We focus on the Great Lakes Region of North America, analyzing both surface measurement data and numerical model simulations. Ten years worth of PM10 and PM2.5 data at a selection of sites in the Upper Midwest as reported by the U.S. Environmental Protection Agency (EPA) are correlated with large-scale meteorology over the region from the National Centers for Environmental Prediction (NCEP) reanalysis data. We use two statistical downscaling methods (multiple linear regression and analog) to identify which processes have the greatest impact on aerosol concentration variability. Based on these correlations, we can estimate PM over periods when no data are available, and over periods where data were withheld for model verification.
We find that sites in urban or industrial areas tend to exhibit correlations between PM concentrations and surface winds and atmospheric stability. Sites in suburban and rural areas have PM correlated with sea level pressure, geopotential height, and mid-level winds, indicative of synoptic-scale processes. Preliminary results suggest that regional meteorological processes account for 54 – 71% of aerosol concentration variability at study sites.
Using the Weather Research and Forecasting (WRF) model, an advanced mesoscale numerical weather prediction model, we simulate regional conditions conducive to high PM episodes to further examine how meteorology across scales contributes to elevated aerosol levels.