淮北平原基于ARIMA模型的冬小麦日土壤水分预测
|
摘要
准确预测土壤水分动态变化对农作物生长以及节水灌溉至关重要。为反映逐日土壤水分动态变化,利用五道沟水文实验站蒸渗仪2017-2018年土壤水实测资料,采用时间序列分析方法,分别建立了冬小麦全生育期10、30、50 cm土层的土壤水分计算模型。结果表明:10、30、50 cm土壤含水量变异系数有明显差异,随土层深度增加逐渐减小,分别为0.190、0.103、0.040。利用ARIMA模型对土壤水分进行拟合,10、30、50 cm土层土壤水分计算模型分别为ARIMA(4,1,7)、ARIMA(1,1,2)、ARIMA(2,1,3),拟合优度R~2均大于0.95;不同土层土壤水分计算模型均具有较好的预测能力,且随深度增加预测精度提高,由10 cm增至50 cm最大相对误差从15.6%降至5.1%。研究成果为进一步制定淮北平原节水灌溉制度,提高田间水利用率具有重要意义。
Accurate prediction of soil moisture dynamics is crucial for crop growth and water-saving irrigation. In order to reflect the daily dynamic changes of soil moisture,the calculation models of soil moisture in 10,30 and 50 cm soil layers during the whole growth period of winter wheat were established by using time series analysis method and the actual measured data of soil water from 2017 to 2018 in Wudaogou hydrological experimental station. The results showed that the variation coefficient of soil moisture content at 10,30 and 50 cm was significantly different and gradually decreased with the increase of soil depth,which was 0. 190,0. 103 and 0. 040,respectively. ARIMA model was used to fit soil moisture. Calculation models of soil moisture in 10,30 and 50 cm soil layers were ARIMA( 4,1,7),ARIMA( 1,1,2) and ARIMA( 2,1,3),respectively. R~2 of goodness of fit was greater than 0.95.The calculation models of soil moisture in different soil layers had good prediction ability,and the prediction accuracy increased with the increase of depth,and the maximum relative error increased from 10 cm to 50 cm,and decreased from 15.6% to 5.1%. The research results are of great significance to further formulate the water-saving irrigation system in Huaibei plain and improve the utilization rate of field water.
Accurate prediction of soil moisture dynamics is crucial for crop growth and water-saving irrigation. In order to reflect the daily dynamic changes of soil moisture,the calculation models of soil moisture in 10,30 and 50 cm soil layers during the whole growth period of winter wheat were established by using time series analysis method and the actual measured data of soil water from 2017 to 2018 in Wudaogou hydrological experimental station. The results showed that the variation coefficient of soil moisture content at 10,30 and 50 cm was significantly different and gradually decreased with the increase of soil depth,which was 0. 190,0. 103 and 0. 040,respectively. ARIMA model was used to fit soil moisture. Calculation models of soil moisture in 10,30 and 50 cm soil layers were ARIMA( 4,1,7),ARIMA( 1,1,2) and ARIMA( 2,1,3),respectively. R~2 of goodness of fit was greater than 0.95.The calculation models of soil moisture in different soil layers had good prediction ability,and the prediction accuracy increased with the increase of depth,and the maximum relative error increased from 10 cm to 50 cm,and decreased from 15.6% to 5.1%. The research results are of great significance to further formulate the water-saving irrigation system in Huaibei plain and improve the utilization rate of field water.
引文
[1]冀荣华,张舒蕾,郑立华,等.基于多值神经元复数神经网络的土壤水分预测[J].农业工程学报,2017,33(S1):126-131.
[2]侯晓丽,冯跃华,吴光辉,等.基于人工神经网络土壤水分动态预测模型应用研究[J].节水灌溉,2016(7):70-72,76.
[3]孙景生,康绍忠.我国水资源利用现状与节水灌溉发展对策[J].农业工程学报,2000,16(2):1-5.
[4]蔡甲冰,刘钰,雷廷武,等.精量灌溉决策定量指标研究现状与进展[J].水科学进展,2004,15(4):531-537.
[5]黄治平,王铄今,侯彦林,等.统计法土壤水分诊断模型[J].生态学杂志,2017,36(12):3 345-3 351.
[6]冀荣华,李鑫,张舒蕾,等.基于GM(1,1)模型群的农田土壤水分短期预测[J].农业机械学报,2016,47(S1):401-407.
[7]霍轶珍,李生勇,郭彦芬.ISAREG模型在灌区土壤水分预报中的应用研究[J].人民黄河,2017,39(12):157-160.
[8]吕庆玉.河北省平原区土壤水分预报研究[J].水科学与工程技术,2012(S1):113-117.
