Regional Assessment of GCM-Simulated Current Climate over Northern Canada

Abstract

Several international Global Climate Models (GCMs) are evaluated on their ability to simulate the mean values and spatial variability of current (1961–90) temperature and precipitation over four regions across Canada’s North. A number of observed climate data sets for Arctic Canada are also assessed. Results reveal a close correspondence, particularly for temperature, among the four observed climate data sets assessed. However, the various GCM simulations of this observed climate show considerable inter-regional and seasonal variability, with temperature more accurately simulated than precipitation. Temperature findings indicate that the British HadCM3, German ECHAM4, and Japanese CCSR-98 models best replicate annual and seasonal values over all sub-regions. The Canadian CGCM2 and U.S. NCAR-PCM models have intermediate accuracy, and the Australian CSIRO-Mk2b and U.S. GFDL-R30 models are least representative. Temperature simulations from the various GCMs collectively display a similar degree of accuracy over all sub-regions, with no clear evidence of superiority in any given area. Precipitation, conversely, is accurately simulated by the majority of models only over northern Quebec/Labrador. All GCMs substantially overestimate annual and seasonal precipitation amounts in the western and central Canadian Arctic.