基于Google Earth Engine平台的关中冬小麦面积时空变化监测

以关中地区为研究区,基于Google Earth Engine(GEE)平台,根据冬小麦生育期内归一化植被指数(NDVI)时序曲线和物候特征,采用NDVI重构增幅算法和光谱突变斜率,构建了关中地区冬小麦提取模型并实现了冬小麦种植面积的提取。用农业统计面积验证提取结果表明:在市级和县级尺度上,决定系数R~2分别为0.82和0.62,一致性指标d分别为0.95和0.84,提取结果与实地调查数据的空间一致性精度为93.4%。结果显示:关中地区冬小麦主要分布在中部关中平原,冬小麦种植面积在2011—2017年呈下降趋势,减少了83.22×10~3 hm~2(8.47%)。综合考虑冬小麦NDVI时序曲线的"峰""谷"特征,具有一定的普适性,可为大面积连续年份冬小麦种植面积时空监测提供参考。

Monitoring of spatial\|temporal change of winter wheat area in Guanzhong Region based on Google Earth Engine

A method based on Google Earth Engine (GEE) was developed and tested in this study for mapping winter wheat planting area over Guanzhong Region. After analyzing the winter wheat MODIS NDVI time series curve and its phenology calendar in Guanzhong Region, the extraction model was established based on NDVI remodel amplification and NDVI increase/decrease slope threshold. The result showed that area of winter wheat extracted by remote sensing agreed well with statistics, with the determination coefficient R² of 0.82 and 0.62, and the index of agreement d of 0.95 and 0.84 at city level and county level, respectively. Further, the spatial consistency accuracy of extraction results and field survey data was 93.4%. Winter wheat was mainly distributed on Guanzhong Plain, whose terrain is relatively flat. In 2011-2017, winter wheat planting area in Guanzhong Region was showing a downward trend. There was a series decline by 8.47% (83.22×10³hm²) in planting area of winter wheat in 7 years. In this study, the peak and valley of NDVI curve were considered as never before. This study can serve as a reference for large\|scale and long\|time monitoring of spatial and temporal information of winter wheat planting area with remote sensing data.