Sensitivity of simulated global-scale freshwater fluxes and storages to input data, hydrological model structure, human water use and calibration

• Main Authors: Müller Schmied, Hannes, Eisner, Stephanie, Franz, Daniela, Wattenbach, Martin, Portmann, Felix Theodor, Flörke, Martina, Döll, Petra
• Format: Article in Journal/Newspaper
• Language: English
• Published: 2014
• Subjects: ddc:550; Greenland; Nash; Sutcliffe; Antarc*; Antarctica
• Online Access: http://publikationen.ub.uni-frankfurt.de/frontdoor/index/index/docId/34404; https://nbn-resolving.org/urn:nbn:de:hebis:30:3-344049; https://doi.org/10.5194/hessd-11-1583-2014; http://publikationen.ub.uni-frankfurt.de/files/34404/hessd-11-1583-2014.pdf

Global-scale assessments of freshwater fluxes and storages by hydrological models under historic climate conditions are subject to a variety of uncertainties. Using the global hydrological model WaterGAP 2.2, we investigated the sensitivity of simulated freshwater fluxes and water storage variations to five major sources of uncertainty: climate forcing, land cover input, model structure, consideration of human water use and calibration (or no calibration). In a modelling experiment, five variants of the standard version of WaterGAP 2.2 were generated that differed from the standard version only regarding the investigated source of uncertainty. Sensitivity was analyzed by comparing water fluxes and water storage variations computed by the variants to those of the standard version, considering both global averages and grid cell values for the time period 1971–2000. The basin-specific calibration approach for WaterGAP, which forces simulated mean annual river discharge to be equal to observed values at 1319 gauging stations (representing 54% of global land area except Antarctica and Greenland), has the highest effect on modelled water fluxes and leads to the best fit of modelled to observed monthly and seasonal river discharge. Alternative state-of-the-art climate forcings rank second regarding the impact on grid cell specific fluxes and water storage variations, and their impact is ubiquitous and stronger than that of alternative land cover inputs. The diverse model refinements during the last decade lead to an improved fit to observed discharge, and affect globally averaged fluxes and storage values (the latter mainly due to modelling of groundwater depletion) but only affect a relatively small number of grid cells. Considering human water use is important for the global water storage trend (in particular in the groundwater compartment) but impacts on water fluxes are rather local and only important where water use is high. The best fit to observed time series of monthly river discharge (Nash–Sutcliffe criterion) ...