汉江流域安康站日径流预测的LSTM模型初步研究

doi:10.18306/dlkxjz.2020.04.010

Abstract

Abstract:

Based on the hydrological data from 2003 to 2014, Long-Short Term Memory (LSTM) was used to construct a daily runoff prediction model for the Ankang discharge station in the upper reaches of the Hanjiang River. The accuracy of daily runoff prediction was evaluated under different input conditions. The result shows that when the foreseeing period is one day, the efficiency coefficient of the LSTM in the calibration period and the validation period can reach 0.68 and 0.74 respectively under the condition that only the previous daily runoff of the Ankang Station is used as input. When the previous areal rainfall of the catchment and the previous daily runoff of the upstream Shiquan Station were added to the LSTM model as input variables, the daily runoff prediction precision was improved. The efficiency coefficient of the training period and the validation period could reach 0.83 and 0.84, respectively. The root mean square error was also significantly reduced. The accuracy of the main flood peak flow forecasting also increased. The LSTM can effectively avoid the problem of over-fitting, and has better generalization performance. However, when the foreseeable period is extended from one day to two or three days, the performance of LSTM is significantly reduced.

Key words: long-short term memory, daily runoff forecast, Hanjiang River Basin, Ankang Station

HU Qingfang, CAO Shiyi, YANG Huibin, WANG Yintang, LI Linjie, WANG Lihui. Daily runoff predication using LSTM at the Ankang Station, Hanjing River[J].PROGRESS IN GEOGRAPHY, 2020, 39(4): 636-642.

Figures/Tables 9

Fig.1

Fig.2

Fig.3

Tab.1

Tab.2

Tab.3

Fig.4

Fig.5

Fig.6

References 18

[1]	Liang X, Liang Y . Applications of data mining in hydrology[C]// Proceedings 2001 IEEE International Conference on Data Mining. San Jose, USA, 2001: 617-620.
[2]	Kurtulus B, Razack M . Modeling daily discharge responses of a large karstic aquifer using soft computing methods: Artificial neural network and neuro-fuzzy[J]. Journal of Hydrology, 2010,381(1):101-111. doi: 10.1002/(sici)1096-9861(19970428)381:1&lt;101::aid-cne8&gt;3.0.co;2-5 pmid: 9087422
[3]	Tiwari M K, Chatterjee C . A new wavelet-bootstrap-ANN hybrid model for daily discharge forecasting[J]. Journal of Hydroinformatics, 2011,13(3):500-519. doi: 10.2166/hydro.2010.142
[4]	赵铜铁钢, 杨大文 . 神经网络径流预报模型中基于互信息的预报因子选择方法[J]. 水力发电学报, 2011,30(1):24-30.
	[ Zhao Tongtiegang, Yang Dawen . Mutual information-based input variable selection method for runoff-forecasting neural network model. Journal of Hydroelectric Engineering, 2011,30(1):24-30. ]
[5]	刘柯 . 基于主成分分析的BP神经网络在城市建成区面积预测中的应用: 以北京市为例[J]. 地理科学进展, 2007,26(6):131-139.
	[ Liu Ke . Application of BP Neural Network based on principal component analysis in urban area prediction: A case study of Beijing. Progress in Geography, 2007,26(6):131-139. ]
[6]	李爱云, 张红霞 . 小波人工神经网络在径流预报中的应用[J]. 人民黄河, 2011(10):37-38, 41.
	[ Li Aiyun, Zhang Hongxia . Wavelet neural network model and its application in runoff forecasting. Yellow River, 2011(10):37-38, 41. ]
[7]	郭炅, 张艳军, 王俊勃 , 等. 长短期记忆模型在小流域洪水预报上的应用研究[J]. 水资源研究, 2019(1):24-32.
	[ Guo Jiong, Zhang Yanjun, Wang Junbo , et al. Application of the long short-term memory networks for flood forecast. Journal of Water Resources Research, 2019(1):24-32. ]
[8]	周研来, 郭生练, 张斐章 , 等. 人工智能在水文预报中的应用研究[J]. 水资源研究, 2019(1):1-12.
	[ Zhou Yanlai, Guo Shenglian, Zhang Feizhang , et al. Hydrological forecasting using artificial intelligence techniques. Journal of Water Resources Research, 2019(1):1-12. ]
[9]	Kratzert F, Klotz D, Brenner C , et al. Rainfall-runoff modelling using long short-term memory (LSTM) networks[J]. Hydrology and Earth System Sciences, 2018,22(11):6005-6022. doi: 10.5194/hess-22-6005-2018
[10]	顾逸 . 基于长短期记忆循环神经网络及其结构约减变体的中长期径流预报研究[D]. 武汉: 华中科技大学, 2018.
	[ Gu Yi . Medium and long-term runoff prediction based on long-term and short-term memory-cycle neural networks and their structural reductions. Wuhan, China: Huazhong University of Science and Technology, 2018. ]
[11]	冯钧, 潘飞 . 一种LSTM-BP多模型组合水文预报方法[J]. 计算机与现代化, 2018(7):82-85.
	[ Feng Jun, Pan Fei . A hydrologic forecast method based on LSTM-BP. Computer and Modernization, 2018(7):82-85. ]
[12]	刘桐彤 . 基于长短期记忆神经网络的短期负荷预测方法[J]. 黑龙江科技信息, 2016(31):81.
	[ Liu Tongtong . Short-term load forecasting method based on long-term and short-term memory neural network. Heilongjiang Science and Technology Information, 2016(31):81. ]
[13]	刘炳春, 齐鑫, 王庆山 . 北京城市代谢预测研究: 基于长短期记忆神经网络模型[J]. 地理科学进展, 2019,38(6):851-860. doi: 10.18306/dlkxjz.2019.06.006
	[ Liu Bingchun, Qi Xin, Wang Qingshan . Urban metabolism prediction of Beijing City based on long short-term memory neural network. Progress in Geography, 2019,38(6):851-860. ] doi: 10.18306/dlkxjz.2019.06.006
[14]	晏臻, 于重重, 韩璐 , 等. 基于CNN+LSTM的短时交通流量预测方法[J]. 计算机工程与设计, 2019,40(9):2620-2624, 2659.
	[ Yan Zhen, Yu Chongchong, Han Lu , et al. Short-term traffic flow prediction method based on CNN+LSTM. Computer Engineering and Design, 2019,40(9):2620-2624, 2659. ]
[15]	陶晔, 杜景林 . 基于随机森林的长短期记忆网络气温预测[J]. 计算机工程与设计, 2019,40(3):144-150.
	[ Tao Ye, Du Jinglin . Temperature prediction using long short term memory network based on random. Computer Engineering and Design, 2019,40(3):144-150. ]
[16]	Guo J M, Markoni H . Driver drowsiness detection using hybrid convolutional neural network and long short-term memory[J]. Multimedia Tools and Applications, 2019,78(20):29059-29087. doi: 10.1007/s11042-018-6378-6
[17]	Hochreiter S, Schmidhuber J . Long short-term memory[J]. Neural Computation, 1997,9(8):1735-1780. doi: 10.1162/neco.1997.9.8.1735 pmid: 9377276
[18]	Ritter A, Muñoz-Carpena R . Performance evaluation of hydrological models: Statistical significance for reducing subjectivity in goodness-of-fit assessments[J]. Journal of Hydrology, 2013,480:33-45. doi: 10.1016/j.jhydrol.2012.12.004

