Abstract. This paper investigates the value of observed river discharge data for global-scale hydrological modeling of a number of flow characteristics that are required for assessing water resources, flood risk and habitat alteration of aqueous ecosystems. An improved version of WGHM (WaterGAP Global Hydrology Model) was tuned in a way that simulated and observed long-term average river discharges at each station become equal, using either the 724-station dataset (V1) against which former model versions were tuned or a new dataset (V2) of 1235 stations and often longer time series. WGHM is tuned by adjusting one model parameter (γ) that affects runoff generation from land areas, and, where necessary, by applying one or two correction factors, which correct the total runoff in a sub-basin (areal correction factor) or the discharge at the station (station correction factor). The study results are as follows. (1) Comparing V2 to V1, the global land area covered by tuning basins increases by 5%, while the area where the model can be tuned by only adjusting γ increases by 8% (546 vs. 384 stations). However, the area where a station correction factor (and not only an areal correction factor) has to be applied more than doubles (389 vs. 93 basins), which is a strong drawback as use of a station correction factor makes discharge discontinuous at the gauge and inconsistent with runoff in the basin. (2) The value of additional discharge information for representing the spatial distribution of long-term average discharge (and thus renewable water resources) with WGHM is high, particularly for river basins outside of the V1 tuning area and for basins where the average sub-basin area has decreased by at least 50% in V2 as compared to V1. For these basins, simulated long-term average discharge would differ from the observed one by a factor of, on average, 1.8 and 1.3, respectively, if the additional discharge information were not used for tuning. The value tends to be higher in semi-arid and snow-dominated regions where hydrological models are less reliable than in humid areas. The deviation of the other simulated flow characteristics (e.g. low flow, inter-annual variability and seasonality) from the observed values also decreases significantly, but this is mainly due to the better representation of average discharge but not of variability. (3) The optimal sub-basin size for tuning depends on the modeling purpose. On the one hand, small basins between 9000 and 20 000 km2 show a much stronger improvement in model performance due to tuning than the larger basins, which is related to the lower model performance (with and without tuning), with basins over 60 000 km2 performing best. On the other hand, tuning of small basins decreases model consistency, as almost half of them require a station correction factor.
Supplementary material to "Global-scale evaluation of 23 precipitation datasets using gauge
observations and hydrological modeling"
