基于灰色关联法的月降雨量预测
|
摘要
【目的】提高降雨量预测精度,为农业、水利等相关部门提供决策依据。【方法】鉴于月降雨量时间序列具有显著的多尺度特征,开展了数据驱动下基于模型融合的月降雨量预测研究,应用灰色EGM(1,1)模型和自适应模糊神经网络系统(ANFIS)分别预测了年尺度与月尺度下的月降雨量,采用灰色关联法将2个预测结果进行数据融合。利用澳大利亚维多利亚8个站点降雨数据验证所提出方法,并将预测结果进行了与单一灰色EGM(1,1)、ANFIS、人工神经网络(ANN)、自回归积分滑动平均模型(ARIMA)与聚类回归法(CLR)模型预测结果对比。【结果】模型融合预测结果精度高于单一EGM(1,1)、ANFIS、ANN及ARIMA模型预测结果,并在8个站点中的5个取得了最佳预测效果,其中中部地区(Ballarat和Cape Otway站点)及东部地区(Dookie,Wangaratta和Orbost站)预测均方根误差为28.2～37.2 mm,西部地区(Dimboola,Edenhope和Dunkeld站点)预测均方根误差为20.8～23.4 mm。【结论】所提出的模型融合预测法可行,为月降雨量预测提供了新思路。
【Objective】Precipitation is the main input to many catchments and ecosystems, and catchment management and design of irrigation schedule need to predict it. This paper presents a method to predict monthly precipitation.【Method】Considering that temporal change of monthly precipitation is nonlinear and multi-scale, this paper proposed a data-driven based fusion method, in which the grey model(EGM) and the adaptive neuro-fuzzy inference system(ANFIS) were used to predict the rainfall at yearly and monthly scales. The two predictions were then fused by the grey correlation method. The model was evaluated against the rainfall data collected from eight sites in Vitoria of Australia. We also compared the proposed model with the so-called sole prediction models which include EGM(1,1), ANFIS, artificial neural network(ANN) and the autoregressive integrated moving average(ARIMA).【Result】The proposed model was more accurate than EGM(1,1), ANFIS and ANN and ARIMA model, giving the most accurate predication for five out of the eight sites, in which the root mean square errors of the proposed model for the central region(sites at Ballarat and Cape Otway) and the eastern region(Dookie, Wangaratta and Orbost stations) was 28.2 to 37.2 mm respectively, while for the western region(Dimboola, Edenhope and Dunkeld sites) it was 20.8 to 37.2 mm.【Conclusion】The proposed model is adequate and reliable, offering an alternative to predict monthly precipitation.
【Objective】Precipitation is the main input to many catchments and ecosystems, and catchment management and design of irrigation schedule need to predict it. This paper presents a method to predict monthly precipitation.【Method】Considering that temporal change of monthly precipitation is nonlinear and multi-scale, this paper proposed a data-driven based fusion method, in which the grey model(EGM) and the adaptive neuro-fuzzy inference system(ANFIS) were used to predict the rainfall at yearly and monthly scales. The two predictions were then fused by the grey correlation method. The model was evaluated against the rainfall data collected from eight sites in Vitoria of Australia. We also compared the proposed model with the so-called sole prediction models which include EGM(1,1), ANFIS, artificial neural network(ANN) and the autoregressive integrated moving average(ARIMA).【Result】The proposed model was more accurate than EGM(1,1), ANFIS and ANN and ARIMA model, giving the most accurate predication for five out of the eight sites, in which the root mean square errors of the proposed model for the central region(sites at Ballarat and Cape Otway) and the eastern region(Dookie, Wangaratta and Orbost stations) was 28.2 to 37.2 mm respectively, while for the western region(Dimboola, Edenhope and Dunkeld sites) it was 20.8 to 37.2 mm.【Conclusion】The proposed model is adequate and reliable, offering an alternative to predict monthly precipitation.
引文
[1]薛胜军,张佩云,陈静怡.云平台下基于半朴素贝叶斯的降雨量预测[J].系统仿真学报,2016,28(5):1 117-1 123.