站点	最大值 /(m³/s)	最小值 /(m³/s)	平均值 /(m³/s)	标准差 /(m³/s)	变差系数
石泉	9770.0	3.6	315.2	628.9	2.0
安康	16800.0	0	588.7	965.8	1.6

预测方案	模型输入变量			PT/d
预测方案	安康站前期径流序列	安康站集水范围前期面雨量序列	安康站上游石泉站前期径流序列	PT/d
方案1	√			1~3
方案2		√		1~3
方案3			√	1~3
方案4	√	√		1~3
方案5	√		√	1~3
方案6		√	√	1~3
方案7	√	√	√	1~3

预测方案	精度指标	PT=1 d		PT=2 d		PT=3 d
预测方案	精度指标	训练期	检验期	训练期	检验期	训练期	检验期
方案1	NSE	0.68	0.74	0.43	0.58	0.34	0.47
方案1	RMSE/(m³/s)	529	403	711	511	763	573
方案2	NSE	0.65	0.66	0.54	0.54	0.39	0.43
方案2	RMSE/(m³/s)	554	456	638	534	734	594
方案3	NSE	0.64	0.72	0.46	0.60	0.36	0.54
方案3	RMSE/(m³/s)	566	419	694	499	755	532
方案4	NSE	0.77	0.77	0.58	0.62	0.43	0.49
方案4	RMSE/(m³/s)	456	373	610	487	713	559
方案5	NSE	0.72	0.76	0.49	0.62	0.39	0.54
方案5	RMSE/(m³/s)	496	381	669	484	735	532
方案6	NSE	0.79	0.81	0.61	0.65	0.43	0.51
方案6	RMSE/(m³/s)	430	347	590	463	712	548
方案7	NSE	0.83	0.84	0.62	0.67	0.43	0.56
方案7	RMSE/(m³/s)	384	316	581	454	708	525

Daily runoff predication using LSTM at the Ankang Station, Hanjing River

RichHTML

PDF (PC)

Like

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 9

References 18

Related Articles 1

Metrics

Comments

Recommended 0