Rice Fields Mapping Using Multi-Temporal GF-6 Remote Sensing Images

Abstract

Rice is one of the most important crops in the world, meanwhile, the rice field is also an important contributor to greenhouse gas methane emission. Therefore, it is important to get an accurate estimation of rice acreage for both food pro-duction and climate change related studies. This study aims to quickly and accurately map the rice planting area using GF-6 WFV time-series images in Hongze District, Jiangsu Province of China. Field campaigns were carried out during the rice growing season and ground-truth data were collected for classification accuracy assessments in 2019. The land cover types were classified as rice, natural vegetation, water bodies, economic crops and other nonvegetated areas. The Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI), Ratio Vegetation Index (RVI), and Normalized Differ-ence Red Edge 1 Index (NDRE1) were established by the GF-6 WFV images taken in several key growth stages of rice. As such, a remote sensing extraction of rice was established, according to the dynamic change of NDVI, NDWI, RVI, and NDREI of var-ious feature types over time. A stepwise classification strategy utilizing the VIs signatures during key growth stages of rice was proposed. After the accuracy verification by visual interpretation points using RapidEye high-resolution image, the overall classification accuracy was calculated. Results showed that the extraction area of rice was 332.35 km <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> in the study area, and the overall classification accuracy was 92.43 %. At the same time, the distribution map of rice without red edge bands was ob-tained using the same remote sensing images and masks, substi-tuting NDVI for NDREI in the preliminary rice extraction. More importantly, the extraction with red edge bands showed increases of 0.50%, 4.71 %, 5.93% and 0.05 in the overall accu-racy, producer's accuracy, user's accuracy of rice, and kappa coefficient, respectively. By contrast, the extraction with or without red edge bands was superimposed on the remote sensing image, indicating that the rice distributions were similar, but the extraction without red edge bands presented an obvious omission. This finding proved that the red edge bands effectively re-duced the classification error and omission of crops. Conse-quently, the domestic red edge satellite data can provide a great application potential to crop classification and area extraction.