Fit-for-Purpose Approach for the Detection and Analysis of Earthquake Surface Ruptures Using Satellite Images

Abstract

Coseismic surface ruptures are the ground truth of earthquake rupture along faults, and detailed mapping of earthquake surface ruptures provides an opportunity to understand the relationship between earthquakes and faults. One of the most effective tools for mapping earthquake surface ruptures is the use of remote sensing data, such as high-resolution satellite imagery. This study proposes a crack detection model based on deep learning and analysis tools to extract the geometrical characteristics of surface rupture using high-resolution satellite imagery (Pléiades-1B) for an earthquake that occurred in the Bulnay region of Mongolia. By comparing the prediction result with a line map of the previous study (Choi, 2018), it was possible to confirm the reliability of the fault detection. The model achieved a rupture detection accuracy of approximately 90% and an extraction of characteristic features of crack error level of 5% or less. To assess the broad applicability of the proposed model regardless of image size, we evaluated the model performance through quantitative and qualitative methods. The model accurately calculated essential characteristics, such as the orientation and length of a diverse range of surface ruptures. These results confirm the general effectiveness of the proposed model in detecting and characterizing surface ruptures caused by earthquakes. The suggested model for automated target detection utilizing satellite imagery can serve as a fit-for-purpose solution for conducting field surveys and acquiring fundamental earthquake-related data. The proposed model can provide valuable insights into the aftermath of seismic events by identifying a range of surface ruptures and deformations induced by earthquakes.

Keywords

• Coseismic surface ruptures
• deep learning
• earthquake fault mapping
• morphology feature extraction
• satellite imagery