A two‐step‐ahead recurrent neural network for stream‐flow forecasting

Chang, Li-Chiu; Chang, Fi‐John; Chiang, Yi‐Ming

doi:10.1002/hyp.1313

Cited by 84 publications

(34 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…RMSE can provide a good measure of model performance for high flows (Karunanithi et al, 1994), but significant variations in the assessment of different catchments will occur, since the evaluation metric is dependent on the scale of the dataset that is being analysed. It is perhaps better to report RMSE, rather than Mean Squared Error (MSE; Chang et al, 2004;Chen et al, 2006;Furundzic, 1998;Riad et al, 2004;), because RMSE is measured in the same units as the original data, rather than in squared units, and is thus more representative of the size of a "typical" error. MSE was at one point the most widely used measure of overall accuracy for a forecasting method but it is also the method that has incurred the most criticism (e.g.…”

Section: Statistical Parameters Of Observed and Modelled Time Series mentioning

confidence: 99%

“…Metric Example Studies 1* AME None found 2* PDIFF None found 3 MAE Chang et al (2004); Chen et al (2006); Dawson et al (2006a); Karunasinghe and Liong (2006); Liong et al (2000); Supharatid (2003) (Gupta et al 1998). ME and RMSE are expressed in terms of daily measures by the National Weather Service but the reported equations can be applied to different temporal periods.…”

mentioning

confidence: 99%

See 1 more Smart Citation

HydroTest: A web-based toolbox of evaluation metrics for the standardised assessment of hydrological forecasts

Dawson

Abrahart

See

2007

Environmental Modelling & Software

454

244

View full text Add to dashboard Cite

Section: Statistical Parameters Of Observed and Modelled Time Series mentioning

confidence: 99%

mentioning

confidence: 99%

HydroTest: A web-based toolbox of evaluation metrics for the standardised assessment of hydrological forecasts

Dawson

Abrahart

See

2007

Environmental Modelling & Software

454

244

View full text Add to dashboard Cite

“…Chang et al 2004, Pan and Wang 2005, Muluye and Coulibaly 2007, Carcano et al 2008, Coulibaly 2010. Furthermore, none of these studies has investigated the EKF algorithm to perform training of recurrent neural networks for use in hydrology-related problems.…”

Section: Introductionmentioning

confidence: 99%

Improving long-range hydrological forecasts with extended Kalman filters

Muluye

2011

Hydrological Sciences Journal

View full text Add to dashboard Cite

There is a continuing effort to advance the skill of long-range hydrological forecasts to support water resources decision making. The present study investigates the potential of an extended Kalman filter approach to perform supervised training of a recurrent multilayer perceptron (RMLP) to forecast up to 12-month-ahead lake water levels and streamflows in Canada. The performance of the RMLP was compared with the conventional multilayer perceptron (MLP) using suites of diagnostic measures. The results of the forecasting experiment showed that the RMLP model was able to provide a robust modelling framework capable of describing complex dynamics of the hydrological processes, thereby yielding more accurate and realistic forecasts than the MLP model. The performance of the method in the present study is very promising; however, further investigation is required to ascertain the versatility of the approach in characterizing different water resources and environmental problems.Key words extended Kalman filter; hydrological forecast; recurrent multilayer perceptron; streamflow Amélioration des prévisions hydrologiques à longue échéance par filtres de Kalman étendusRésumé Il y a un effort constant pour faire progresser la qualité des prévisions hydrologiques à longue échéance en vue d'aider la prise de décision en matière de ressources en eau. La présente étude examine le potentiel d'une approche par filtre de Kalman étendu pour effectuer un apprentissage supervisé d'un perceptron multicouches récurrent (RMLP) pour prévoir jusqu'à 12 mois à l'avance des niveaux de lacs et des débits au Canada. La performance du RMLP a été comparée à celle du perceptron multicouches conventionnel (MLP) en utilisant un ensemble de mesures de diagnostic. Les résultats de l'expérience ont montré que le modèle RMLP a pu fournir un cadre de modélisation robuste, capable de décrire la dynamique complexe des processus hydrologiques, donnant ainsi des prévisions plus précises et réalistes que le modèle MLP. Les performances de la méthode dans la présente étude sont très prometteuses, mais des investigations complémentaires sont nécessaires pour s'assurer de la versatilité de l'approche dans la caractérisation de divers problèmes environnementaux et de ressources en eau.

show abstract

“…Multi-stepahead prediction is a challenging task which attempts to make predictions several time steps into the future. Chang et al (2004) developed a two-step-ahead recurrent neural network for streamflow forecasting. Later they explored three types of multi-step-ahead (MSA) neural networks, viz.…”

Section: Introductionmentioning

confidence: 99%

Effect of data time interval on real-time flood forecasting

Remesan

Ahmadi

Shamim

et al. 2010

Journal of Hydroinformatics

View full text Add to dashboard Cite

Rainfall-runoff is a complicated nonlinear process and many data mining tools have demonstrated their powerful potential in its modelling but still there are many unsolved problems. This paper addresses a mostly ignored area in hydrological modelling: data time interval for models. Modern data collection and telecommunication technologies can provide us with very high resolution data with extremely fine sampling intervals. We hypothesise that both too large and too small time intervals would be detrimental to a model's performance, which has been illustrated in the case study. It has been found that there is an optimal time interval which is different from the original data time interval (i.e. the measurement time interval). It has been found that the data time interval does have a major impact on the model's performance, which is more prominent for longer lead times than for shorter ones. This is highly relevant to flood forecasting since a flood modeller usually tries to stretch his/her model's lead time as far as possible. If the selection of data time interval is not considered, the model developed will not be performing at its full potential. The application of the Gamma Test and Information Entropy introduced in this paper may help the readers to speed up their data input selection process.

show abstract

A two‐step‐ahead recurrent neural network for stream‐flow forecasting

Cited by 84 publications

References 30 publications

HydroTest: A web-based toolbox of evaluation metrics for the standardised assessment of hydrological forecasts

HydroTest: A web-based toolbox of evaluation metrics for the standardised assessment of hydrological forecasts

Improving long-range hydrological forecasts with extended Kalman filters

Effect of data time interval on real-time flood forecasting

Contact Info

Product

Resources

About