River flow prediction using artificial neural networks: generalisation beyond the calibration range

Imrie, Claire E.; Durucan, Şevket; Korre, Anna

doi:10.1016/s0022-1694(00)00228-6

Cited by 311 publications

(142 citation statements)

References 20 publications

Supporting

Mentioning

136

Contrasting

Unclassified

Order By: Relevance

“…Therefore the data must be scaled so that they always fall within the specified range. The disadvantages of sigmoid activation functions in the output layer, concerning the model's ability to generalise beyond the calibration range, are given by Imrie et al (2000). Solomatine and Dulal (2003) suggest to use an unbounded linear function in the output layer because it is able, to a certain extent, to extrapolate beyond the range of the training data.…”

Section: Neural Network Modelmentioning

confidence: 99%

See 1 more Smart Citation

A neural network model for short term river flow prediction

Teschl

Randeu

2006

Nat. Hazards Earth Syst. Sci.

View full text Add to dashboard Cite

Abstract. This paper presents a model using rain gauge and weather radar data to predict the runoff of a small alpine catchment in Austria. The gapless spatial coverage of the radar is important to detect small convective shower cells, but managing such a huge amount of data is a demanding task for an artificial neural network. The method described here uses statistical analysis to reduce the amount of data and find an appropriate input vector. Based on this analysis, radar measurements (pixels) representing areas requiring approximately the same time to dewater are grouped.

show abstract

Section: Neural Network Modelmentioning

confidence: 99%

“…For identifying the architecture of an ANN associated with determining the number of neurons in each layer, the trial-and-error approach is still the most common (Imrie et at.,2000;Pan and Wang, 2004;Toth et al 2000). Some software packages perform the trial-and-error optimisation automatically.…”

Section: Ann Architecturementioning

confidence: 99%

A neural network model for short term river flow prediction

Teschl

Randeu

2006

Nat. Hazards Earth Syst. Sci.

View full text Add to dashboard Cite

show abstract

“…Imrie et al, 2000;Tingsanchali & Gautam, 2000 1972 -1977 1977 -1983 1983 -1986 Fig. 4 Long-term annual pseudo-precipitation (rain plus melt) of the Mesochora catchment.…”

Section: Runoff Modelling In Climatically Varied Regimesmentioning

confidence: 99%

“…A standard procedure is cross-validation, whereby another set of data is used to monitor the generality of the model during training. The study by Imrie et al (2000), revealed that the generalization beyond the calibration range depends on the use of specific functions with respect to inherent hydrological properties of the catchment and calibration data. However, the aforementioned issues concern the modelling of the absolute extreme flows (high or low), rather than the identification of extreme events in climatically different rainfall-runoff (R-R) processes.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Artificial neural networks and high and low flows in various climate regimes

Panagoulia

2006

Hydrological Sciences Journal

View full text Add to dashboard Cite

An algorithm coupling linear least squares and simplex optimization (LLSSIM) is used to examine the ability of a three-layer feedforward artificial neural network (ANN) to simulate the high and low flows in various climate regimes over a mountainous catchment (the Mesochora catchment in central Greece). The plot of the long-term annual catchment pseudo-precipitation (rain plus snowmelt) simulated by the snow accumulation and ablation model (SAA) of the US National Weather Service (US NWS) showed trends of three climatically distinct periods, described by clearly descending, rising and moderately descending segments in pseudo-precipitation. The ANN model was calibrated for each of the three climate types and each was validated against the others. A set of statistical measures and graphs adapted for high and low flows showed the robustness of the ANN model under various climates and transient conditions. The ANN model proved capable of simulating well the daily high and low flows when it is calibrated for increasing pseudo-precipitation and validated for moderately decreasing pseudo-precipitation. For the entire period, the ANN model provided a better simulation of high and low flows than the conceptual soil moisture accounting (SMA) model of the US NWS, which was also employed in this study. Because the ANN is not a physically-based model, it is by no means a substitute for the SMA model. However, it is concluded that the ANN approach is an effective alternative for daily high-and low-flow simulation and forecasting in climatically varied regimes, particularly in cases where the internal dynamics of the catchment do not require an explicit representation.Key words artificial neural network; conceptual modelling; high flows; linear least squares; low flows; simplex optimization Réseaux de neurones artificiels et crues et étiages en régimes climatiques variésRésumé Un algorithme couplé de moindres carrés linéaires et d'optimisation simplex (LLSSIM) a été utilisé afin d'examiner l'aptitude d'un réseau de neurones artificiel (RNA) sans rétroaction à trois niveaux à simuler les crues et les étiages selon les régimes climatiques variés d'un bassin versant montagneux (le bassin versant de Mesochora en Grèce centrale). Les graphiques de la pseudoprécipitation (pluie plus fonte nivale) à long terme, simulée par le modèle d'accumulation et d'ablation de la neige (SAA) du service national météorologique des Etats Unis (US NWS), ont révélé les tendances de trois périodes climatiques distinctes, correspondant à des segments de décroissance forte, croissance et décroissance modérée. Le modèle RNA a été calé pour chacun des trois types climatiques et validé par rapport aux deux autres. Un ensemble d'indices statistiques et de graphiques adaptés aux crues et aux étiages a montré la robustesse du modèle RNA selon différents climats ainsi qu'en conditions transitoires. Il est apparu que le modèle RNA est apte à bien simuler les crues et les étiages journaliers lorsqu'il est calé en période de pseudo-précipitation croissan...

show abstract

Artificial Neural Network Concepts in Hydrology

Minns¹,

Hall

2005

Encyclopedia of Hydrological Sciences

View full text Add to dashboard Cite

The solution of many applied hydrological problems, such as the forecasting of floods, has for several decades been based upon the concepts of linear systems analysis. However, the introduction of informatics tools, such as Artificial Neural Networks (ANNs), with their origins in cognitive sciences and pattern recognition, has made available new lines of investigation. Nevertheless, despite their apparent structural simplicity, the use of ANNs to encapsulate the transformation of rainfall over a catchment into streamflow at its outlet requires at least as much hydrological insight as a conventional physical/conceptual hydrological model. Of particular importance are: the choice of the input data streams; and the maintenance of the ability of the ANN to generalize and extrapolate beyond its training data set. Attention to these aspects invariably provides a model whose goodness‐of‐fit to an independent testing data set is superior to that of parameter‐based hydrological modeling systems.

show abstract

River flow prediction using artificial neural networks: generalisation beyond the calibration range

Cited by 311 publications

References 20 publications

A neural network model for short term river flow prediction

A neural network model for short term river flow prediction

Artificial neural networks and high and low flows in various climate regimes

Artificial Neural Network Concepts in Hydrology

Contact Info

Product

Resources

About