earthquake and Tsunami, Japan Agency for Marine-Earth Science and Technology)

Keywords:Gaussian Process, Machine Learning, Disaster prevention, tsunami height prediction

Tsunami early warning systems using water pressure gauges operate around the world to cope with damage caused by tsunami waves. The systems use a correlation between observed pressure gauges value around coast and tsunami height at prediction points near shore, because tsunami height basically depends on the topography (bathymetry) during its propagation.
In predicting the tsunami height, it is important both avoiding underestimation and increasing accuracy in order to minimize the damage. The conventional method selects an scenario which has the largest tsunami height of near the observed pressure gauges value in the tsunami database that contains pre-computed tsunamis offshore and nearshore from 1506 earthquake scenarios. Although this conventional method can avoid under estimation, it puts the prediction accuracy second.
In this study, we extended tsunami height prediction method using Gaussian Process regression and proposed a prediction method with less underestimation and higher accuracy. We investigate the prediction accuracy and the possibility of underestimation by our proposed method that uses pressure gauges data from the Dense Ocean-floor Network System for Earthquakes and Tsunamis (DONET) in the Nankai trough.