Tsunami Simulations of the Sulawesi M w 7.5 Event: Comparison of Seismic Sources Issued from a Tsunami Warning Context Versus Post-Event Finite Source

The 28 September 2018 Sulawesi earthquake generated a much larger tsunami than expected from its M w = 7.5 magnitude and from its dominant strike-slip mechanism. Within a few minutes after the earthquake, the tsunami devastated the seafront of Palu bay, destroying houses and infrastructures over a few hundred meters. Coastal subsidence and slumping at various locations around the bay were also observed. There is debate in the scientific community as to whether submarine landslides and shore collapses contributed to the generation of strong and destructive waves locally. The objective of this study is threefold: first, to determine whether standard seismic inversions could predict the source in the context of tsunami early warning; second, to define a new seismic source built from optical image correlation and based on the geological and tectonic context; third, to assess whether the earthquake alone is able to generate up to 9-m wave heights at the coast. Numerical simulations of the tsunami propagation are performed for different seismic dislocation sources. Nonlinear shallow water equations are solved by a finite-difference method in grids with 200-m and 10-m resolutions. The early CMT focal solutions calculated by seismological institutes show dominant strike-slip mechanisms with a homogenous slip distribution. These sources produce maximum tsunami heights of 40-cm on the coast of Palu city. Two heterogeneous sources are tested and compared: the USGS “finite fault” model calculated from seismic inversion and a new “hybrid” source inferred from different techniques. The latter is based on a segmented fault in agreement with the geological context and built from both from seismic parameters of a CMT solution and the observed horizontal ground displacements. This source produces water wave heights of 4 to 5-m in the Palu bay. The observed inundation heights and distances are reproduced satisfactorily by the model at Pantoloan and at the southwestern tip of Palu bay. However, the “hybrid” source is unable to reproduce the largest 8 to 12-m water heights as reported from field surveys. Thus, even though this “hybrid” source produces most of the reported tsunami energy, we cannot exclude that the numerous coastal collapses observed in Palu bay contributed to increase the local tsunami run-up.