基于不同ELM的西北旱区参考作物蒸散量模拟模型
|
摘要
参考作物蒸散量(reference crop evapotranspiration,ET0)的精准模拟是智慧灌溉、农田水分高效利用及水资源优化调度的重要依据。为有效提高气象资料缺乏情况下对西北旱区ET0的模拟精度,在西北旱区选取4个代表性站点,并以FAO 56 Penman-Monteith(P-M)模型的ET0计算结果为标准值,基于"sin"、"radbas"和"hardlim"3种激活函数构建27种极限学习机模型(extreme learning machine,ELM,分别记为ELM-sini、ELM-radj、ELM-hardk),并将其模拟结果与Hargreaves-Samani(H-S)、Makkink(MK)、Irmark-Allen(I-A)模型进行比较。结果表明:ELM-sin7(输入u2、Tmax和Tmin)的R2和NSE均大于0.96,RMSE小于0.35 mm/d,其GPI排名第4,模型模拟精度较高; ELM-rad5(输入Tmax、Tmin和n)和ELM-sin8(输入Tmax和Tmin)的R2和NSE分别大于0.78和0.76,RMSE小于0.93 mm/d; H-S、MK和I-A模型的R2和NSE分别小于0.77和0.63,RMSE大于1.00 mm/d,可见ELM-rad5和ELM-sin8模型精度明显高于相同输入下的其他物理模型;基于ELM-sin7探究隐含层节点数对模型精度的影响发现隐含层节点数为60～100时模型精度最高;基于ELMsin7模型进行可移植性分析发现,ELM-sin7在西北旱区内各训练站点和模拟站点组合下模拟精度较高。因此,在相同气象因子组合输入下,ELM-sini和ELM-radj模型模拟精度明显高于ELM-hardk,其中ELM-sin7模拟精度较高适用于西北旱区气象因子较少时的ET0模拟;而较传统物理模型,仅有温度和日照时数时ELM-rad5模型在西北旱区适用性更好,仅有温度时ELM-sin8模型在西北旱区适用性更强。
The accurate simulation of reference crop evapotranspiration( ET0) is an important basis for intelligent irrigation,efficient utilization of water resources in farmland and optimal scheduling of water resources. With the lack of meteorological data,to enhance the simulation accuracy of ET0 in the arid regions of Northwest China effectively,4 representative sites are selected,the ET0 calculations of the FAO 56 Penman Monteith( P-M) model are set as the values of standard,and then 27 kinds of extreme learning machine model( extreme learning machine,ELM,respectively for the ELM-sini,ELM-radj,ELM-hardk) are built based on 3 activation functions( "sin""radbas"and"hardlim"). Next,the results of the simulation are compared together with the models of Hargreaves-Samani( H-S),Makkink( MK)and Irmark-Allen( I-A). The results are as follows: in the case of ELM-sin7( input u2,Tmaxand Tmin) model,the value of R2 and NSE are more than 0.96,while the RMSE is less than 0.35 mm/d. This model gets the fourth greatest GPI among all the models,indicating its higher accuracy. For the models of ELM-rad5( input Tmax,Tmin and n) and ELM-sin8( input Tmaxand Tmin),the values of R2 and NSE are more than 0.78 and 0.76,while RMSE is less than 0.93 mm/d. As for H-S,MK and I-A models,the values of R2 and NSE are less than 0.77 and 0.63,while their RMSE is more than 1.00 mm/d. Thus,the ELM-rad5 and ELM-sin8 models are obviously more accurate compared with the other physical models under the same input. What's more,it is discovered that with the ELM-sin7 model,when the number of hidden layer nodes are from 60 to 100,the model has the highest accuracy,and the results of ELM-sin7 model portability show that ELM-sin7 model in the arid north-western China has simulation accuracy based on different sites' data. Therefore,under the same input of meteorological factor combination,the simulation accuracy of ELM-siniand ELM-radjmodels are remarkably higher than that of ELM-hardk. And ELMsin7 model has the highest accuracy among those models,which can be suitable for the ET0 simulation in the arid northwest area when the meteorological factors are not rich. In addition,compared with the traditional physical models,the ELM-rad5 model has a better applicability in the arid northwest when just the input temperature and the hours of sunshine are known. And the ELM-sin8 model is more suitable when temperature is the only meteorological factor provided.
