基于HJ卫星CCD数据的冬小麦病虫害面积监测

遥感在农作物病虫害监测中广泛采用的简单植被指数阈值法难以判别冬小麦的健康状况。该研究选择二值逻辑回归法，分别建立实测光谱得到的植被指数与其健康状况之间的关系模型。结果表明，重归一化植被指数RDVI模型和三角植被指数TVI模型可信度较好。考虑到遥感监测冬小麦病虫害时，涉及的地域范围广，冬小麦生长状况存在明显的局域差异，采用了3×3邻域像元的一致性假设消除局部环境差异。将模型应用到中国新近发射的环境与灾害监测预报小卫星星座HJ-CCD传感器数据，得到提取的冬小麦受病虫害胁迫范围与枣阳市植保站普查结果相符，也与地面实测结果相一致，其中，TVI模型结果的精度达到76.47%，能够满足农作物病虫害面积遥感监测要求。

Diseases and insect pests area monitoring for winter wheat based on HJ-CCD imagery

Previous studies of remote sensing on monitoring crop diseases show that simple vegetation index threshold methods are not effective to discriminate stressed winter wheat from the healthy ones. In this paper, binary logistic regression was utilized to establish the relational model between field measured vegetation indexes and health status of winter wheat. In order to reduce the growing variations between different regions, an algorithm was implemented based on an assumption of neighborhood consistency among 3×3 pixels. Results indicate that renormalised difference vegetation index (RDVI) model and triangular vegetation index (TVI) model perform satisfactorily and have high reliability. These models were then applied to CCD image of the newly-launched environment monitoring and disaster forecasting microsatellite of China. Results showed that the area of stressed winter wheat detected by RDVI and TVI models were consistent with the census by the Zaoyang Plant Protection Station as well as the field investigation. Besides, TVI model has the accuracy of 76.47%, which can meet the needs of operational crop diseases and insect pests area monitoring.