Effect of grid size on effective parameters and model performance of the MIKE-SHE code

R.F. Vázquez, L. Feyen, J. Feyen, J.C. Refsgaard

Research output: Contribution to journalArticleResearchpeer-review

130 Citations (Scopus)


The present study discusses the application of the physically based distributed MIKE-SHE code to a medium size catchment using different grid sizes to investigate scale effects on the model results. First a 600 m grid-square model was calibrated. This was then subjected to a multi-resolution (MR) validation test by using the effective parameters of the calibrated model in a 300 m and a 1200 m grid-square model. The MR test indicated that the models for the resolutions analysed only differ marginally. Secondly, the effect of grid size on both the calibrated effective model parameters and the model performance was analysed in the scope of a multi-calibration test in which the calibration and validation processes were kept as unique and standard as possible for every grid size. The model was calibrated and validated for every grid size against both daily catchment discharge measurements and observed water levels using both a split sample procedure and a multi-site validation test. The investigation indicated that the best validation results, in terms of river discharge, were obtained with a 600 m grid-resolution, independent of the stream-flow station. This, together with the exponential increase in computation time when reducing the grid size, indicates that, for the given level of data input and quality, the model type and structure, and the time step, a 600 m grid-resolution is most appropriate for the catchment under study.

Original languageEnglish
Pages (from-to)355-372
Number of pages18
JournalHydrological Processes
Issue number2
Publication statusPublished - 15 Feb 2002


  • Distributed modelling
  • Effective parameters
  • Grid size
  • Performance

Programme Area

  • Programme Area 2: Water Resources


Dive into the research topics of 'Effect of grid size on effective parameters and model performance of the MIKE-SHE code'. Together they form a unique fingerprint.

Cite this