基于辐射改进Penman-Monteith模型估算粮食主产区参考作物蒸散量  

Estimation of Reference Crop Evapotranspiration in Major Grain-producing Areas Based on Improved Penman-Monteith Model

作　　者：马钊 李鹏程[2,3] 刘洪伟 孟静 MA Zhao;LI Peng-cheng;LIU Hong-wei;MENG Jing(Shandong Provincial Water Resources Survey and Design Institute Co.,Ltd,Jinan 250013,China;School of Geographic Sciences,Hebei Normal University,Shijiazhuang 050010,China;Hebei University of Water Resources and Electric Engineering,Cangzhou 061001,Hebei Province,China;Water Resources Survey and Design Office of Taierzhuang District,Zaozhuang 277000,Shandong Province,China)

机构地区：[1]山东省水利勘测设计院有限公司,济南250013 [2]河北师范大学地理科学学院,石家庄050010 [3]河北水利电力学院,河北沧州061001 [4]枣庄市台儿庄区水利勘测设计室,山东枣庄277000

出　　处：《节水灌溉》2024年第3期24-33,共10页Water Saving Irrigation

基　　金：国家科技支撑计划项目(31410005321403);河北省高等学校科学技术研究项目资助(BJ2021256);2020年河北水利电力学院河北省高等学校基本科研业务费研究项目(202004)。

摘　　要：为进一步提高Penman-Monteith模型估算参考作物蒸散量(Reference crop evapotranspiration,ET0)的精度,以中国粮食主产区为研究对象,将其划分为温带湿润半湿润地区(THSZ)、温带干旱半干旱地区(TASZ)、暖温带半湿润地区(WTSZ)和亚热带湿润地区(SHZ),基于32个气象站点1994-2020年长序列实测逐日气象数据,将猎豹算法(CO)、沙猫算法(SCSO)、野狗算法(DOA)优化的时间卷积神经网络模型(TCN)和3种基于日照时数、3种基于温度的经验模型估算的辐射(R_(s))值与PM模型进行融合,得到改进PM模型。以均方根误差(RMSE)、决定系数(R^(2))、平均绝对误差(MAE)和效率系数(E_(NS))为精度评价体系,找出了粮食主产区不同分区的ET0最优估算模型,结果表明:基于日照时数模型的计算精度要优于温度模型,其中CO-TCN模型在全区内均表现出了较高的精度,在不同分区的RMSE、MAE、R^(2)和E_(NS)中位数取值分别为0.099~0.171 mm/d、0.057~0.111mm/d、0.984~0.998、0.983~0.997,由此可将CO-TCN模型估算的辐射值与PM模型融合,作为标准值用于估算粮食主产区ET0。To further improve the accuracy of estimating regional reference crop evapotranspiration(ET0)by Penman-Monteith model,we focused on the main grain producing areas in China.The region was stratified into temperate humid and semi-humid areas(THSZ),temperate arid and semi-arid areas(TASZ),warm temperate and semi-humid areas(WTSZ),and subtropical humid areas(SHZ).Based on the daily meteorological data from 32 meteorological stations spanning the period from 1994 to 2020,a time Convolutional neural network model(TCN)optimized by Cheetah algorithm(CO),Sand Cat algorithm(SCSO)and Wild dog algorithm(DOA)were employed.Additionally,radiation(R_(s))values estimated by 3 kinds of sunshine hours empirical models and 3 kinds of temperature empirical models were merged with the PM model to get an improved PM model.Root mean square error(RMSE),determination coefficient(R^(2)),mean absolute error(MAE),and efficiency coefficient(E_(NS))were used as precision evaluation systems.The results showed that:the calculation accuracy of the sunshine hours models was better than that of the temperature models.The CO-TCN models showed the highest accuracy in the whole region.The median values of RMSE,MAE,R^(2),and E_(NS)in different partitions are 0.099~0.171 mm/d,0.057~0.111 mm/d,0.984~0.998,and 0.983~0.997,respectively.The radiation values estimated by the CO-TCN model can be integrated with the PM model as standard values for estimating ET0 in the main grain-producing areas.

关 键 词：粮食主产区 参考作物蒸散量 辐射 Penman-Monteith模型 时间卷积神经网络 

分 类 号：S271[农业科学—农业水土工程]