Regional-Scale Monitoring of Rice Flood Disaster Based on Multi-Temporal Remote Sensing Images

摘要

Rice flood disaster causes wide range of damage in the rice production worldwide. To obtain the spatial distribution of damage area and access the damage severity, a fast and accurate loss assessment method is of great importance for effective administration. The objective of this study was to determine suitable spectral features for rice flood disaster detecting and to develop a mapping method at regional scale based on satellite remote sensing image data. Within a number of vegetation indices (VIs) that were examined for their sensitivity to the rice flood disaster, Normalized Difference Water Index (NDWI) and enhanced vegetation index (EVI) were identified as the optimal VIs for detecting rice flood disaster. Based on the two-date satellite images and multi-temporal analysis, a mapping method for monitoring the rice flood disaster at a regional scale were developed based on support vector machine (SVM) and maximum likelihood classifier (MLC). And the model that relies on SVM method significantly outperformed the MLC method, with the overall accuracy increased from 0.64 to 0.82. The successful application of this method in a typical rice flood disaster in Anhui Province, China demonstrated the feasibility of the method and showed its promising potential to be implemented in practice.