A Rapid Rice Growth Monitoring Method Based on Sentinel-L SAR Data and Deep Learning

Close growth monitoring of crops such as rice using remote sensing data can help ensure food security and contribute to the achievement of the United Nations Sustainable Development Goals 2 (SDG 2) by 2030. To address the problems of insufficient timeliness of crop growth monitoring and one-sided portrayal of crop growth features, a fast crop growth monitoring method based on Synthetic Aperture Radar (SAR) data and deep learning algorithms is proposed in this paper. Based on Sentinel-1 dual-polarization SLC SAR data, a variety of crop features are extracted to comprehensively and meticulously portray the crop growth status. Then, the improved Deeplabv3+ model is used to map the SAR feature covariates to the graded crop growth labels, realizing the rapid and accurate monitoring of rice growth. Suihua City, Heilongjiang Province, China, was used as the study area to conduct experiments on rice at the tasseling stage. The results show that the proposed method portrays the growth of rice more comprehensively, and is capable of fast and accurate crop growth grading with an overall accuracy of 92.74% for the test area, which has a high potential for application.