The potential health impact of ambient ozone and PM2.5 concentrations modulated by climate change over the United States is investigated using combined atmospheric and health modeling. Regional air quality modeling for 2001 and 2050 was conducted using CMAQ Modeling System with meteorology from the GISS Global Climate Model, downscaled regionally using MM5,keeping boundary conditions of air pollutants, emission sources, population, activity levels, and pollution controls constant. BenMap was employed to estimate the air pollution health outcomes at the county, state, and national level for 2050 caused by the effect of meteorology on future ozone and PM2.5 concentrations. The changes in calculated annual mean PM2.5 concentrations show a relatively modest change with positive and negative responses (increasing PM2.5 levels across the northeastern U.S.) although average ozone levels slightly decrease across the northern sections of the U.S., and increase across the southern tier. Results suggest that climate change driven air quality-related health effects will be adversely affected in more then 2/3 of the continental U.S. Changes in health effects induced by PM2.5 dominate compared to those caused by ozone. PM2.5-induced premature mortality is about 15 times higher then that due to ozone. Nationally the analysis suggests approximately 4000 additional annual premature deaths due to climate change impacts on PM2.5 vs 300 due to climate change-induced ozone changes. However, the impacts vary spatially. Increased premature mortality due to elevated ozone concentrations will be offset by lower mortality from reductions in PM2.5 in 11 states. Uncertainties related to different emissions projections used to simulate future climate, and the uncertainties forecasting the meteorology, are large although there are potentially important unaddressed uncertainties (e.g., downscaling, speciation, interaction, exposure, and concentration-response function of the human health studies).

Heat-related mortality in US cities is expected to more than double by the mid-to-late 21st century. Rising heat exposure in cities is projected to result from: 1) climate forcings from changing global atmospheric composition; and 2) local land surface characteristics responsible for the urban heat island effect. The extent to which heat management strategies designed to lessen the urban heat island effect could offset future heat-related mortality remains unexplored in the literature. Using coupled global and regional climate models with a human health effects model, we estimate changes in the number of heat-related deaths in 2050 resulting from modifications to vegetative cover and surface albedo across three climatically and demographically diverse US metropolitan areas: Atlanta, Georgia, Philadelphia, Pennsylvania, and Phoenix, Arizona. Employing separate health impact functions for average warm season and heat wave conditions in 2050, we find combinations of vegetation and albedo enhancement to offset projected increases in heat-related mortality by 40 to 99% across the three metropolitan regions. These results demonstrate the potential for extensive land surface changes in cities to provide adaptive benefits to urban populations at risk for rising heat exposure with climate change.

The relative contributions of PM2.5 and ozone precursor emissions to air pollution-related premature mortality modulated by climate change are estimated for the U.S. using sensitivities of air pollutants to precursor emissions and health outcomes for 2001 and 2050. Result suggests that states with high emission rates and significant premature mortality increases induced by PM2.5 will substantially benefit in the future from SO2, anthropogenic NOX and NH3 emissions reductions while states with premature mortality increases induced by O3 will benefit mainly from anthropogenic NOX emissions reduction. Much of the increase in premature mortality expected from climate change-induced pollutant increases can be offset by targeting a specific precursor emission in most states based on the modeling approach followed here.

No abstract