Medium to long term runoff forecast for the Huai River Basin based on machine learning algorithm

doi:10.13745/j.esf.sf.2021.10.2

Abstract

Abstract:

Accurate and reliable medium to long term runoff prediction is key to supporting scientific allocation of water resources and improving water resources utilization efficiency. In this study, the AdaBoost (AdB), Random forest (RF) and Support vector machine (SVM) algorithms were used to make medium to long term runoff predictions at the Wangjiaba and Bengbu stations, Huai River Basin, from November to October of the following year. The permutation accuracy importance measure was used to select the key factors affecting monthly runoff from 1562 factor variables constructed from 130 meteorological-climate factors and previous precipitation/runoff fluxes, and the monthly runoff forecasts based on the aforementioned three machine learning algorithms were made for each month, using model parameters determined by way of random search combined with cross-validation. The performances of the three algorithms were evaluated using the variable amplitude error qualification rate indicator and the grade (five level) forecast qualification rate indicator. The former indicator shows that the 12-month average qualification rates of prediction for the three algothrims at the Wangjiaba station were, respectively, 99.8% (AdB), 96.6% (RF) and 95.5% (SVM) and at the Bengbu station 100% (AdB), 94.8% (RF) and 93.8% (SVM); the results using the latter indicator at the Wangjiaba station were, respectively, 79.0% (AdB), 76.4% (RF) and 79.9% (SVM), and at the Bengbu Station 81.0% (AdB), 75.6% (RF) and 76.6% (SVM). The three machine learning algorithms all performed well, but RF and SVM had low prediction rates on high discharge values, and AdB performed better than RF and SVM overall.

Key words: machine learning, AdaBoost algorithm, random forest algorithm, support vector machine algorithm, runoff prediction, Huai River Basin

CLC Number:

P338.2

HU Yiming, CHEN Teng, LUO Xuyi, TANG Chao, LIANG Zhongmin. Medium to long term runoff forecast for the Huai River Basin based on machine learning algorithm[J]. Earth Science Frontiers, 2022, 29(3): 284-291.

Figures/Tables 10

References 23

[1]	郦于杰, 梁忠民, 唐甜甜. 基于支持向量回归机的长期径流预报及不确定性分析[J]. 南水北调与水利科技, 2018, 16(3): 45-50.
[2]	BENNETT J C, WANG Q J, LI M, et al. Reliable long-range ensemble streamflow forecasts: combining calibrated climate forecasts with a conceptual runoff model and a staged error model[J]. Water Resources Research, 2016, 52(10): 8238-8259. DOI URL
[3]	LIANG Z M, TANG T T, LI B Q, et al. Long-term streamflow forecasting using SWAT through the integration of the random forests precipitation generator: case study of Danjiangkou Reservoir[J]. Hydrology Research, 2018, 49(5): 1513-1527. DOI URL
[4]	刘甜, 梁忠民, 邱辉, 等. 基于CFS的汉江上游梯级水库系统月入库径流预测[J]. 水电能源科学, 2019, 37(8): 14-17, 10.
[5]	HU Y M, SCHMEITS M J, VAN ANDEL S J, et al. A stratified sampling approach for improved sampling from a calibrated ensemble forecast distribution[J]. Journal of Hydrometeorology, 2016, 17(9): 2405-2417. DOI URL
[6]	赵铜铁钢, 杨大文, 蔡喜明, 等. 基于随机森林模型的长江上游枯水期径流预报研究[J]. 水力发电学报, 2012, 31(3): 18-24, 38.
[7]	ADAMOWSKI J, FUNG CHAN H, PRASHER S O, et al. Comparison of multiple linear and nonlinear regression, autoregressive integrated moving average, artificial neural network, and wavelet artificial neural network methods for urban water demand forecasting in Montreal, Canada[J]. Water Resources Research, 2012, 48(1): W01528.
[8]	农振学, 王超, 雷晓辉. 基于主成分分析和BP神经网络的赣江流域中长期径流预报[J]. 水电能源科学, 2018, 36(1): 16-19.
[9]	范钟秀. 中长期水文预报[M]. 南京: 河海大学出版社, 1999.
[10]	谢帅, 黄跃飞, 李铁键, 等. 不同流域的自回归径流预报效果对比[J]. 应用基础与工程科学学报, 2018, 26(4): 723-736.
[11]	李福兴, 陈伏龙, 蔡文静, 等. 基于EMD组合模型的径流多尺度预测[J]. 地学前缘, 2021, 28(1): 428-437.
[12]	HUANG S Z, CHANG J X, HUANG Q, et al. Monthly streamflow prediction using modified EMD-based support vector machine[J]. Journal of Hydrology, 2014, 511: 764-775. DOI URL
[13]	YASEEN Z M, JAAFAR O, DEO R C, et al. Stream-flow forecasting using extreme learning machines: a case study in a semi-arid region in Iraq[J]. Journal of Hydrology, 2016, 542: 603-614. DOI URL
[14]	李伶杰, 王银堂, 胡庆芳, 等. 基于随机森林与支持向量机的水库长期径流预报[J]. 水利水运工程学报, 2020(4): 33-40.
[15]	陶思铭, 梁忠民, 陈在妮, 等. 长短期记忆网络在中长期径流预报中的应用[J]. 武汉大学学报(工学版), 2021, 54(1): 21-27.
[16]	NI L L, WANG D, SINGH V P, et al. Streamflow and rainfall forecasting by two long short-term memory-based models[J]. Journal of Hydrology, 2020, 583: 124296.
[17]	SHEN C P. A transdisciplinary review of deep learning research and its relevance for water resources scientists[J]. Water Resources Research, 2018, 54(11): 8558-8593. DOI URL
[18]	王健宗, 瞿晓阳. 深入理解AutoML和AutoDL: 构建自动化机器学习与深度学习平台[M]. 北京: 机械工业出版社, 2019.
[19]	HAM Y G, KIM J H, LUO J J. Deep learning for multi-year ENSO forecasts[J]. Nature, 2019, 573(7775): 568-572. DOI URL
[20]	洪瑾, 甘成势, 刘洁. 基于机器学习的洋岛玄武岩主量元素预测稀土元素[J]. 地学前缘, 2019, 26(4): 45-54.
[21]	游文霞, 申坤, 杨楠, 等. 基于AdaBoost集成学习的窃电检测研究[J]. 电力系统保护与控制, 2020, 48(19): 151-159.
[22]	中华人民共和国水利部.SD 250-2000水文情报预报规范[S]. 北京: 中国水利水电出版社, 2000.
[23]	BERGSTRA J, BENGIO Y. Random search for hyper-parameter optimization[J]. Journal of Machine Learning Research, 2012, 13: 281-305.

