Uncertainty Analysis of Back-Projection Methods

Date: 4/25/2019

Time: 08:45 AM

Room: Cascade I

Back-Projecting high frequency (HF) waves is routinely used to image rupture processes of large earthquakes. However, HF waves are strongly affected by source depth, focal mechanisms and the Earth’s 3D velocity structures, causing large uncertainties in back-projection (BP). So far these uncertainties have not been thoroughly investigated. Here we perform 1D and 3D synthetic tests to investigate uncertainties in two representative BP methods, Multiple SIgnal Classification and Compressive Sensing. We generate synthetics for sources embedded in 1D or 3D velocity models with different depths and focal mechanisms, and then back-project them using the two methods based on array configurations. The focal depth test shows that depth phases can be back-projected as artifacts swimming towards the array with offsets proportional to the source depth, suggesting that depth phases are mapped to their reflection points in the surface. For a multiple point source model involving varied focal mechanisms, the seismograms differ significantly between arrays. Using the 2016 Mw7.8 Kaikoura earthquake as a scenario, both real and synthetic data show that the South American and Australian arrays image inconsistent rupture evolution, especially for late segments. We attribute this to varied focal mechanisms at varied rupture stages. We also test the impact of 3D velocity structures by simulating an event in the Java subduction zone. The 3D trench and coast structure can generate strong and long lasted codas, which are mirrored as artifacts far from the input. Finally, we model an offshore event in the Sumatran subduction zone and show that the generated wavefields feature the frequency-dependence, leading to frequency-dependent BP results. In summary, our analyses indicate that the preceding factors can affect various aspects of BP results. Thus, we suggest target-oriented synthetic tests, such as simulations of megathrust earthquakes with 3D source-side velocity structures, should be conducted when we interpret detailed BP images to infer the earthquake kinematics and dynamics.

Presenting Author: Hongyu Zeng

Authors