The accurate simulation of reference crop evapotranspiration( ET0) is an important basis for intelligent irrigation,efficient utilization of water resources in farmland and optimal scheduling of water resources. With the lack of meteorological data,to enhance the simulation accuracy of ET0 in the arid regions of Northwest China effectively,4 representative sites are selected,the ET0 calculations of the FAO 56 Penman Monteith( P-M) model are set as the values of standard,and then 27 kinds of extreme learning machine model( extreme learning machine,ELM,respectively for the ELM-sini,ELM-radj,ELM-hardk) are built based on 3 activation functions( "sin""radbas"and"hardlim"). Next,the results of the simulation are compared together with the models of Hargreaves-Samani( H-S),Makkink( MK)and Irmark-Allen( I-A). The results are as follows: in the case of ELM-sin7( input u2,Tmaxand Tmin) model,the value of R2 and NSE are more than 0.96,while the RMSE is less than 0.35 mm/d. This model gets the fourth greatest GPI among all the models,indicating its higher accuracy. For the models of ELM-rad5( input Tmax,Tmin and n) and ELM-sin8( input Tmaxand Tmin),the values of R2 and NSE are more than 0.78 and 0.76,while RMSE is less than 0.93 mm/d. As for H-S,MK and I-A models,the values of R2 and NSE are less than 0.77 and 0.63,while their RMSE is more than 1.00 mm/d. Thus,the ELM-rad5 and ELM-sin8 models are obviously more accurate compared with the other physical models under the same input. What's more,it is discovered that with the ELM-sin7 model,when the number of hidden layer nodes are from 60 to 100,the model has the highest accuracy,and the results of ELM-sin7 model portability show that ELM-sin7 model in the arid north-western China has simulation accuracy based on different sites' data. Therefore,under the same input of meteorological factor combination,the simulation accuracy of ELM-siniand ELM-radjmodels are remarkably higher than that of ELM-hardk. And ELMsin7 model has the highest accuracy among those models,which can be suitable for the ET0 simulation in the arid northwest area when the meteorological factors are not rich. In addition,compared with the traditional physical models,the ELM-rad5 model has a better applicability in the arid northwest when just the input temperature and the hours of sunshine are known. And the ELM-sin8 model is more suitable when temperature is the only meteorological factor provided.
引文
[1]丁加丽,彭世彰,徐俊增,等.基于温度资料的参考作物蒸发蒸腾量计算方法[J].河海大学学报:自然科学版,2007,(6):633-637.
[2]陈玉民,郭国双.中国主要作物需水量与灌溉[M].北京:中国水利水电出版社,1995.
[3] Richard G.Crop evapotranspiration; Guidelines for computing Crop Water Requirements[J].Fao Irrigation&Drainage Paper,1998,56.
[4] Hargreaves G H. Reference crop evapotranspiration from temperature[J].Applied Eng in Agric,1985,1(2).
[5] Blaney H F.Determining water requirements in irrigated areas from climatological and irrigation data[R]. T3chnical Paper No.96,US Department of Agriculture,Soil Conservation Service,Washington,DC,1950,48.
[6] Priestley C H B,Taylor R J.On the Assessment of Surface Heat Flux and Evaporation Using Large-Scale Parameters[J]. Monthly Weather Review,1972,100(2):81-92.
[7]李志.参考作物蒸散简易估算方法在黄土高原的适用性[J].农业工程学报,2012,(6):106-111.
[8] Irmak S,Irmak A,Allen R G,et al.Solar and Net Radiation-Based Equations to Estimate Reference Evapotranspiration in Humid Climates[J]. Journal of Irrigation&Drainage Engineering,2003,129(5):336-347.