[2] CHAU K W, WU C L. A hybrid model coupled with singular spectrum analysis for daily rainfall prediction[J]. Journal of Hydroinformatics, 2010, 12(4):458-473.
[3] NARAYANAN P, BASISTHAA, SARKAR S, et al. Trend analysis andARIMAmodelling of pre-monsoon rainfall data for western India[J]. Comptesrendus Géoscience, 2013, 345(1):22-27.
[4] KIM B S, HOSSEIN S Z, CHOI G. Evaluation of temporal-spatial precipitation variability and prediction using seasonal ARIMA model in Mongolia[J].Ksce Journal of Civil Engineering, 2011, 15(5):917-925.
[5] FAULINA R. Hybrid ARIMA-ANFIS for Rainfall Prediction in Indonesia[J]. International Journal of Science&Research, 2013.
[6] ABBOT J, MAROHASY J. Input selection and optimisation for monthly rainfall forecasting in Queensland, Australia, using artificial neural networks[J]. Atmospheric ReseIarch, 2014, 138(2):166-178.
[7] DEO R C, SAHN M. Application of the Artificial Neural Network model for prediction of monthly Standardized Precipitation and Evapotranspiration Index using hydrometeorological parameters and climate indices in eastern Australia[J]. Atmospheric Research, 2015, 161/162:65-81.
[8] BAGIROV A M, MAHMOOD A, BARTON A. Prediction of monthly rainfall in Victoria, Australia:Cluster wise linear regression approach[J].Atmospheric Research, 2017, 188:20-29.
[9]郑玉胜,黄介生.基于神经网络的灌溉用水量预测[J].灌溉排水学报, 2004, 23(2):59-61.
[10]廖凯华,徐绍辉,吴吉春,等.基于PCR和ANN构建的土壤转换函数的适用性研究[J].灌溉排水学报, 2014, 33(1):17-21.
[11]王翠翠,魏永霞.基于GM(1,1)-AR模型和AR模型的黑土区坡耕地土壤水分预测[J].灌溉排水学报, 2014, 33(z1):414-418.
[12] ZHANG Jixian. Multi-source remote sensing data fusion:status and trends[J]. International Journal of Image and Data Fusion, 2010, 1(1):5-24.
[13] ABABAEI B, MIRZAEI F, SOHRABI T, et al. RESERVOIR DAILY INFLOW SIMULATION USING DATA FUSION METHOD[J]. Irrigation&Drainage, 2013, 62(4):468-476.
[14] BAI Y, CHEN Z, XIE J, et al. Daily reservoir inflow forecasting using multiscale deep feature learning with hybrid models[J]. Journal of Hydrology,2016, 532:193-206.
[15] SIVAKUMAR B, BERNDTSSON R. Advances in data-based approaches for hydrologic modeling and forecasting[M]. Singapore:WORLD SCIEN TIFIC, 2010.
[16]任刚红,杜坤,刘年东,等.局域法邻近点选取对供水量预测精度的影响[J].土木建筑与环境工程, 2017, 39(4):102-106.
[17]邓聚龙.灰色系统理论教程[M].武汉:华中理工大学出版社, 1990.
[18] JIN S H, LIN P, HALLETT M. Linear and nonlinear information flow based on time-delayed mutual information method and its application to cortico muscular interaction[J]. Clinical Neurophysiology Official Journal of the International Federation of Clinical Neurophysiology, 2010, 121(3):392-401.
[19] AKRAMI S A, NOURANI V, HAKIM S J S. Development of nonlinear model based on wavelet-ANFIS for rainfall forecasting at Klang Gates Dam[J].Water Resources Management, 2014, 28(10):2 999-3 018.
[20] JANG J S R, SUN C T,MIZUTANI E. Neuro-fuzzy and soft computing-a computational approach to learning and machine intelligence[J]. IEEE Trans.Autom. Control, 1997, 42(10):1 482-1 484.
[21] LIU S, FORREST J, YANG Y. A brief introduction to grey systems theory[C]//IEEE International Conference on Grey Systems and Intelligent Services[J].IEEE, 2011,3:2 403-2 408.
