南方丘陵地区水稻种植面积遥感信息提取的试验

EXPERIMENTAL RESEARCH ON RICE PLANTING AREA OF HILLY REGION IN SOUTHERN CHINA USING REMOTELY SENSED DATA

以浙江省为试验区，针对水稻种植面积遥感信息提取的业务化运行问题，进行了以下试验：（１）以传统的单象元统计分类识别方法为分类器，在地理信息系统支持下，提取丘陵地区大范围水稻种植面积信息的可行性；（２）在遥感资料的基础上，结合地形数据综合提取水稻种植面积专题信息的可行性和有效性；（３）混合象元分解方法在丘陵地区的有效性和适用性．结果表明，用最大似然法提取大范围水稻种植面积信息的精度可满足业务化运行的要求；模糊监督分类有较高的分类精度和较好的稳定性，具有较强的适应性；坡度数据作为遥感影像分类的辅助数据层，可以有效地提高丘陵地区水稻种植面积信息的提取精度，还可以提高分类的稳定性和空间位置精度．

Taking Zhejiang Province as the experimental area, the experimental research on rice planting area of hilly region in southern China is carried out by using NOAA/AVHRR data.The main contents concern the contrast tests on the practical approaches.Both digital elevation model (DEM) and digital slope model (DSM) derived from the digital relief map are used for the purpose of improving the classification accuracy of AVHRR data in large area hilly region.The results show that the accuracies of maximum likelihood (MLH) classification could satisfy the professional requirements of estimating rice planting area, and fuzzy supervised classification (FSC) has better classification accuracy and stability than that of MLH. In addition, DSM may improve the results of extracting paddy field signatures from AVHRR, particularly may improve the spatial precision.