Global water-balance modelling with WASMOD-M: Parameter estimation and regionalisation

SummaryLimitations in water quantity and quality are among the greatest social and economic problems facing mankind. However, difficulties in estimating the global long-term average runoff have led to differences of as much as 30% when integrated to the whole earth. Model estimates of runoff are especially uncertain for the 50% of the global land surface lacking consistent runoff data. In this study, we present the WASMOD-M global water-balance model, constructed to provide robust runoff estimates both for gauged and ungauged basins. WASMOD-M is a conceptual water-budgeting model with two state-variables and five tunable parameters. A simple parameter-value estimation procedure allowed “acceptable” parameter values to be identified both for the majority of gauged basins, and for most ungauged basins. Acceptable global simulations could be accomplished with continentally constant parameter values but at the cost of compensating errors on a basin scale. A “standard”, spatially-distributed parameter-value set was derived for a ”best” global simulation. Of the simulated 59132 0.5° × 0.5° cells, 45% got “good” parameter values as a by-product of regionalisation, 41% from regionalisation, whereas 14% were given a default value set. This global set allowed simulation of the 1915–2000 world water balance. The simulation was in the same range as previously published model results and compilations of runoff measurements. Long-term average within-year runoff variations agreed well with previously published results for most of the studied runoff stations although WASMOD-M was only calibrated against long-term average runoff. Improvement of WASMOD-M and other global water-balance models should be simplified by a common definition of basin boundaries and areas, as well as runoff. Further, modelling progress will depend on improved global datasets of precipitation and runoff regulation.