基于sentinel-1和sentinel-2数据的衡水市冬小麦越冬前面积提取
2021-01-15刘馨江亚军
刘馨 江亚军


摘要:以9月上旬至12月下旬sentinel-1A GRD为主要数据,提取小麦等不同地物后向散射系数变化曲线,根据当地种植制度和农事活动分析其后向散射系数曲线变化特点,结合11月下旬NDVI通过基于专家知识的决策树分类法进行冬小麦越冬前面积提取。经过精度验证,分类精度88.52%,Kappa 系数为 0.81。结果表明,该方法可在目标区冬小麦进入分蘖期后较准确地提取种植面积,为越冬期相关服务提供基础。
关键词:sentinel-1A 冬小麦面积提取决策树
中图分类号:P237文献标识码:A 文章编号:1672-3791(2021)11(c)-0000-00
Abstract: Taking sentinel-1a GRD from early September to late December as the main data, the backscattering coefficient change curves of different ground objects such as wheat are extracted. The variation characteristics of backscattering coefficient curve are analyzed according to local planting system and agricultural activities. Combined with NDVI in late November, the area of winter wheat before overwintering is extracted by decision tree classification based on expert knowledge. After accuracy verification, the classification accuracy is 88.52%, and the kappa coefficient is 0.81. The results show that this method can accurately extract the planting area after the winter wheat enters the tillering stage in the target area, and provide the basis for relevant services in the overwintering period.
KeyWords:Sentinel-1a; Winter wheat; Area extraction; Decision tree
冬小麦是衡水市主要农作物之一,其种植面积占全市粮食播种面积一半以上,在当地农业生产中占有重要比重[1]。近年来,随着气候变化和极端天气增加,给冬小麦气象服务带来严峻挑战,准确估算冬小麦种植面积是做好相关气象服务的基础。目前遥感技术已被广泛应用于农作物监测领域,与传统地面观测手段相比,遥感技术可大范围地获取地面信息,已成为一种有效的大面积农作物监测方法[2]。
在冬小麦面积遥感监测中通常采用光学卫星在拔节期进行反演,其余时段则由于其他绿色植被干扰难以有效地进行面积提取,这给冬小麦拔节期之前状态监测带来不便,而冬小麦在越冬期前生长状态决定其能否安全越冬,因此在越冬期前获取冬小麦种植范围信息对做好冬季相关农业气象服务具有重要意义。
在遥感监测中光学遥感容易受到云的影响,对于冬小麦来说最佳面积提取时间在3月下旬至4月上旬之间,时间窗口小。……