[9]陈晓,赵悦华,安国庆.衡水市土壤水分变化规律及预报分析[J].水科学与工程技术,2014(6):49-52.
[10]王晓燕,陈洪松,王克林,等.红壤坡地土壤水分时间序列分析[J].应用生态学报,2007,18(2):297-302.
[11]张和喜,杨静,方小宇,等.时间序列分析在土壤水分预测中的应用研究[J].水土保持研究,2008,15(4):82-84.
[12]杨绍辉,王一鸣,郭正琴,等.ARIMA模型预测土壤水分研究[J].干旱地区农业研究,2006,24(2):114-118.
[13]李军,张和喜,蒋毛席,等.基于ARIMA模型的贵州省黄壤水分预测研究[J].人民黄河,2010,32(8):73-75.
[14]周振民,刘荻.基于Matlab的人工神经网络用水量预测模型[J].中国农村水利水电,2007(4):45-47,49.
[15]马建成,黄诗峰,胡健伟,等.基于MODIS数据的山东省土壤水分遥感动态监测与分析[J].水文,2013,33(3):29-33,42.
[16]刘洪斌,武伟,魏朝富,等.土壤水分预测神经网络模型和时间序列模型比较研究[J].农业工程学报,2003,19(4):33-36.
[17]哈建强,祝明,朱艳飞.沧州市土壤水分变化规律研究[J].水文,2017,37(6):74-79.
[18]尚松浩,毛晓敏,雷志栋,等.冬小麦田间水分预报的BP神经网络模型[J].水利学报,2002(4):60-63,68.
[19] Mohammad Mahfuzur Rahman,Minjiao Lu,Khin Htay Kyi. Variability of soil moisture memory for wet and dry basins[J]. Journal of Hydrology,2015(523):107-118.
[2]侯晓丽,冯跃华,吴光辉,等.基于人工神经网络土壤水分动态预测模型应用研究[J].节水灌溉,2016(7):70-72,76.
[3]孙景生,康绍忠.我国水资源利用现状与节水灌溉发展对策[J].农业工程学报,2000,16(2):1-5.
[4]蔡甲冰,刘钰,雷廷武,等.精量灌溉决策定量指标研究现状与进展[J].水科学进展,2004,15(4):531-537.
[5]黄治平,王铄今,侯彦林,等.统计法土壤水分诊断模型[J].生态学杂志,2017,36(12):3 345-3 351.
[6]冀荣华,李鑫,张舒蕾,等.基于GM(1,1)模型群的农田土壤水分短期预测[J].农业机械学报,2016,47(S1):401-407.
[7]霍轶珍,李生勇,郭彦芬.ISAREG模型在灌区土壤水分预报中的应用研究[J].人民黄河,2017,39(12):157-160.
[8]吕庆玉.河北省平原区土壤水分预报研究[J].水科学与工程技术,2012(S1):113-117.
[9]陈晓,赵悦华,安国庆.衡水市土壤水分变化规律及预报分析[J].水科学与工程技术,2014(6):49-52.
[10]王晓燕,陈洪松,王克林,等.红壤坡地土壤水分时间序列分析[J].应用生态学报,2007,18(2):297-302.
[11]张和喜,杨静,方小宇,等.时间序列分析在土壤水分预测中的应用研究[J].水土保持研究,2008,15(4):82-84.
[12]杨绍辉,王一鸣,郭正琴,等.ARIMA模型预测土壤水分研究[J].干旱地区农业研究,2006,24(2):114-118.
[13]李军,张和喜,蒋毛席,等.基于ARIMA模型的贵州省黄壤水分预测研究[J].人民黄河,2010,32(8):73-75.
[14]周振民,刘荻.基于Matlab的人工神经网络用水量预测模型[J].中国农村水利水电,2007(4):45-47,49.
[15]马建成,黄诗峰,胡健伟,等.基于MODIS数据的山东省土壤水分遥感动态监测与分析[J].水文,2013,33(3):29-33,42.
[16]刘洪斌,武伟,魏朝富,等.土壤水分预测神经网络模型和时间序列模型比较研究[J].农业工程学报,2003,19(4):33-36.
[17]哈建强,祝明,朱艳飞.沧州市土壤水分变化规律研究[J].水文,2017,37(6):74-79.
[18]尚松浩,毛晓敏,雷志栋,等.冬小麦田间水分预报的BP神经网络模型[J].水利学报,2002(4):60-63,68.
[19] Mohammad Mahfuzur Rahman,Minjiao Lu,Khin Htay Kyi. Variability of soil moisture memory for wet and dry basins[J]. Journal of Hydrology,2015(523):107-118.