分级	要素距平值/%	分级	要素距平值/%
枯水	距平<-20	偏丰	10<距平≤20
偏枯	-20≤距平<-10	丰	距平>20
正常	-10≤距平≤10

分级	要素距平值/%	分级	要素距平值/%
枯水	距平<-20	偏丰	10<距平≤20
偏枯	-20≤距平<-10	丰	距平>20
正常	-10≤距平≤10

时间	王家坝指标1/%
时间	Ada	RF	SVM		Ada	RF	SVM	Ada		RF		SVM		Ada	RF	SVM
当年11月	100.0	93.8	100.0	70.8		68.8	75.3		100.0		94.0		94.0	82.0	78.0	86.0
当年12月	100.0	97.9	93.8	77.1		70.8	76.4		100.0		100.0		98.0	82.0	74.0	82.0
次年1月	100.0	98.0	100.0	69.4		63.3	70.6		100.0		98.0		94.1	76.5	78.4	62.7
次年2月	100.0	98.0	100.0	69.4		83.7	86.8		100.0		94.1		88.2	76.5	66.7	58.8
次年3月	100.0	89.8	93.9	83.7		83.7	82.8		100.0		98.0		86.3	92.2	86.3	72.5
次年4月	100.0	100.0	100.0	83.7		87.8	81.6		100.0		90.2		88.2	72.5	66.7	72.5
次年5月	100.0	98.0	87.8	81.6		75.5	73.5		100.0		90.2		100.0	72.5	68.6	80.4
次年6月	100.0	98.0	100.0	87.8		75.5	89.8		100.0		98.0		98.0	82.4	78.4	86.3
次年7月	100.0	98.0	100.0	69.4		65.3	81.6		100.0		98.0		94.1	74.5	76.5	82.4
次年8月	98.0	95.9	91.8	83.7		81.6	85.8		100.0		82.4		100.0	92.2	62.7	90.2
次年9月	100.0	93.9	83.7	81.6		71.4	67.3		100.0		96.1		100.0	80.4	82.4	80.4
次年10月	100.0	98.0	100.0	89.8		89.8	87.4		100.0		98.0		84.3	88.2	88.2	64.7
平均	99.8	96.6	95.9	79.0		76.4	79.9		100.0		94.8		93.8	81.0	75.6	76.6

时间	王家坝指标1/%
时间	Ada	RF	SVM		Ada	RF	SVM	Ada		RF		SVM		Ada	RF	SVM
当年11月	100.0	93.8	100.0	70.8		68.8	75.3		100.0		94.0		94.0	82.0	78.0	86.0
当年12月	100.0	97.9	93.8	77.1		70.8	76.4		100.0		100.0		98.0	82.0	74.0	82.0
次年1月	100.0	98.0	100.0	69.4		63.3	70.6		100.0		98.0		94.1	76.5	78.4	62.7
次年2月	100.0	98.0	100.0	69.4		83.7	86.8		100.0		94.1		88.2	76.5	66.7	58.8
次年3月	100.0	89.8	93.9	83.7		83.7	82.8		100.0		98.0		86.3	92.2	86.3	72.5
次年4月	100.0	100.0	100.0	83.7		87.8	81.6		100.0		90.2		88.2	72.5	66.7	72.5
次年5月	100.0	98.0	87.8	81.6		75.5	73.5		100.0		90.2		100.0	72.5	68.6	80.4
次年6月	100.0	98.0	100.0	87.8		75.5	89.8		100.0		98.0		98.0	82.4	78.4	86.3
次年7月	100.0	98.0	100.0	69.4		65.3	81.6		100.0		98.0		94.1	74.5	76.5	82.4
次年8月	98.0	95.9	91.8	83.7		81.6	85.8		100.0		82.4		100.0	92.2	62.7	90.2
次年9月	100.0	93.9	83.7	81.6		71.4	67.3		100.0		96.1		100.0	80.4	82.4	80.4
次年10月	100.0	98.0	100.0	89.8		89.8	87.4		100.0		98.0		84.3	88.2	88.2	64.7
平均	99.8	96.6	95.9	79.0		76.4	79.9		100.0		94.8		93.8	81.0	75.6	76.6