On the evaluation of sediment transport models in tidal environments

The application of numerical models for the simulation of coastal hydro-and sediment dynamics requires model verification, calibration and validation with field data. Yet, no commonly accepted rules for the evaluation of sediment transport models exist. This paper discusses the significance of statistical parameters and their limitations considering common time lags in tidal environments. It is shown that the occasionally used discrepancy ratio lacks quantitative and qualitative information on model performance, as the time context information on time series characteristics is lost. As an initial measure of association, the simple linear correlation coefficient r2 is proposed. To account for time lag errors in suspended transport models, a separate cross-correlation analysis for the flood and ebb tidal phase is proposed. For a comparison with other model applications, a concluding rating of model performance can be expressed by a dimensionless error definition which takes into account the quality of field data.