The dynamic processes in the tidal reaches of the Yangtze River lead to the complexity of short-term water level forecasting. Historical data of daily water level are obtained for the lower reaches (Anqing–Wuhu–Nanjing) of the Yangtze River. Stationary time series of water level is derived by making the first-order difference with the raw datasets. An artificial neural network–Kalman hybrid model is proposed for water level forecasting, in which the Kalman filtering is introduced for partial data reconstruction. The model is calibrated with the hydrologic daily water level data of years 2014–2016 for MaAnshan station. Comparing with the traditional artificial neural network model, daily water level predictions are improved by the hybrid algorithm. Discrepancies appear under the circumstance of sharp variations of water level observations. Moreover, the implementation strategy of Kalman filtering algorithm is explored, which indicates the superiority of local Kalman filtering.
ZhuFHuangWCaiYet al. Development of a river hydrodynamic model for studying surface-ground water interactions affected by climate change in Heihe river, China. J Coastal Res2014; 68: 129–135.
2.
ValipourMBanihabibMEBehbahaniSMR. Comparison of the ARMA, ARIMA, and the autoregressive artificial neural network models in forecasting the monthly inflow of Dez dam reservoir. J Hydrol2013; 476: 433–441.
3.
ZhangWYuH.Transitive property of correlation coefficient. Coll Math J2013; 29(1): 91–94.
4.
De GooijerJGHyndmanRJ. 25 years of time series forecasting. Int J Forecasting2006; 22(3): 443–473.
ZhangGP.Time series forecasting using a hybrid ARIMA and neural network model. Neurocomputing2003; 50(1): 159–175.
7.
GardnerMWDorlingSR.Artificial neural networks (the multilayer perceptron)—a review of applications in the atmospheric sciences. Atmos Environ1998; 32(14): 2627–2636.
8.
FarukDÖ. A hybrid neural network and ARIMA model for water quality time series prediction. Eng Appl Artif Intel2010; 23(4): 586–594.
9.
ChauKW.A review on integration of artificial intelligence into water quality modelling. Mar Pollut Bull2006; 52(7): 726–733.
10.
CheeneebashJGalanisGGopaulAet al. Improving artificial neural network forecasts with Kalman filtering. Chinese J Pract Hand Surg2007; 13A: 27–33.
11.
EvoraNDCoulibalyP.Recent advances in data-driven modeling of remote sensing applications in hydrology. J Hydroinform2009; 11(3): 194–201.
12.
LiCBaiYZengB.Deep feature learning architectures for daily reservoir inflow forecasting. Water Resour Manag2016; 30: 5145–5161.
13.
BaiYChenZXieJet al. Daily reservoir inflow forecasting using multiscale deep feature learning with hybrid models. J Hydrol2016; 532: 193–206.
14.
ShukurOBLeeMH.Daily wind speed forecasting through hybrid KF-ANN model based on ARIMA. Renew Energ2015; 76: 637–647.
15.
KhasheiMBijariM.An artificial neural network (p, d, q) model for timeseries forecasting. Expert Syst Appl2010; 37(1): 479–489.
16.
BonventureMMWaitituAGWanjoyaAK.Application of Kalman filter to artificial neural networks prediction for foreign exchange rates. Math Theor Model2013; 3(14): 24–33.
17.
YangFTanYJiangTet al. Study on change trend of harmonic constant of Sa tidal constituent at Yangtze River estuary. Yangtze River2014; 45(11): 44–47, 67.
18.
PindyckRobert S.RubinfeldDaniel L. (1998), Econometric Models and Economic Forecasts. Boston: Irwin/McGraw-Hill.
19.
BoxGEPJenkinsGMReinselGC.Time series analysis forecasting and control. J Oper Res Soc1994; 22(2): 199–201.
20.
TaoDLiXHuiYet al. Study on the correlation between water area and multi site water level of Dongting Lake based on MODIS data. Resour Environ Yangtze River Bas2012; 21(6): 756–765.
21.
YangZZhuJWangBet al. Analysis of tidal characteristics of the tide gauges in the Changjiang Estuary. J East China Normal Univ2012; 3: 111–119.
22.
MurataNYoshizawaSAmariS.Network information criterion-determining the number of hidden units for an artificial neural network model. IEEE T Neural Networ1996; 5(6): 865–872.
23.
ZhangGPatuwoBEHuMY.Forecasting with artificial neural networks: the state of the art. Int J Forecasting1998; 14(1): 35–62.
24.
AnsongMOYaoHXHuangJS.Radial and sigmoid basis function neural networks in wireless sensor routing topology control in underground mine rescue operation based on particle swarm optimization. Int J Distrib Sens N2013; 2: 264–273.
25.
OliveiraMAD. An application of neural networks trained with Kalman filter variants (EKF and UKF) to heteroscedastic time series forecasting. Appl Math Sci2012; 6(74): 3675–3686.
26.
DaubechiesI.The wavelet transform, time-frequency localization and signal analysis. IEEE T Inform Theory1990; 36(5): 961–1005.
27.
FildesR.Forecasting, structural time series models and the Kalman filter, by Andrew C. Harvey; Bayesian forecasting and dynamic models, by Mike West; Jeff Harrison. In: HarveyAC (ed.) Forecasting, structural time series models and the Kalman filter. Cambridge: Cambridge University Press, 1990, pp.1031–1033.
28.
AlmeidaLB.The fractional Fourier transform and time-frequency representations. IEEE T Signal Proces1994; 42(11): 3084–3091.
29.
MoradkhaniHGuptaHSorooshianSVet al. Combined parameter and state estimation of hydrological models using ensemble Kalman filter. Adv Water Resour2005; 28(2): 135–147.