[9] Kumar M,Raghuwanshi N S,Siongh R. Estimating evapotranspiration using artificial neuaral network[J]. Journal of Irrigation and Drainage Engineering,ASCE,2002,128(4):224-233.
[10] Landeras G,Ortiz-Barredo A,López J J. Comparison of artificial neural network models and empirical and semi-empirical equations for daily reference evapotranspiration estimation in the Basque Country(Northern Spain)[J]. Agricultural Water Management,2008,95(5):553-565.
[11] Tabari H,Kisi O,Ezani A,et al.SVM,ANFIS,regression and climate based models for reference evapotranspiration modeling using limited climatic data in a semi-arid highland environment[J].Journal of Hydrology,2012,444-445(10):78-89.
[12]崔远来,马承新,沈细中,等.基于进化神经网络的参考作物腾发量模拟[J].水科学进展,2005,16(1):76-81.
[13]张育斌,魏正英,马胜利,等.极端天气下作物参照蒸散量计算方法研究[J].中国农村水利水电,2014,(12):64-68.
[14] Abdullah S S,Malek M A,Abdullah N S,et al.Extreme Learning Machines:A new approach for prediction of reference evapotranspiration[J].Journal of Hydrology,2015,527:184-195.
[15] Feng Yu,Cui Ningbo,Zhao Lu,et al.Comparison of ELM,GANN,WNN and empirical models for estimating reference evapotranspiration in humid region of Southwest China[J].Quaternary International,2016,441:129-139.
[16] Feng Yu,Peng Yong,Cui Ningbo,et al.Modeling reference evapotranspiration using extreme learning machine and generalized regression neural network only with temperature data[J]. Computers and Elctronics in Agriculture,2017,136,71-78.
[17]邹伟东,张百海,姚分喜,等.基于改进型极限学习机的日光温室温湿度模拟与验证[J].农业工程学报,2015,31(24):194-200.
[18]刘海启,王迎春.我国西北旱区农牧业综合开发潜力分析及水资源支撑能力研究[J].中国农业资源与区划,2015,(3):1-9.
[19]苑津莎,张利伟,王瑜,等.基于极限学习机的变压器故障诊断方法研究[J].电测与仪表,2013,(12):21-26.
[20]王小川.MATLAB神经网络43个案例分析[M].北京:北京航空航天大学出版社,2013.
[21] Jamil B,Akhtar N. Estimation of diffuse solar radiation in humidsubtropical climatic region of India:Comparison of diffuse fraction and diffusion coefficient models[J].Energy,2017,131:149-164.
[22]汪彪,曾新民,刘正奇,等.中国西北地区参考作物蒸散量的估算与变化特征[J].干旱气象,2016,(2):243-251.
[23]刘宪锋,潘耀忠,张锦水,等.1960—2011年西北五省潜在蒸散的时空变化[J].应用生态学报,2013,(9):2 564-2 570.
[24] Eslamian S,Khordadi M J,Abedi-Koupai J.Effects of variations in climatic parameters on evapotranspiration in the arid and semi-arid regions[J].Global&Planetary Change,2011,78(3-4):188-194.
[25]张青雯,崔宁博,冯禹,等.中国西南五省参考作物蒸散量时空变化分析[J].灌溉排水学报,2016,35(11):80-87.
[26]曹雯,申双和,段春锋.西北地区生长季参考作物蒸散发的敏感性分析[C]∥中国气象学会年会现代农业气象防灾减灾与粮食安全分会场,2010.
[2]陈玉民,郭国双.中国主要作物需水量与灌溉[M].北京:中国水利水电出版社,1995.
[3] Richard G.Crop evapotranspiration; Guidelines for computing Crop Water Requirements[J].Fao Irrigation&Drainage Paper,1998,56.
[4] Hargreaves G H. Reference crop evapotranspiration from temperature[J].Applied Eng in Agric,1985,1(2).
[5] Blaney H F.Determining water requirements in irrigated areas from climatological and irrigation data[R]. T3chnical Paper No.96,US Department of Agriculture,Soil Conservation Service,Washington,DC,1950,48.
