AquaCrop模型在东北黑土区作物产量预测中的应用研究

东北黑土区是我国玉米和大豆生产基地,为了实现利用AquaCrop模型优化管理和预测产量,本文基于作物小区田间试验和大田观测数据,采用OAT(one factor at a time)法分析了该模型参数的敏感性,率定了敏感性高的参数,并对率定后的模型进行了验证。结果表明:玉米和大豆产量均对影响经济产量的收获指数十分敏感,二者虽然对冠层和根系生长参数都敏感,但有所差异:玉米对冠层衰减系数(canopy decline coefficient,CDC)更为敏感,而大豆则对限制冠层伸展的水分胁迫系数曲线的形状因子(shape factor for water stress coefficient for canopy expansion,Pexshp)更为敏感;玉米因根系深对最大有效根深(maximum effective rooting depth,Zx)更敏感,大豆因根系浅对根区根系伸展曲线的形状因子(shape factor describing root zone expansion,Rexshp)更敏感。由于玉米需水量大,对冠层形成和枯萎前的作物系数(crop coefficient before canopy formation and senescence,KcTr,x)和归一化水分生产力(normalized water productivity,WP*)很敏感,大豆则是一般敏感。率定后模型模拟玉米产量与实测产量的回归系数由0.34提升至0.89,模拟大豆产量与实测产量的回归系数由0.80提升至0.88。进一步用大田实测产量的验证结果表明:预测的玉米与大豆产量与实测产量间回归方程的决定系数(coefficient of determination,R2)分别为0.775和0.779,均方根误差(root mean square error,RMSE)分别为1.076 t hm^–2和0.299 t hm^–2,标准均方根误差(normalized root mean square error,NRMSE)分别为0.097和0.178,模拟效率(model efficiency,ME)分别为0.747和0.730,率定后的AquaCrop模型能较精准地模拟东北黑土区玉米和大豆产量,可用于产量预测或优化管理。

Application study of crop yield prediction based on AquaCrop model in black soil region of Northeast China

Northeast black soil area is the production area of maize and soybean in China.In order to optimize the agricultural management and forecast crop yield with AquaCrop model,we use OAT(one factor at a time)method to analyze the sensitivity of the model parameters based on the experiment and field observation data,and to validate the model after calibrated the high sensitivity parameters.The results of sensitivity analysis showed that the yields of maize and soybean were both extremely sensitive to the reference harvest index(HI0)and the parameters of canopy growth and root growth.The difference was that maize was more sensitive to the canopy decline coefficient(CDC),while soybean was more sensitive to the shape factor for water stress coefficient for canopy expansion(Pexshp).Maize was more sensitive to the maximum effective rooting depth(Zx)because of its deep root,while soybean was more sensitive to the shape factor describing root zone expansion(Rexshp)because of its short roots.Maize was extremely sensitive to the crop coefficient before canopy formation and senescence(KcTr,x)and the normalized water productivity(WP*)due to the large water demand,while soybean was only generally sensitive.After calibrated the high sensitivity parameters with experiment data,the regression coefficient of simulated yield and measured yield of maize increased from 0.34 to 0.89,and the regression coefficient of simulated yield and measured yield of soybean increased from 0.80 to 0.88.Furthermore,the validation results of field observation data indicated that the determination coefficients(R2),the root mean square error(RMSE),the normalized root mean square error(NRMSE)and the model efficiency(ME)of the AquaCrop model of maize and soybean were 0.775 and 0.779,1.076 t hm^–2 and 0.299 t hm^–2,0.097 and 0.178,0.747 and 0.730,respectively.The calibrated AquaCrop model can accurately simulate the yield of corn and soybean in the black soil area of Northeast China,and is useful for yield prediction and optimal management.