Identification of Rice Planting Areas Based on GEE Platform:A Case Study of Heilongjiang Province
Accurate identification and extraction of rice plant-ing areas are of great value for grain yield estimation and effec-tive use of cultivated land. Previous studies focused on the identification of rice in a single key phenological phase,result-ing in poor spectral differentiation between rice and some land features in a single phase. Based on the sentinel-2 MSI data,this paper takes Heilongjiang Province as the research area,and uses the cloud platform of Google Earth Engine (GEE) to support the sentinel-2 MSI data. One-class Support Vec-tor Machine (OCSVM) is used for remote sensing recogni-tion and extraction of rice planting range. In this paper,a fu-sion phenological feature recognition algorithm based on pix-els is designed. Firstly,the timing analysis of rice in the study area is carried out. Then,the phenological features of rice and other land features in the study area are selected. Finally,the selected phenological features are fused to extract rice planting range. The results show that the accuracy of OA and Kappa are 0.969 and 0.93 respectively compared with the field sur-vey data.
万方数据
