考虑气象因子不确定性的参考作物蒸散量预报方法

参考作物蒸散量(ET_0)的准确预测预报对于制定作物灌溉制度与实时灌溉调度具有重要意义,然而气象因子的不确定性极大的影响着ET_0的预测精度。因此本研究采用马尔科夫蒙特卡罗模拟与自适应采样算法相结合的方法(AM-MCMC)对气象因子的不确定性进行修正,以气象站实测ET_0作为标准值,利用径向基神经网络(RBF)模型建立气象因子与ET_0的映射关系,建立基于气象因子不确定性的ET_0不确定性预测模型(CU-RBF),并以华北平原农田下垫面为例进行验证。结果表明,与传统的RBF确定性预报结果相比,CU-RBF预测结果的各精度评价效果均有所提高,纳什系数提高了10%,均方根误差和平均绝对误差分别降低了16.94%、17.05%,且单独修正平均风速的CU-RBFWs预测模型效果比分别单独修正最高温度、平均相对湿度的预测模型效果好。考虑气象因子不确定性开展ET_0的预报研究,减小了预测值与实际值的误差,可为农田下垫面的未来水分管理提供科学依据。

Prediction method of reference crop evapotranspiration considering the uncertainty of meteorological factors

The accurate prediction of reference crop evapotranspiration (ET₀) is of great significance for the formulation of crop irrigation schedule and real-time irrigation scheduling. However,the uncertainty of meteorological factors greatly affects the prediction accuracy of ET₀. Therefore,in this study,Markov Monte Carlo simulation combined with adaptive sampling algorithm (AM-MCMC) was used to modify the uncertainty of meteorological factors. ET₀ measured by meteorological station was taken as standard value,and radial basis function neural network (RBF) model was used to describe the mapping relationship between meteorological factors and ET₀. Then ET₀ uncertainty prediction model (CU-RBF) considering the uncertainty of meteorological factors was established. The farmland-underlying surface of North China Plain was taken as an example to verify the model. The results show that,compared with the traditional RBF deterministic prediction results, the accuracy evaluation effects of CU-RBF prediction results are improved, the NSE is increased by 10%,the RMSE and MAE are reduced by 16.94% and 17.05% respectively. Moreover,the prediction model with correction of average Ws is more effective than that with correction of T_max and RH. Considering the uncertainty of meteorological factors,the prediction of ET₀ is more consistent with the actual growth status of the crop,which provides a scientific basis for improved water management during crop production.