Multi-Sensor Satellite-Imaging Data Fusion for Earthquake Damage Assessment and the Significant Features

With the recent developments in remote sensing technologies, the resolution and detection capabilities of both synthetic aperture radars (SAR) and optical sensors have increased significantly enough to detect damages of earthquakes. However, extensive human labor and domain expertise is required for processing and analyzing large areas affected by disasters. We propose a multi-sensor data fusion-based neural network, which combines the datasets from a SAR and an optical sensor for the damage detection of the 2018 Sulawesi earthquake in Indonesia. We also try to understand which input features are the significant contributors for which classification outputs.