An Analysis of Possible Climate Change Impacts on Human Mortality in South Central Canada

A synoptic climatological classification approach, using a combination of principal components analysis, an average linkage clustering procedure (hierarchical clustering), and discriminant function analysis (nonhierarchical reclassification), was used to automatically classify distinctive synoptic categories for the period 1953-2001 at each of four selected cities (Montreal, Ottawa, Toronto, and Windsor). Excess mortality within the identified synoptic weather types were shown to be associated with temperature extremes (heat and cold) and air pollution. Using downscaled GCM scenarios from the Canadian GCMs (CGCM1 GHG+A and CGCM2 A2) and the U.S. GCM (GFDL R30 Coupled Climate Model), discriminant function analysis was then used to estimate the projected frequencies of excess mortality-related weather types for future climate scenarios (2040-59, 2070-89). The projected air pollution concentrations were estimated using the historical regression models from weather typing applied to the downscaled climate change scenarios. This model can also incorporate various air pollution emission scenarios. The expected mortality rates due to modeled global warming and pollutant effects for both warm and cold seasons were then estimated. Preliminary results show that under climate change, the future total seasonal elevated mortality rates caused by extreme weather would be increased in summer and decreased in winter due to the expected rise in temperatures.