The need to incorporate water resources into regional and global integ
rated assessment models for evaluating global change impacts has been
identified as being important, but given the spatial and temporal vari
ability of freshwater resources and limited data availability, the inc
lusion of water within these large scale models has proven to be diffi
cult. A compromise between accurate representation of complex processe
s and simplistic parameterization is necessary for capturing regional
and temporal variability and to minimize computational requirements. A
n analysis of Western Europe's and Africa's freshwater runoff, which s
pans a range of climate variability, was performed at varying levels o
f spatial aggregation and at both monthly and annual time steps. Model
results showed that regional runoff characteristics were lost beyond
a data aggregation of 1 degrees x 1 degrees resolution. A monthly soil
moisture model proved to be adequate for assessing annual water avail
ability, which is the scale often used in integrated assessment models
. Simpler methods, such as an empirical annual model and a simple mont
hly model did not consistently replicate annual historic runoff. (C) 1
997 Elsevier Science B.V.