An ensemble forecasting method for tsunami warning

Tsunami warning driven by the earthquake rapid report can provide rapid prediction of tsunami threats after an earthquake occurs. However, there is significant uncertainty in the earthquake data obtained immediately after an earthquake occurs, and further information such as slip heterogeneity cannot be obtained in a timely manner. The uncertainty of earthquake rapid report consequently results in significant uncertainty in tsunami forecasting. Thus, an ensemble forecasting method for tsunamis has been proposed to quantitatively estimate uncertainty and tsunami hazard, considering the uncertainty of magnitude, epicenter, and heterogeneous slip. The deviation distribution of magnitude and epicenter in earthquake rapid report is approximately adopted based on the historical data statistics between widely used earthquake datasets. The heterogeneous slip is generated using the well-known CERS (Codes for Earthquake Rupture and ground-motion Simulation) model. The ensemble simulation can be completed in a short period of time to obtain the probability distribution of tsunami height, using a fast simulation method based on the unit source reconstruction. Taking a hypothetical M_w 9.0 earthquake tsunami event as an example, the forecasting results and efficiency of the ensemble method have been discussed. An optimized scenario sampling has been suggested by comparing with a statistically stable results obtained from refined scenarios sampling.