Uncertainty in global hydrological modeling related to meteorological forcing and river routing characteristics.

Abstract

Global hydrological models (GHM) can be used in studies on a range of topics related to land surface hydrology. With growing computational power and availability of better input data, efforts have to be made to improve such models in order to use their full potential. Two factors that determine the quality of GHM simulations are meteorological forcing and parameterization of flow characteristics. Question is how model results respond to changes in these features. In the first part of this study two sources of meteorological input are compared, i.e. the reanalysis products ERA-Interim and MERRA, using a reference dataset (CRU TS2.1). Both the actual variable fields, like precipitation, and GHM output based on the products are analyzed. The model that is used is PCR-GLOBWB. It solves the land surface water balance and has a river routing module that simulates discharge and flooding. One of the objectives is to find out which of the reanalysis products is most suitable for further use in this study and in other future research. This turns out to be ERA-Interim, mainly because MERRA suffers from a problem with cloud cover in tropical regions. The second part of the study consists of a sensitivity analysis, in which the influence of changes in channel and floodplain properties on river discharge and flooding is investigated. PCR-GLOBWB is run with 36 different sets of parameters. The resulting discharge and flood extent about the possibilities of improving the model by adjusting them. Comparing the influence of forcing and parameter variations on model performance will tell something about what the focus should be on in future studies on model improvements.