[2] CHAU K W, WU C L. A hybrid model coupled with singular spectrum analysis for daily rainfall prediction[J]. Journal of Hydroinformatics, 2010, 12(4):458-473.
[3] NARAYANAN P, BASISTHAA, SARKAR S, et al. Trend analysis andARIMAmodelling of pre-monsoon rainfall data for western India[J]. Comptesrendus Géoscience, 2013, 345(1):22-27.
[4] KIM B S, HOSSEIN S Z, CHOI G. Evaluation of temporal-spatial precipitation variability and prediction using seasonal ARIMA model in Mongolia[J].Ksce Journal of Civil Engineering, 2011, 15(5):917-925.
[5] FAULINA R. Hybrid ARIMA-ANFIS for Rainfall Prediction in Indonesia[J]. International Journal of Science&Research, 2013.
[6] ABBOT J, MAROHASY J. Input selection and optimisation for monthly rainfall forecasting in Queensland, Australia, using artificial neural networks[J]. Atmospheric ReseIarch, 2014, 138(2):166-178.
[7] DEO R C, SAHN M. Application of the Artificial Neural Network model for prediction of monthly Standardized Precipitation and Evapotranspiration Index using hydrometeorological parameters and climate indices in eastern Australia[J]. Atmospheric Research, 2015, 161/162:65-81.
[8] BAGIROV A M, MAHMOOD A, BARTON A. Prediction of monthly rainfall in Victoria, Australia:Cluster wise linear regression approach[J].Atmospheric Research, 2017, 188:20-29.
[9]郑玉胜,黄介生.基于神经网络的灌溉用水量预测[J].灌溉排水学报, 2004, 23(2):59-61.
[10]廖凯华,徐绍辉,吴吉春,等.基于PCR和ANN构建的土壤转换函数的适用性研究[J].灌溉排水学报, 2014, 33(1):17-21.
[11]王翠翠,魏永霞.基于GM(1,1)-AR模型和AR模型的黑土区坡耕地土壤水分预测[J].灌溉排水学报, 2014, 33(z1):414-418.
[12] ZHANG Jixian. Multi-source remote sensing data fusion:status and trends[J]. International Journal of Image and Data Fusion, 2010, 1(1):5-24.
[13] ABABAEI B, MIRZAEI F, SOHRABI T, et al. RESERVOIR DAILY INFLOW SIMULATION USING DATA FUSION METHOD[J]. Irrigation&Drainage, 2013, 62(4):468-476.
[14] BAI Y, CHEN Z, XIE J, et al. Daily reservoir inflow forecasting using multiscale deep feature learning with hybrid models[J]. Journal of Hydrology,2016, 532:193-206.
[15] SIVAKUMAR B, BERNDTSSON R. Advances in data-based approaches for hydrologic modeling and forecasting[M]. Singapore:WORLD SCIEN TIFIC, 2010.
[16]任刚红,杜坤,刘年东,等.局域法邻近点选取对供水量预测精度的影响[J].土木建筑与环境工程, 2017, 39(4):102-106.
[17]邓聚龙.灰色系统理论教程[M].武汉:华中理工大学出版社, 1990.
[18] JIN S H, LIN P, HALLETT M. Linear and nonlinear information flow based on time-delayed mutual information method and its application to cortico muscular interaction[J]. Clinical Neurophysiology Official Journal of the International Federation of Clinical Neurophysiology, 2010, 121(3):392-401.
[19] AKRAMI S A, NOURANI V, HAKIM S J S. Development of nonlinear model based on wavelet-ANFIS for rainfall forecasting at Klang Gates Dam[J].Water Resources Management, 2014, 28(10):2 999-3 018.
[20] JANG J S R, SUN C T,MIZUTANI E. Neuro-fuzzy and soft computing-a computational approach to learning and machine intelligence[J]. IEEE Trans.Autom. Control, 1997, 42(10):1 482-1 484.
[21] LIU S, FORREST J, YANG Y. A brief introduction to grey systems theory[C]//IEEE International Conference on Grey Systems and Intelligent Services[J].IEEE, 2011,3:2 403-2 408.