Dynamical Downscaling of a Global Chemistry-climate Model to Study the Influence of Climate Change and Variability on Mid-21st Century PM2.5 and Associated Human Mortality in the Continental US

In the coming decades, anthropogenic climate change and associated natural feedback emissions like biogenic VOCs and wildfires can potentially affect air quality, but the climate change signal can be obscured by the noise of climate variability. In this study, we statistically combine probability distributions from multi-year global model ensembles with dynamical downscaling over the continental US, with the goal of quantifying the impacts of climate change and variability on US PM_2.5 levels at fine spatial resolution.

Our approach involves using a 3-member ensemble of GFDL-CM3 global chemistry-climate model simulations. The GFDL-CM3 simulations vary only in initial conditions for the period 2006-2100 under the RCP8.5 scenario, with aerosoland O₃ precursor emissions fixed at 2005 levels to isolate the impact of climate change on air quality. Empirical Orthogonal Function (EOF) analysis of the simulations identifies eastern US regions that vary coherently in PM_2.5, and we select four present (2006-2020) and four future (2050-2060) years that are representative of high and medium annual mean PM_2.5 levels in each region.

We dynamically downscale the GFDL-CM3 meteorology and chemistry of the selected eight years to 12-km with the regional models WRF and CMAQ. The 3-member GFDL-CM3 model simulations (2^o spatial resolution) from 2006 to 2060 provide a broader context for the downscaled CMAQ air quality simulations. Another global model NCAR CESM (12-member ensemble, 1^o spatial resolution) run under the same future scenario as GFDL-CM3 also provides additional context and statistics. Using probability distributions of PM_2.5 from the coarse global models, we construct fine scale mean annual PM_2.5 probability distributions for the present and future in individual 12 km grid cells of the downscaled CMAQ simulations. By examining the differences in fine scale mean annual PM_2.5 distributions between the present and the future, we quantify the effects of climate change and climate variability on PM_2.5 and ultimately, on human mortality associated with PM_2.5.