The sensitivity of tsunami models and tsunami energy to geologically constrained fault sources and shallow bathymetry

• Main Author: Felix, Raquel Picar
• Other Authors: Adam D. Switzer
• Format: Thesis-Doctor of Philosophy
• Language: English
• Published: Nanyang Technological University 2023
• Subjects: Science::Geology::Volcanoes and earthquakes
• Online Access: https://hdl.handle.net/10356/168435
• Institution: Nanyang Technological University

The field of tsunami research has advanced significantly in the past 18 years since the 2004 Indian Ocean tsunami, but mismatches between source mechanisms and coastal wave heights demonstrate that important knowledge gaps remain. In this thesis, I worked on topics related to the tsunami source and the coastal region. I then applied what I learned by addressing these knowledge gaps to assess the tsunami hazard in Bali and Lombok in Indonesia due to a potential earthquake on the Flores back-arc thrust. I first explored the tsunami source region by quantifying the role of the frontal thrusts and other splay faults in the generation of tsunami earthquakes. In order to compare the ability of an earthquake to generate a tsunami, I measured the energy of the initial tsunami wave using a modified tsunami energy equation that includes the water damping effect. This damping largely depends on the ratio of the uplift patch width to the water depth; based on my results, when this ratio is close to 1, <50% of the energy is transmitted from seafloor to the sea surface. The ratio is generally close to or less than 1 for frontal thrusts, because their narrow uplift patches are only few kilometers wide, close to the values of the water depth (average ocean depth is at ∼4 km; deepest ocean depth at the trench is at ∼11 km). Because of this, frontal thrusts generate relatively little tsunami energy, and cannot be responsible for tsunami earthquakes. In contrast, the d´ecollement which is much wider (tens to hundreds of kilometers) is not significantly affected by water damping, and it can more easily trigger large tsunamis. Overall, the frontal thrust has a minimal contribution to tsunami generation, and the focus should be on broader uplift source such as the d´ecollement. I next evaluated the impact of shallow bathymetry on tsunami models, to under-stand how tsunami waves can be amplified or diminished by details in the coastal bathymetry. For this study, I used the COMCOT software to numerically model tsunamis at four study sites. I compared the tsunamis propagating across bathymetry models with resolutions of 5, 10, 20, 30, 40, 50, 100, 200 and 300 m, and ∼455 m (GEBCO 2021 dataset). Using the 5 m resolution bathymetry as a baseline refer-ence model, the results showed that the 10 m – 50 m bathymetry models produce reasonably accurate coastal waves, with maximum wave heights and first wave arrival times that are only ≤10% and ≤5% lower than expected, respectively. In contrast, the maximum wave heights in the coarser bathymetry resolution models are underes-timated by as much as 30%, 40%, 60%, and 70% for the 100 m, 200 m, 300 m, and GEBCO bathymetry models, respectively. In general, this work shows that when it is not possible to use a very high-resolution bathymetry, a bathymetry with a res-olution of 10 – 50 m is sufficient to reasonably estimate tsunami wave heights, but coarser models can produce significant underestimation. The choice of specific resolu-tion should also depend on the mean distribution of data points, so that most of the data are represented well in the interpolated bathymetry. The use of publicly avail-able coarse-resolution dataset such as GEBCO, however, will lead to highly unreliable estimations of tsunamis at the coast. This work also highlights the need to provide in-depth details regarding the bathymetry used in the modeling (e.g., data source, data point distribution, interpolation method) as these are as important as detailing the tsunami source setup, and are critical for other researchers to assess the quality of the results of the numerical modeling. Finally, I applied what I learned from addressing the two previous topics to assess the tsunami hazard in Bali and Lombok due to the Flores back-arc thrust. The Flores back-arc thrust is selected as the study site because the 2018 Mw >6 earthquake sequence gave new insight regarding its fault geometry – with an active fault ramp downdip of a d´ecollement – and tsunamigenic potential. Hence, there is a need to re-evaluate the tsunami hazard due to slip on the thrust. To do this, I first generated a 30 m resolution bathymetry for the impacted region that represents the sounding points clustered around the main study sites, the capital cities of Mataram, Lombok and Denpasar, Bali. I applied the tsunami energy calculation to compare the initial tsunami energy triggered by the numerical model with and without the water filter and showed that their difference is only 2 – 3%. Therefore, using the unfiltered model does not have a great impact on the results. I then proceeded to model several earthquake and tsunami scenarios to assess regional hazard. The results showed that both cities could be hit by multiple waves, and that coseismic subsidence of 20 - 40 cm would exacerbate their exposure to tsunami waves and other coastal hazards. The waves that would hit Mataram are higher and arrive earlier (∼1.6 – 2.7 m high; <9 minutes) than the waves in Denpasar (∼0.6 – 1.4 m high; <23 – 27 minutes). I further explored the impact on Mataram and showed that the inundation could affect industrial sites, which in turn could trigger technological disasters, making the scenario more hazardous. Overall, this thesis addressed the sensitivity of tsunami models and tsunami energy to geologically constrained fault sources and shallow bathymetry. I hope that my contributions will help in the advancement of tsunami research and can be applied in future work.