South Java tsunami model using highly resolved data and probable tsunamigenic sources

The huge Sumatra-Andaman earthquake tsunami (M_W > 9.2) on 26 December, 2004 and the Nias earthquake tsunami (M_W = 8.5) on 28 March, 2005, which occurred off Sumatra Island were followed by another earthquake off Java Island (M_W = 7.8), which generated a tsunami in the Indian Ocean on 17 July, 2006. The epicenter of this earthquake was located near the trench of the Sunda subduction zone south of Java Island. Unlike the two previous tsunamis off Sumatra Island, this earthquake generated a sizable tsunami whose run-up were larger than expected based on the given seismic waves emanating from the slow-weak earthquake, and was classed as a tsunami earthquake.

Following this series of tragic events, which caused huge losses of life and enormous damage to infrastructure, initiatives and efforts were made by government and non-government institutions as well as scientists from a wide range of disciplines to gain a greater understanding of the earthquake-generating mechanisms along the Java Trench and increase the ability and knowledge required to identify hazards, impacts, and countermeasures.

The absence of accurate near-shore bathymetry and topography data poses a challenge for the tsunami model for the 17 July, 2006 Java is. Furthermore, the performances of the available models following such events, displayed inconsistencies when compared to the field observation data. In addition, the detailed analysis of tsunami models in terms of the variety of source models and their parameters affecting the tsunami run-up heights and their distribution along the coastline as well as model validations were limited.

In the present study, the tsunami of 17 July, 2006 is studied using a broad range of tsunami sources, highly resolved bathymetry and topography data, and extensive post-tsunami field observations for model validation. The numerical modeling of the tsunami was carried out using TUNAMI code, which is based on non-linear shallow water equations (NSWE). For the tsunami sources, the present study uses various fault models that were estimated using the empirical scaling laws and the inversion model from the tsunami mareograms as well as the finite fault model from broadband networks of seismic waveforms. The deformation model used the analytical expressions derived by Mansinha & Smylie (1971) and Okada (1985), while the GITEWS project established the RuptGen tool to determine seafloor displacement in the Sunda arc region.

The reconstruction of the 2006 Java tsunami model suggests that source models imposing low-rigidity material and higher slip are able to model fields data and are more comparable to the run-up heights than those using normal values. In addition, the comparison of tsunami amplitudes near-shore between numerical simulations and field observation shows that the distributed slip of the multi-fault model gives the better result than those from the uniform slip of the single-fault model. Furthermore, by evaluating the fitting curves' shape and the discriminant parameters estimated from the empirical formula, it suggests that the cause of the event was not so clear whether those due to solely seismic dislocations or associated with submarine landslides. The limitation of the NSWE model seems to have caused a deficiency of tsunami run-up in hilly areas where steep slopes exist.

Based on the proposed synthetic model for the effect of complex ruptures, the ratio between tsunami run-up resulting from uniform slip and distributed slip is in the range \~1–2.5 along the coastline, whose length is four times that of the fault. While, the variation tsunami run-up normalized by average slip along the coastline follows Gaussian curves, with those at the center being the maximum with a ratio value varying in the range \~0.6–1.55. The rigidity value ratio of 2, 3, and 4 times will reduce the tsunami amplitudes on average by 61%, 47%, and 37%, respectively. The significantly different geometric data input of about 27% only showed a negligible deviation in tsunami amplitudes of 2.4% and 1.5% for distributed and uniform slips. The tests for the 2006 Java tsunami source model using the proposed dimensionless graphs resulting from the present study, namely the ratio of rigidities, the ratio of uniformly distributed slips, and the variation of run-up, indicated that both display consistent trends.

To determine the level of future tsunami hazard in the study area, the hypothetic model based on several limited scenarios has been constructed. It revealed that the tsunami inundation concentrated in certain locations consists of four clusters. One of the farthest and biggest inundations is in the river mouth of the Serayu and its surrounding low-lying land up to \~15 km further east with the tsunami penetration reaching \~5 km inland. The other clusters, their tsunami inundation is varying less than \~1 km.

The effectiveness of the proposed mitigation measures varies by about \~7.5–27% in terms of the inundation area. The greenbelt of “Waru” trees is able reducing tsunami \~7.5%, while the sand dunes are of about 27%. These results suggest the mitigation measures by utilizing the artificial greenbelt and sand dunes are insufficient, hence the mitigation programs of vertical and horizontal evacuation in study area is highly demanded.