基于5种人工智能模型计算重庆地区参考作物蒸散量

(1.重庆市水利电力建筑勘测设计研究院, 重庆 400020; 2.西南大学 资源环境学院, 重庆 400715)

重庆; 参考作物蒸散量; 人工智能模型; 经验模型; 日照时数

Calculation of Reference Crop Evapotranspiration in Chongqing Based on 5 Artificial Intelligent Models
BAO Lingling1, YANG Yonggang1, LIU Jianjun1, ZHANG Weihua2

(1.Chongqing Surveying and Design Institute of Water Resources, Electric Power and Architecture, Chongqing 400020, China; 2.College of Resources and Environment, Southwest University, Chongqing 400715, China)

Chongqing; reference crop evapotranspiration; artificial intelligent models; empirical model; sunshine hours

备注

为获得计算重庆地区参考作物蒸散量(Reference crop evapotranspiration,ET0)的最优模型,选用支持向量机模型(SVM)、高斯指数模型(GEM)、随机森林模型(RF)、极限学习机模型(ELM)和广义回归神经网络模型(GRNN)5种人工智能模型,以丰都、奉节、沙坪坝、万州、酉阳共5个站点1991—2016年的逐日气象数据为基础,估算ET0日值、月值,并与Penman-Monteith(P-M)计算结果进行了对比,结果表明:不同模型精度存在差异,在相同气象参数输入的情况下,人工智能模型计算精度要高于经验模型,在相同参数输入的情况下,GEM模型误差指标最低而一致性指标最高,日照时数n是影响重庆地区ET0变化和影响模型精度的最关键因素,而GEM模型为重庆地区ET0估算的最优人工智能模型。

In order to obtain the optimal model for calculating reference crop evapotranspiration(ET0)in Chongqing, the five artificial intelligent models such as support vector machines(SVM), Gaussian exponential model(GEM), random forest model(RF), extreme learning machine model(ELM)and generalized regression neural network model(GRNN)were used as the calculation models. Based on the daily meteorological data from Fengdu, Fengjie, Shapingba, Wanzhou, and Youyang from 1991 to 2016, the daily and monthly ET0 under different combinations of meteorological parameter inputs were estimated, and compared with the calculation results of the standard model Penman-Monteith(PM). The results show that the accuracies of different models are different; under the input of the same meteorological parameters, the calculation accuracy of the artificial intelligence model is higher than that of the empirical model; the error index of the GEM model is the lowest and the consistency index is the highest. Sunshine duration is the most critical factor impacting the accuracy of modeling and change in ET0 in Chongqing. GEM model is the optimal artificial intelligence model to estimate ET0 in Chongqing.