Estimation of winter wheat yield based on coupling remote sensing information and WheatSM model
Estimation of winter wheat yield based on coupling remote sensing information and WheatSM model
复制标题
DOI：
10.13292/j.1000-4890.201907.039
复制
发表时间：
2019-07-01
期刊：
影响因子：
--
通讯作者：
Yu Wei-dong
中科院分区：
文献类型：
--
作者：
Li Ying;Chen Huai-liang;Yu Wei-dong
中国为不同类型冬小麦开发的小麦生长模型WheatSM在科学研究和公共服务中得到了应用。遥感信息与作物生长模型的耦合在大面积作物生长监测和估产中具有重要的应用价值。河南省鹤壁市是中国的冬小麦主产区。以鹤壁市为研究区域，将2013-2017年优化重构的时间序列MODIS LAI数据与WheatSM模型相结合，结合SCE-UA最优同化和EnKF同化方法，从站点和区域两个尺度上估计冬小麦产量。结果表明，在站点尺度上严格标定WheatSM参数的前提下，引入具有不确定性的遥感数据并没有提高作物模型的模拟精度。遥感观测数据的质量对EnKF同化方法结果的影响要大于SCE-UA最优同化方法。在区域尺度上，使用SCE-UA和EnKF方法的资料同化结果的精度均高于没有同化的资料同化结果。模拟产量与统计产量之间的均方根误差分别从2036.0 kg“hm-2降至1641 kg”hm-2和1587.7 kg“hm-2，降幅分别为19.4%和22.0%，且EnKF同化方法的同化效率高于SCE-UA最优同化方法，为WheatSM与遥感数据的耦合同化策略的选择提供了依据。
WheatSM, a wheat growth model developed for different types of winter wheat in China, is applied in scientific research and public service. The coupling of remote sensing information with crop growth model has important application value in crop growth monitoring and yield estimation in large area. Hebi City in Henan Province is a main winter wheat producing area of China. With Hebi as the study area, the optimized reconstructed time series MODIS LAI data from 2013 to 2017 was coupled with WheatSM model with both SCE-UA optimal assimilation and EnKF assimilation methods to estimate the yield of winter wheat at both site and regional scales. The results showed that the introduction of remote sensing data with uncertainties did not improve the simulation accuracy of the crop model on the premise of strictly calibrating the parameters of WheatSM at site scale. The quality of remote sensing observation data had greater effects on the results of EnKF assimilation method than that of SCE-UA optimal assimilation method. At regional scale, the accuracy of data assimilation results with both SCE-UA and EnKF methods were higher than that without data assimilation. RMSE between simulated yield and statistical yield decreased from 2036.0 kg " hm-2 to 1641 kg " hm-2 with SCE-UA method and to 1587.7 kg " hm-2 with EnKF method, with a reduction of 19.4% and 22.0%, respectively. The efficiency of EnKF assimilation method was higher than that of SCE-UA optimal assimilation method. Our results could provide a basis for the selection of data assimilation strategies coupling WheatSM with remote sensing data.