Model trees as an alternative to neural networks in rainfall—runoff modelling
- 1 June 2003
- journal article
- Published by Taylor & Francis in Hydrological Sciences Journal
- Vol. 48 (3) , 399-411
- https://doi.org/10.1623/hysj.48.3.399.45291
Abstract
This paper investigates the comparative performance of two data-driven modelling techniques, namely, artificial neural networks (ANNs) and model trees (MTs), in rainfall—runoff transformation. The applicability of these techniques is studied by predicting runoff one, three and six hours ahead for a European catchment. The result shows that both ANNs and MTs produce excellent results for 1-h ahead prediction, acceptable results for 3-h ahead prediction and conditionally acceptable result for 6-h ahead prediction. Both techniques have almost similar performance for 1-h ahead prediction of runoff, but the result of the ANN is slightly better than the MT for higher lead times. However, the advantage of the MT is that the result is more understandable and allows one to build a family of models of varying complexity and accuracy. Résumé Cet article étudie de manière comparative les performances de deux techniques de modélisation pluie débit contraintes par les données, en l'occurrence des réseaux de neurones artificiels (RNA) et des arbres de modèles (AM). L'applicabilité de ces techniques est étudiée pour la prévision des débits d'un bassin versant européen pour des anticipations de une, trois et six heures. Les résultats montrent que les RNA et les AM produisent des résultats excellents pour la prévision à une heure, acceptables pour la prévision à trois heures et acceptables sous condition pour la prévision à six heures. Les deux techniques ont des performances presque similaires pour la prévision des débits à une heure, mais les résultats du RNA sont légèrement meilleurs que ceux de l'AM pour les délais plus longs. Néanmoins l'AM présente les avantages de fournir des résultats plus compréhensibles et de permettre la construction d'une famille de modèles de complexité et de précision variables.Keywords
This publication has 18 references indexed in Scilit:
- Improved non-linear transfer function and neural network methods of flow routing for real-time forecastingJournal of Hydroinformatics, 2001
- A recurrent neural networks approach using indices of low-frequency climatic variability to forecast regional annual runoffHydrological Processes, 2000
- Comparing neural network and autoregressive moving average techniques for the provision of continuous river flow forecasts in two contrasting catchmentsHydrological Processes, 2000
- River flow prediction using artificial neural networks: generalisation beyond the calibration rangeJournal of Hydrology, 2000
- Artificial Neural Networks in HydrologyPublished by Springer Nature ,2000
- An artificial neural network approach to rainfall-runoff modellingHydrological Sciences Journal, 1998
- Artificial neural networks as rainfall-runoff modelsHydrological Sciences Journal, 1996
- Multivariate Adaptive Regression SplinesThe Annals of Statistics, 1991
- Comparative analysis of several conceptual rainfall-runoff modelsJournal of Hydrology, 1991
- Nonlinear flood routing with multilinear modelsWater Resources Research, 1987