基于GEE的宝鸡峡灌区耕地灌溉面积遥感监测方法

准确监测耕地灌溉面积的变化，对于管理有限的农业水资源及保证粮食安全等具有重要意义.选择中国西北干旱半干旱地区的典型灌区陕西省宝鸡峡灌区，基于Google Earth Engine(GEE)云平台的Landsat卫星数据，在土地利用分类提取出耕地范围的基础上，结合实地采样获得的样本点，并基于宝鸡峡灌区实际，提出了一种获取历史样本点的思路和方法；通过计算归一化植被指数(NDVI)、增强植被指数(EVI)、绿色叶绿素植被指数(GCVI)、水调节绿色指数(WGI)等指数作为分类参数，应用随机森林(RF)算法对2010年及2020年耕地灌溉区域进行了识别与提取.结果表明，宝鸡峡灌区2010年和2020年耕地灌溉面积分別为104 211和90 174 hm²,混淆矩阵验证总体精度分别为91.67%和95.35%,Kappa系数为0.832和0.907.

Remote sensing monitoring method of cultivated land irrigation area in Baojixia Irrigation District based on GEE

Accurate monitoring of the changes of cultivated land irrigation area is of great significance for managing limited agricultural water resources and ensuring food security. Baojixia Irrigation District of Shaanxi Province, a typical irrigation district in arid and semi-arid areas of northwest China, was selected. Landsat satellite data based on GEE cloud platform, on the basis of extracting the range of cultivated land use classification, combined with field sampling points, and based on the actual situation of Baojixia Irrigation District, a method for obtaining historical sample points is proposed. By calculating NDVI, EVI, GCVI, WGI and other indexes as classification parameters. The random forest algorithm was used to identify and extract the irrigated areas of cultivated land in 2010 and 2020. The results show that the irrigated area of cultivated land in Baojixia Irrigation District in 2010 and 2020 are 104 211 hm² and 90 174 hm² respectively, and with overall accuracy of 91.67% and 95.35%, Kappa coefficients are 0.832 and 0.907.