For many years, the evaluation of numerical model results was largely based on specialist interpretation of graphical output only. Examples are water-level curves or discharge time series, current vector distributions, spreading patterns of heat, spills, etc. The increasing complexity of model functionality, the present integration into model trains, and the application by (non-specialist) end-users located at a great distance from the software developers has led to the demand of user guidance and quantification in model evaluation. (We would like to note that such formal measures for model assessment are not new as such - they are also used in WL-Delft Hydraulics' (semi-) automated model calibration tools such as adjoint modelling, SCEM, GLUE and DUD.) General use of quantitative evaluation methods in engineering modelling results in a more objective, consistent and reproducible validation and assessment, which are therefore more easily verifiable and justifiable to third parties. Focusing on the various user questions behind the application of models for flows, transports and waves, an list was drawn up of the useful approaches for quantitative model evaluation. Well-chosen secondary quantities can also be suitable for model assessment. A well-known example is tidal constants in a tidal model. Attention was given to the selection of relevant evaluation parameters, suitable norms and criteria, and justifiable exceedance values for acceptance or rejection. It is often argued that visual inspection of model results remains necessary during application and evaluation, for easy identification of deviations and the systematic or random nature of such deviations. In that respect, the human eye can be seen as a rapid and highly trained tool for data assimilation. In most coastal ocean applications, a combined visual and quantitative evaluation may be achieved by presenting the spatial distribution of quantitative model quality indicators. This provides useful additional information such as spatial coherence, correlations and consistency. Interpretation of this will often indicate explanations and origins of the possible deviations, see the example below.
Financed by (a.o.) Ministry of: Transport, Public Affairs and Water Management / Education, Culture and Science |
