Snow Mass over North America: Observations and Results from the Second Phase of the Atmospheric Model Intercomparison Project

Abstract

Abstract Eighteen global atmospheric general circulation models (AGCMs) participating in the second phase of the Atmospheric Model Intercomparison Project (AMIP-2) are evaluated for their ability to simulate the observed spatial and temporal variability in snow mass, or water equivalent (SWE), over North America during the AMIP-2 period (1979–95). The evaluation is based on a new gridded SWE dataset developed from objective analysis of daily snow depth observations from Canada and the United States with snow density estimated from a simple snowpack model. Most AMIP-2 models simulate the seasonal timing and the relative spatial patterns of continental-scale SWE fairly well. However, there is a tendency to overestimate the rate of ablation during spring, and significant between-model variability is found in every aspect of the simulations, and at every spatial scale analyzed. For example, on the continental scale, the peak monthly SWE integrated over the North American continent in AMIP-2 models varies between ±50% of the observed value of ∼1500 km3. The volume of water in the snowpack, and the magnitudes of model errors, are significant in comparison to major fluxes in the continental water balance. It also appears that the median result from the suite of models tends to do a better job of estimating climatological mean features than any individual model. Year-to-year variations in large-scale SWE are only weakly correlated to observed variations, indicating that sea surface temperatures (specified from observations as boundary conditions) do not drive interannual variations of SWE in these models. These results have implications for simulations of the large-scale hydrologic cycle and for climate change impact assessments.