Large-Scale Rice Mapping of Thailand using Sentinel-1 Multi-Temporal SAR Data

Abstract

With synthetic aperture radar (SAR) gradually entering the big data era, large-scale crop mapping research has important scientific significance and application prospects. Thailand is the largest rice exporter of the world, controlling a third of the global market. Therefore, this paper proposed a large-scale rice mapping method for rice monitoring in Thailand. Three simple but effective temporal features were defined based on the phenological information of rice. In order to improve the accuracy of rice mapping, this paper introduced the deep learning semantics segmentation methods, which have already achieved tremendous success in the field of computer vision, into the large-scale rice mapping using the SAR time series. The experimental result showed that the proposed large-scale rice mapping method can achieve a satisfactory result, which achieved 92% overall accuracy.