This dissertation tests whether or not mercury emissions from electric power plants are not a significant contributor to mercury measurements in rainfall and argues that the current United States (U.S.) Environmental Protection Agency (EPA) proposed regulatory scheme for controlling mercury from electric power plants, the Clean Air Mercury Rule (CAMR), is an effective regulatory mechanism by using a number of ordinary least square (OLS) and spatial regression models. Two dependent variables are tested, mercury concentration (the average mercury concentration measured in rainfall in nanograms per liter, ng/L) and mercury deposition (the total annual mercury falling at each measurement site in nanograms per square meter, ng/m2), with mercury concentration determined to be the more valid dependent variable. The source for the mercury concentration and deposition data is the Mercury Deposition Network (MDN), part of the National Atmospheric Deposition Program (NADP), with the data obtained for between 46 and 75 sites operating from 2001 through 2005.;Independent variables include: (1) emissions to the air from power plants, (2) emissions to the air from other industrial sites, (3) emissions to the land from the mining industry, (4) population as a proxy variable for vehicle emissions, (5) burned area from wildfires, (6) precipitation and (7) dummy variables for year and EPA region. Data for independent variables 1, 2, and 3 were obtained from the EPA's Toxic Release Inventory (TRI) program. Population for each county in the U.S. was obtained from the Census Bureau, and wildfire data was obtained from the U.S. Department of Agriculture satellite based fire mapping system, Moderate Resolution Imaging Spectroradiometer (MODIS). Microsoft Access was utilized to summarize and total the independent variables within a variable radius of the MDN measurement sites, ranging from 25 to 500 miles. The software tool GeoDa 0.95i, made available by the University of Illinois, was used to perform the OLS, spatial lag, and spatial error regressions.;After changing the functional form of the equation to a log-linear model (using the natural log form of the dependent variable and the linear forms of the independent variables) to deal with heteroskedasticity, the results indicate a strong spatial component to the model. Other than precipitation, the most significant predictor of mercury concentration is fire area burned between 50 and 75 miles of the MDN measurement site (z = 3.08, p0.01). Other positive and significant predictors in this model include all other industry emissions between 25 and 50 miles (z = 2.71, p0.01), fire area burned between 75 and 100 miles (z = 2.64, p0.01), population within 25 miles (z = 1.91, p0.10), utility emissions between 25 and 50 miles (z = 1.88, p0.10), and population between 50 and 75 miles (z = 1.71, p0.10). Two of the independent variables are significant and have negative coefficients. These are utility emissions between 50 and 75 miles (z = -2.49, p0.05), and fire area burned between 25 and 50 miles (z = -2.12, p0.05).;Several conclusions are drawn from this research, including: (1) that utility mercury emissions are marginally significant as a predictor of mercury concentration in rainfall, but only at distances under 50 miles from the measurement point, (2) that there is no known best method for controlling mercury emissions from all utility plants at high levels of collection efficiency (90 percent) although research is ongoing, and (3) that the cap-and-trade provisions of CAMR would be unlikely to result in the creation of new or the exacerbation of existing mercury hotspots. Given that the U.S. District of Columbia Circuit Court of Appeals set aside the CAMR rule in early 2008, two policy prescriptions are provided. One approach makes an economic argument for revising the cap-and-trade provisions of CAMR to include transfer coefficients. The second suggestion involves a less complicated and more politically acceptable change to the trading rules for mercury.
展开▼
