Aerosol characterization in the Southeastern U. S. using satellite data for applications to air quality and climate

The goal of this thesis is to develop and implement methodologies for using satellite remotely sensed data in conjunction with ground based observations and modeling for characterization of regional aerosol variations with applications to air quality and climate studies in the Southeastern U. S. This region is of special interest because of distinct aerosol types, less warming climate trends compared to the rest of U.S., and growing population.;To support this primary goal, a technique is developed that exploits the statistical relationship between PM2.5 (particulate matter that has an aerodynamic radius of 2.5 µm or less) and satellite AOD (Aerosol Optical Depth) from MODIS (Moderate resolution Imaging Spectroradiometer) where a probabilistic approach is used for air quality assessments in the metropolitan Atlanta area.;Examination of the use of multiple satellite sensors to aid in investigating the impacts of biomass burning in the region is performed. The utility of data fusion is evaluated in understanding the effects of the large wildfire that burned in May 2007. This wildfire caused PM2.5 in the metropolitan Atlanta area to exceed healthy levels with some measurements surpassing 150 µg/m3 during the month. OMI (Ozone Monitoring Instrument) AI (Aerosol Index), which qualitatively measures absorbing aerosols, have high values of more than 1.5 during May 26 – 31, 2007. CALIPSO (Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations) a space based lidar was used to determine the vertical structure of the atmosphere across the region during the active fire period. CALIPSO was able to identify wildfire aerosols both within the planetary boundary layer (likely affects local air quality) and aloft where aerosol transport occurs. This has important implications for climatic studies specifically aerosol radiative effects.;In-depth analysis of the satellite and ground based aerosol data records over the past decade (2000 – 2009) are performed from a climatic perspective. The long temporal scale allowed for better characterization of seasonality, interannual variability, and trends. Spatial analysis of ten years of AOD from both MODIS and MISR (Multi-angle Imaging Spectroradiometer) showed little variability of AOD during the winter with mean AOD below 0.1 for the entire region, while the summer had decidedly more variability with mean AOD around 0.33 for MODIS and 0.3 for MISR. Seasonal analysis of the PM2.5 revealed that summer means are twice as high as winter means for PM2.5 . All of the datasets show interannual variability that suggests with time AOD and PM2.5 are decreasing, but seasonal variability obscured the detection of any appreciable trends in AOD; however, once the seasonal influence was removed through the creation of monthly anomalies there were decreasing trends in AOD, but only MODIS had a trend of -0.00434 (per month) that statistically significant at the 95% confidence level.;Satellite and ground-based data are used to assess the radiative impacts of aerosols in the region. The regional TOA (Top Of the Atmosphere) direct radiative forcing is estimated by utilizing satellite AOD from MODIS and MISR both on Terra, along with satellite derived cloud fraction, surface albedo (both from MODIS), and single scattering albedo (SSA) from MISR data from 2000 – 2009. Estimated TOA forcing varied from between -6 to -3 W/m 2 during the winter, and during the warmer months there is more variation with ΔF varying between -28 to -12.6 W/m2 for MODIS and -26 to -11 W/m2 for MISR. The results suggest that when AOD, cloud fraction and surface albedo are all consider they add an additional 6 W/m2 of TOA forcing compared to TOA forcing due to aerosol effects only. Varying SSA can create changes in TOA forcing of about 5 W/m2. With removal of the seasonal variability timeseries anomaly trend analysis revealed that estimated TOA forcing is decreasing (becoming less negative) with MODIS based estimates statistically significant at the 95% confidence level. (Abstract shortened by UMI.).