Daily Evapotranspiration Simulation Study of Apple Trees Based on Stacking Ensemble Learning Model
In order to accurately simulate the daily evapotranspiration of apple trees,an ensemble learning model(LSM)based on the stacking strategy was established with support vector machine(SVM),multilayer perceptron(MLP),random forest(RF)and gradient boosting decision tree(GBDT)as the primary learner and multiple linear re-gression(MLR)as the secondary learner.The simulation accuracy of the LSM model was compared with that of MLR,SVM,MLP,RF and GBDT models.The results show that the main factors affecting evapotranspiration of apple trees were the daily average solar radiation,relative humidity,wind speed,temperature and daily ordinal number.The maxi-mal information coefficient values were 0.97,0.72,0.63,0.62 and 0.60.Surface soil temperature and soil moisture content had little impact on evapotranspiration.Compared with MLR,SVM,MLP,RF and GBDT models,the LSM model had the highest simulation accuracy,and the MLR model had the lowest simulation accuracy.The five characteris-tics parameters of the daily average solar radiation,relative humidity,wind speed,temperature and daily ordinal number could accurately simulate the evapotranspiration of apple trees while reducing the acquisition cost of features.This study can provide an effective method for accurate simulation of daily evapotranspiration of apple trees.
万方数据
维普
