Journal cover Journal topic
Hydrology and Earth System Sciences An interactive open-access journal of the European Geosciences Union
Journal topic

Journal metrics

Journal metrics

  • IF value: 4.936 IF 4.936
  • IF 5-year value: 5.615 IF 5-year
  • CiteScore value: 4.94 CiteScore
  • SNIP value: 1.612 SNIP 1.612
  • IPP value: 4.70 IPP 4.70
  • SJR value: 2.134 SJR 2.134
  • Scimago H <br class='hide-on-tablet hide-on-mobile'>index value: 107 Scimago H
    index 107
  • h5-index value: 63 h5-index 63
Volume 11, issue 6 | Copyright

Special issue: Data-driven approaches, optimization and model integration:...

Hydrol. Earth Syst. Sci., 11, 1869-1881, 2007
© Author(s) 2007. This work is licensed under
the Creative Commons Attribution-NonCommercial-ShareAlike 2.5 License.

  04 Dec 2007

04 Dec 2007

Hydrological model coupling with ANNs

R. G. Kamp1,2 and H. H. G. Savenije1 R. G. Kamp and H. H. G. Savenije
  • 1Section of Water Resources, Delft University of Technology, Delft, The Netherlands
  • 2MX. Systems B.V., Rijswijk, The Netherlands

Abstract. There is an increasing need for model coupling. However, model coupling is complicated. Scientists develop and improve models to represent physical processes occurring in nature. These models are built in different software programs required to run the model. A software program or application represents part of the system knowledge. This knowledge is however encapsulated in the program and often difficult to access.

In integrated water resources management it is often necessary to connect hydrological, hydraulic or ecological models. Model coupling can in practice be difficult for many reasons related to data formats, compatibility of scales, ability to modify source codes, etc. Hence, there is a need for an efficient and cost effective approach to model-coupling. Artificial neural networks (ANNs) can be used as an alternative to replace a model and simulate the model's output and connect it to other models.

In this paper, we investigate an alternative to traditional model coupling techniques. ANNs are four different models: a rainfall runoff model, a river channel routing model, an estuarine salt intrusion model, and an ecological model. The output results of each model is simulated by a neural network that is trained on corresponding input and output data sets. The models are connected in cascade and their input and output variables are connected.

To test the results of the coupled neural network also a coupled system of four sub-system models has been set-up. These results have been compared to the results of the coupled neural networks. The results show that it is possible to train neural networks and connect these models. The results of the salt intrusion model was however not very accurate. It was difficult for the neural network to represent both short term (tidal) and long term (hydrological) processes.

Publications Copernicus
Special issue