基于快速设定决策阈值的大范围作物种植分布的遥感监测研究

[目的]作物分布是研究作物种植结构的基础,利用遥感进行大范围作物布局的监测识别,对推进农业种植结构研究、分析农业模式和制定农业政策都具有重要的意义。为了更好地适应作物生产的需求,解决大范围作物种植分布遥感监测方法复杂的问题,亟待构建一种快速实用的作物提取方法,实现作物种植信息的快速高效获取。[方法]以江苏省水稻、小麦和玉米为研究对象,利用作物关键生育期内的多时相中分辨率遥感影像,针对作物生长特点进行影像的特征转换,以行政区县为基础的作业单元进行区域划分及阈值设定,构建多时相阈值决策提取模型,并提出一种基于少量样本投射的阈值快速确定的方法,实现大范围作物分布的快速识别。[结果]该方法能够快速分单元确定模型的阈值,

MONITORING CROP PLANTING DISTRIBUTION IN A LARGE AREA BASED ON REMOTE SENSING RAPID DECISION MAKING THRESHOLD

Crop distribution is the basis for studying crop planting structure. The use of remote sensing technology for monitoring and identification of a large area crop distribution is of great significance for promoting agricultural planting structure research, analyzing agricultural pattern and formulating agricultural policies. In order to better adapt to the needs of crop production and solve the complex problem of remote sensing monitoring methods for large scale crop planting distribution, it is urgent to construct a fast and practical crop extraction method to achieve crop planting information rapidly and efficiently. The rice, wheat and maize in Jiangsu province were took as the target crops. The multi temporal medium resolution remote sensing images during the critical growth period of crops were used as the data sources. The images were transformed according to the characteristics of crop growth and development and divided into working unit based on administrative districts. Then threshold decision extraction model was built and a method which could set threshold quickly based on small number of sample projection was proposed. The threshold of the model could be determined quickly by dividing the working unit. Through the results of verification of the area and positioning accuracy of the working units, the area relative error of planting rice, wheat and maize were all within 11%, and the positioning accuracy was better than 88%, which was basically consistent with the field survey. The method of quickly set the model threshold based on a small number of sample projection can be applied to monitor crops in a large area quickly, meet the application requirements, and has practicality to some extent.