Monitoring Daily Nighttime Light Based on Modis and Deep Learning: A Belgium Case Study

Satellite-observed night-time light has been utilized as a significant indicator for human activities and its impact on environment. Up to present, existing nighttime light (NTL) data still faces challenges in detecting the short-term human-related events due to limitation of the satellite revisit time and data quality. In this paper, we propose a promising approach for monitoring daily light during night based on deep learning and Moderate Resolution Imaging Spectroradiometer (MODIS). By modelling the relationship between MODIS and observed nighttime light, our proposed approach achieves the capability for conversion from MODIS image to Luojia-I-like daily NTL images. The quantitatively assessment of our generated NTL images demonstrates its great performance in terms of both similarity and NTL pattern.