[H-DS27_1PO1] Tsunami and its Forecast

[HDS27-P06] Pre-computed Tsunami Database with Additional Slip Near to the Trench for Tsunami Early Warning in Southern Java

*Haris SUNENDAR¹, Yuichiro TANIOKA¹, Aditya GUSMAN¹, Hamzah LATIEF² (1.Institute of Seismology and Volcanology, Hokkaido University, 2.Bandung Institute of Technology)

Keywords:pre-computed tsunami database, tsunami earthquake, tsunami early warning

We build tsunami database based on simple fault model scenarios for the Java trench subduction zone. We have 480 points along the subduction zone with distance between each other of 20 arc-min. This points are used as the center of simple fault model scenarios. Each point is the center of several fault models with different moment magnitudes.We used a magnitude to fault length and width scaling relationship for the fault model scenario. The moment magnitudes for the fault model scenarios are from Mw 6.3 to Mw 9.0 with interval of 0.3. The fault depth parameter is also a variable for the fault model scenario. We used depth between 10 km to 60 km with interval of 10 km.From each fault model scenario we simulate tsunami propagation by solving the linear shallow water equations. We used bathymetry data based on Indonesian navy chart and GEBCO bathymetric dataset. The grid size for the tsunami simulation is 1 arc-min. The pre-computed maximum tsunami heights and tsunami arrival time at every point along the coast are stored in a database. If a real earthquake occur at any location in the forecasting domain then the pre-computed tsunami heights from 16 scenarios are retrieved from the database. Theses 16 scenarios are those that are surrounding the actual hypocenter and each of the scenario has the closest higher or closest lower magnitude to the actual one. Then the tsunami heights from these scenarios are used in interpolation methods to get the tsunami height forecast. The tsunami heights from two scenarios with a same hypocenter and different moment magnitudes are interpolated by logarithmic interpolation. Then the tsunami heights with different depths and different epicenters are interpolated using linear interpolation and bilinear interpolation, respectively. The the interpolated tsunami heights is group into district administrative regions, then the maximum height for each administrative region is selected. The selected tsunami heights are categorized into three different warning levels. These levels are tsunami smaller than 0.5 m, between 0.5 m to 3 m, and larger than 3 m. We apply this method to forecast the tsunami generated by the 1994 East Java earthquake. The 1994 earthquake is classify as a tsunami earthquake (Newman and Okal, 1998; Pollet and Kanamori, 2000). The earthquake moment magnitude was estimated to be Mw 7.6 (Abercrombie et al., 2001), Mw 7.8 (Bilek et al., 2006; USGS), Mw 7.9 (Pollet and Kanamori, 2000). Our result shows that the forecasted tsunami heights underestimate the actual tsunami heights. One of the main cause could be the fact that we used simple fault model scenarios which sizes were estimated from scaling relationship of magnitude to fault dimension of regular earthquake but not tsunami earthquake. Previous studies shows tsunami earthquake may generate large slip near the trench (Tanioka and Satake, 1996; Satake et al., 2013). Therefore to obtain a more accurate forecast, the fault model scenarios near the trench should represent fault model for tsunami earthquake event.