Characterizing tsunami earthquake rupture parameters with forward modeling and near-field geophysical data
Abstract
Tsunami earthquakes (TsEs) are rare, end-member earthquakes that generate tsunamis much larger than expected for their size. Currently, these events are challenging for local tsunami early warning (TEW) systems. Their algorithms do not discriminate TsEs as threats because of their more moderate magnitude (~ M7-8) compared to typical tsunamigenic events (~ M8.5-9). As such, many casualties are common, and it is crucial to accurately identify TsEs in real-time to warn inhabitants of the inundation zone. Methods involving energy-to-moment and energy-to-duration ratios have previously been used to discriminate TsEs because these events radiate high frequency energy inefficiently. However, these methods use teleseismic data and either need further evaluation after the event or do not stabilize quickly enough to issue a warning before inundation. Thus, it is critical to use near-field data for real-time discrimination. Currently, near-field data only exist for the 2010 M7.8 Mentawai TsE. To compensate for this lack of data and provide estimates on likely rupture parameters for future TsEs, we generate large suites of synthetic rupture models patterned after the Mentawai event and compare them to the observed strong motion and HR-GNSS data. We vary several rupture parameters in the models to determine which combinations best reproduce a TsE scenario.
The rupture parameters we vary are stress drop, rupture velocity, and rise time, because each of these are believed to be characteristically extreme for TsEs because TsEs rupture the shallowest segment of a subduction zone in the compliant accretionary wedge. We have generated hundreds of synthetic rupture scenarios for events with stress drops between 0.1-2 MPa, values that are low compared to typical megathrust earthquakes (~5MPa). The synthetic data have been compared to the observed data, and the residuals are evidently smaller with lower stress drops. We will present these results, as well as results for variable rise time and rupture velocity, as well as all three rupture parameters together to find the most accurate combinations with given uncertainties. By characterizing TsE rupture parameters, we can improve TEW machine learning algorithms to be able to identify these rare, but destructive events.- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2020
- Bibcode:
- 2020AGUFMNH017..04N
- Keywords:
-
- 4315 Monitoring;
- forecasting;
- prediction;
- NATURAL HAZARDS;
- 4333 Disaster risk analysis and assessment;
- NATURAL HAZARDS;
- 4341 Early warning systems;
- NATURAL HAZARDS;
- 4564 Tsunamis and storm surges;
- OCEANOGRAPHY: PHYSICAL