SÉMINAIRE | Franziska Clerc-Schwarzenbach (université de Zurich, Suisse)

The model performance of bucket-type hydrological models is often improved by including an additional path for water to enter or leave the catchment, aside from precipitation, evaporation, and streamflow. We refer to parameters allowing for such an additional path as ‘sweep parameters’. Far from being an exception, sweep parameters are rather ubiquitous. In this commentary, we discuss the relevance and justifiability of sweep parameters. We argue that although the use of sweep parameters can be considered cheating, as it is a way to fix the water balance, it is a transparent way to do so. Furthermore, in some cases they may represent actual processes. We also present some results on the effects of a sweep parameter on the performance and robustness of two bucket-type hydrological models. We find that sweep parameters do not necessarily reduce model robustness towards variations in the meteorological inputs and often improve streamflow simulations. Therefore, sweep parameters should not per se be frowned upon, as long as they are clearly described and not hidden under the rug.