[6] Priestley C H B,Taylor R J.On the Assessment of Surface Heat Flux and Evaporation Using Large-Scale Parameters[J]. Monthly Weather Review,1972,100(2):81-92.
[7]李志.参考作物蒸散简易估算方法在黄土高原的适用性[J].农业工程学报,2012,(6):106-111.
[8] Irmak S,Irmak A,Allen R G,et al.Solar and Net Radiation-Based Equations to Estimate Reference Evapotranspiration in Humid Climates[J]. Journal of Irrigation&Drainage Engineering,2003,129(5):336-347.
[9] Kumar M,Raghuwanshi N S,Siongh R. Estimating evapotranspiration using artificial neuaral network[J]. Journal of Irrigation and Drainage Engineering,ASCE,2002,128(4):224-233.
[10] Landeras G,Ortiz-Barredo A,López J J. Comparison of artificial neural network models and empirical and semi-empirical equations for daily reference evapotranspiration estimation in the Basque Country(Northern Spain)[J]. Agricultural Water Management,2008,95(5):553-565.
[11] Tabari H,Kisi O,Ezani A,et al.SVM,ANFIS,regression and climate based models for reference evapotranspiration modeling using limited climatic data in a semi-arid highland environment[J].Journal of Hydrology,2012,444-445(10):78-89.
[12]崔远来,马承新,沈细中,等.基于进化神经网络的参考作物腾发量模拟[J].水科学进展,2005,16(1):76-81.
[13]张育斌,魏正英,马胜利,等.极端天气下作物参照蒸散量计算方法研究[J].中国农村水利水电,2014,(12):64-68.
[14] Abdullah S S,Malek M A,Abdullah N S,et al.Extreme Learning Machines:A new approach for prediction of reference evapotranspiration[J].Journal of Hydrology,2015,527:184-195.
[15] Feng Yu,Cui Ningbo,Zhao Lu,et al.Comparison of ELM,GANN,WNN and empirical models for estimating reference evapotranspiration in humid region of Southwest China[J].Quaternary International,2016,441:129-139.
[16] Feng Yu,Peng Yong,Cui Ningbo,et al.Modeling reference evapotranspiration using extreme learning machine and generalized regression neural network only with temperature data[J]. Computers and Elctronics in Agriculture,2017,136,71-78.
[17]邹伟东,张百海,姚分喜,等.基于改进型极限学习机的日光温室温湿度模拟与验证[J].农业工程学报,2015,31(24):194-200.
[18]刘海启,王迎春.我国西北旱区农牧业综合开发潜力分析及水资源支撑能力研究[J].中国农业资源与区划,2015,(3):1-9.
[19]苑津莎,张利伟,王瑜,等.基于极限学习机的变压器故障诊断方法研究[J].电测与仪表,2013,(12):21-26.
[20]王小川.MATLAB神经网络43个案例分析[M].北京:北京航空航天大学出版社,2013.
[21] Jamil B,Akhtar N. Estimation of diffuse solar radiation in humidsubtropical climatic region of India:Comparison of diffuse fraction and diffusion coefficient models[J].Energy,2017,131:149-164.
[22]汪彪,曾新民,刘正奇,等.中国西北地区参考作物蒸散量的估算与变化特征[J].干旱气象,2016,(2):243-251.
[23]刘宪锋,潘耀忠,张锦水,等.1960—2011年西北五省潜在蒸散的时空变化[J].应用生态学报,2013,(9):2 564-2 570.
[24] Eslamian S,Khordadi M J,Abedi-Koupai J.Effects of variations in climatic parameters on evapotranspiration in the arid and semi-arid regions[J].Global&Planetary Change,2011,78(3-4):188-194.
[25]张青雯,崔宁博,冯禹,等.中国西南五省参考作物蒸散量时空变化分析[J].灌溉排水学报,2016,35(11):80-87.
[26]曹雯,申双和,段春锋.西北地区生长季参考作物蒸散发的敏感性分析[C]∥中国气象学会年会现代农业气象防灾减灾与粮食安全分会场,2010.