Monitoring of Lake Area Changes from SAR Images Based on Convolutional Neural Networks and Markov Random Field

Monitoring of the distribution and changes of inland water is an important application of Synthetic Aperture Radar (SAR) data interpretation. Water extraction from SAR images is a challenging task because of the complexity of imaging conditions and the various land background. In this paper, a supervised classification method is proposed to delineate the lake water area from SAR images based on Convolutional Neural Networks (CNN) and modified Markov Random Field (MRF). To validate the effectiveness of the proposed method, the Sentinel-1 datasets are used to delineate the water area of Lake Shijiu and Lake Nanyi in the downstream Yangtze River Basin. The experimental results show that the performance of water bodies extraction from the two lakes is satisfactory. Moreover, the changes of water area of the two lakes in the downstream Yangtze River are revealed, and a serious expansion has been found in the lake surface area due to the large rainfall in July 2020.