Simulation of snow mass and extent in general circulation models

An evaluation of the Biosphere-Atmosphere Transfer Scheme (BATS) snow submo del was conducted, both in a stand-alone mode and within the National Cente r for Atmospheric Research (NCAR) Community Climate Model version 3 (CCM3). We evaluated, in the stand-alone mode, the performance of BATS parameteriz ations at local scales using ground-based observations from the former Sovi et Union and from Mammoth Mountain, California. The BATS snow scheme reprod uces well the seasonal evolution of snow water equivalent in both sites, an d the results for the Mammoth Mountain site compare well with those from a more complex, physically based model (SNTHERM). In the coupled mode, we eva luated the modelled snow cover extent, snow mass, precipitation and tempera ture from BATS as linked to the NCAR CCM3 using available observations. The coupled models capture the broad pattern of seasonal and geographical dist ribution of snow cover, with better overall performance than the passive mi crowave snow data derived from the Nimbus-7 Scanning Multi-channel Microwav e Radiometer (SMMR) which generally underestimates snow depth. In terms of continents, the snow mass is better simulated during the accumulation perio d than during the melt period, which is the case for both North America and Eurasia. The simulation of snow mass, precipitation and air temperature fo r North America is slightly better than that for Eurasia. A rigorous evalua tion of snow simulations in coupled land-atmosphere models requires high qu ality global datasets of snow cover extent, snow depth and snow water equiv alent. The available datasets and model outputs are not yet ready to fulfil this objective. Copyright (C) 1999 John Wiley & Sons, Ltd.