A Multi-Fault Model Estimation From Tsunami Data: An Application to the 2018 M7.9 Kodiak Earthquake

Date: 4/24/2019

Time: 03:00 PM

Room: Vashon

A multi-fault model of a complex event, the January 23, 2018 M7.9 offshore Kodiak earthquake, is estimated from tsunami data by utilizing a Green’s function based time reverse imaging (GFTRI) with an adjoint sensitivity (AS) method. The adjoint approach has been used in numerical weather prediction to find the optimal locations for adaptive observations, and has recently been adapted for tsunami studies. The GFTRI method requires Green’s functions (GF) from each source patch obtained by dividing a source region into a regular grid of point sources. GFs are computed using an initial unit source model whose amplitude is concentrated near the grid point. The AS method has been successfully applied to the 2009 Samoa earthquake tsunami (Hossen et al., GRL, 2018).

In this study, we applied the AS method with GFTRI to invert for the source of the January 23, 2018 Kodiak earthquake using tsunami waveforms from DART and tide gauge stations. The M7.9 earthquake occurred 300 km southeast of Kodiak Island, Alaska, on the incoming Pacific plate in the outer rise region of the Alaska-Aleutian subduction zone. The global Centroid Moment Tensor (GCMT) solution indicates faulting occurred on a steeply dipping fault striking either west-southwest (left lateral) or north-northwest (right lateral) while subsequent work (Ruppert et al., GRL, 2018) reveals a more complex pattern of strike-slip faulting. Our results suggest that the rupture occurred on five fault planes oriented in both N-S and E-W fault direction. The result is based on the tsunami source inversion in which we estimated a sea surface displacement using tsunami data recorded by tide gauges and DART buoys located around the source. From the reconstructed sea surface displacement, we estimated the slip distribution on the fault planes chosen based on the fault-parameters suggested by GCMT solution but with different epicentral locations. We carried out a number of source inversions using different combinations of fault planes to find the multi-fault model that provides smallest residual error.

Presenting Author: Md Jakir Hossen

Authors