Atmospheric model data for macroscale hydrology

Conventional climate observation networks are often deficient in both quant ity and quality of data for macroscale hydrological modelling. In this sens e we regard a basin as macroscale if it exceeds 10 000 km(2). Atmospheric m odels such as general circulation models (GCM) and numerical weather predic tion models (NWP) provide alternative sources of data in such cases. While there are many questions concerning the accuracy of the hydrologically-rele vant output from such models, experience has shown good results in hydrolog ical modelling of macroscale northern basins. The spatial scale of climate data used in the hydrological modelling has varied from 3.75 degrees (simil ar to 400 km at 60 degrees) for GCM data to 4 km for data from a high resol ution boundary layer model. The time periods used have varied from a few da ys for NWP data to almost a century for data from the high resolution bound ary layer model. Comparisons of hydrological model outputs using data from various atmospheric models for the 37 000 km(2) upper Columbia Basin and th e 1.8 million km(2) Mackenzie Basin in north-western Canada are given. The experience gained in such one-way off-line coupling of atmospheric and hydr ological models will be useful in developing a global model of the full hyd rological cycle with two-way linking of atmospheric, oceanic and hydrologic al models. (C) 1999 Elsevier Science B.V. All rights reserved.