基于WOFOST模型和遥感数据同化的区域夏玉米产量预测  

Regional Summer Maize Yield Prediction Based on WOFOST Model and Remote Sensing Data Assimilation

作　　者：吴韡[1] 温华炜 李俊博 朱子情 曹良中 Wu Wei;Wen Huawei;Li Junbo;Zhu Ziqing;Cao Liangzhong(College of Agricultural Economics and Technology,Jiujiang Vocational University,Jiujiang 332499,Jiangxi,China;School of Tourism and Geography,Jiujiang University,Jiujiang 332005,Jiangxi,China)

机构地区：[1]九江职业大学农业经济技术学院,江西九江332499 [2]九江学院旅游与地理学院,江西九江332005

出　　处：《绿色科技》2025年第2期271-278,共8页Journal of Green Science and Technology

基　　金：江西省教育厅科学技术研究项目(编号:GJJ21183)。

摘　　要：以山东省德州市禹城市为研究区域,夏玉米为研究对象。首先利用Sobol方法对模型内的作物参数进行全局敏感性分析,筛选出关键的待优化参数,并通过DIRECT-L和SUBPLEX算法对这些参数进行标定。最终,通过集合卡尔曼滤波方法同化遥感信息与作物生长模型,准确预测了区域内夏玉米的单产水平。结果显示:应用DIRECT-L和SUBPLEX算法对选出的敏感性参数标定,标定结果的相对均方根误差(RRMSE)均低于10%,显示出模拟精度高,这表明调整后的模型能够有效地模拟本研究区域内夏玉米的生长状况。通过集合卡尔曼滤波同化GLASS-LAI数据与WOFOST模型,预测区域夏玉米产量,LAI均方根误差由0.23降至0.14,相对均方根误差从9.15%降至5.52%,表明该方法有效提高了产量预测精度。This study takes Yucheng City of Dezhou City,Shandong Province as the research area,and summer maize is selected as the research object.In this paper,Sobol method is used to analyze the global sensitivity of crop parameters in the model,select the key parameters to be optimized,and calibrate these parameters by DIRECT-L and SUBPLEX algorithms.Finally,by integrating remote sensing information and crop growth model with ensemble Kalman filtering method,the yield level of summer maize in the region was accurately predicted.The results showed that the relative root-mean-square error(RRMSE)of the calibration results were lower than 10%by using the DIRECT-L and SUBPLEX algorithms,indicating high simulation accuracy,which indicated that the adjusted model could effectively simulate the growth of summer maize in the study area.The integrated Kalman filter was used to assimilate GLASS-LAI data and WOFOST model to predict regional summer maize yield.The root mean square error of LAI decreased from 0.23 to 0.14,and the relative root mean square error decreased from 9.15%to 5.52%,indicating that this method effectively improved the yield prediction accuracy.

关 键 词：产量预测 WOFOST模型 集合卡尔曼滤波 数据同化 

分 类 号：S513[农业科学—作物学]