基于作物空间物候差异提取黄淮海夏玉米种植面积

考虑大区域内不同纬度间玉米物候差异，利用MODIS EVI时序曲线提取黄淮海夏玉米种植面积。基于Landsat影像和MOD13Q1数据集，提取参考区夏玉米MODIS EVI时间序列曲线；根据研究区内农业气象站夏玉米生育期观测数据，构建夏玉米各物候期与纬度的关系，以纬度作为参数修正参考区夏玉米MODIS EVI时序曲线，获取研究区夏玉米EVI标准时序曲线，结合平均绝对距离（MAD）和p-分位数法提取黄淮海平原夏玉米面积。结果表明，利用遥感影像提取的北京、天津、河北、河南以及山东夏玉米面积分别为125.3×103、162.6×103、2231.8×103、2963.6×103和2731.9×103hm2，各省提取精度均达到80%以上。在市级尺度上，决定系数R2为0.82，均方根误差RMSE为147.8×103hm2；在县级尺度上，决定系数R2为0.62，均方根误差RMSE为17.7×103hm2。说明利用本方法能够准确有效地提取大区域内夏玉米种植面积，为其它农作物在大范围内估计种植面积提供新思路。

Mapping the Cultivation Areas of Summer Maize Using Spatial Variations of Crop Phenology over Huanghuaihai Plain

Considering the phenological differences of maize along with latitude over a large area, the planting area of summer maize was extracted by MODIS EVI time series curve over Huanghuaihai Plain in this paper. Based on Landsat images and MOD13Q1 dataset, the MODIS EVI time series curve of summer maize was obtained in a reference area. The phenology observation data from the agro-meteorological stations were collected to build the relationship between various summer maize growth stages and latitude. The latitude was used to correct the MODIS EVI time series curve of summer maize in the reference area, and then the standard EVI time series curve of summer maize was obtained. Using the mean absolute distance (MAD) method and p-tile algorithm, the summer maize planting area was estimated in the study area. The results showed that the areas of summer maize extracted from remote sensing images in Beijing, Tianjin, Hebei, Henan and Shandong were 125.3×103, 162.6×103, 2231.8×103, 2963.6×103 and 2731.9×103ha, respectively. Meanwhile, the accuracy of extraction result in each province was above 80%. The determinant coefficient (R2) is 0.82 and the root mean square error (RMSE) is 147.8×103ha at city level, and 0.62 and 17.7×103ha at county level, respectively. It indicated that the method in present study has the ability of extracting the summer maize planting area effectively and provides a new idea for estimating the planting area of other crops in a large region.