基于ELM的西北旱区参考作物蒸散量预报模型

为实现气象资料缺失情况下ET₀的精确预报,选取中国西北旱区4个代表性站点的气象资料,建立15种基于极限学习机(ELM)的ET₀预报模型,并通过与其他ET₀计算模型对比和可移植性分析探究ELM在西北旱区的适用性.结果表明:基于温度和风速的ELM₇预报精度较高(整体评价指标GPI排名第4);基于温度和辐射的ELM₅预报精度(GPI排名第6)明显高于Iramk模型和Jensen-Haise模型;仅基于温度的ELM₉预报精度(GPI排名第8)高于Hargreaves-Samani模型.通过模型可移植性分析发现,ELM₇在西北旱区内各训练站点和预测站点组合下预报精度良好.因此,可将ELM₅(输入温度和辐射)、ELM₇(输入温度和风速)和ELM₉(输入温度)作为西北旱区较少气象参数输入情况下精确预报ET₀的推荐模型.

Prediction for reference crop evapotranspiration in arid northwest China based on ELM

To predict ET₀ accurately with limited meteorological data, four representative meteorological sites were selected in arid northwest China to construct extreme learning machine(ELM). This model was used to predict ET₀ and compare with other three models. The results show that when only based on the temperature and the wind speed data, the simulation accuracy of ELM₇ is high(GPI ranks the 4th). When only based on the temperature and the radiation data, the simulation accuracy of ELM₅ model(GPI ranks the 6th)is greater than that of Iramk model and Jensen-Haise model. When only based on the temperature data, the simulation accuracy of ELM₉ model(GPI ranks the 8th)is greater than that of Hargreaves-Samani model. The results of ELM model portability show that ELM₇ model in arid northwestern China has high simulation accuracy based on different sites′ datas. Thus, ELM₅, ELM₇ and ELM₉ models can be used as a recommended model for the prediction of ET₀ based on limited meteorological data in the northwest arid area of China.