On constraining TOPMODEL hydrograph simulations using partial saturated area information

This study uses field observations of the extent of saturated area over limited areas of the small Uhlirska catchment (1.87 km(2)) in the Czech Republic in calibrating the parameters of a version of TOPMODEL. The field information is used within the GLUE methodology, which involves evaluating many different randomly chosen parameter sets within the chosen model structure. The different parameter sets are evaluated on performance in both discharge prediction and prediction of the observed saturated areas using appropriate likelihood measures. The results show that the saturated area information results in a strong constraint of the transmissivity parameter of the model, but that the other parameters show good fits across most of the range over which they are sampled. Quite different posterior distributions for the transmissivity parameter are found for the two different years of data used in conditioning the model. The effect on the prediction bounds for stream discharges is much less, perhaps because the transmissivity parameter combines with different values of the other parameters in the two years to capture the dominant modes of discharge response of the catchment. Copyright (C) 2002 John Wiley Sons, Ltd.