Prisons and Pollutant Plumes: A Spatial Analysis of LULU Coexistence

Abstract

Prison facilities and other locally unwanted land uses (LULUs) lead to a number of health, environmental, and socio-economic impacts on the local community. Prison facilities and other LULUs tend to be sited in locations where less wealth and social capital are available to contest their installation. This causes an increased burden on the local population. The purpose of this study is to address the relationship between prisons, other LULUs, and the health impact on the surrounding community using interdisciplinary approaches including regression analysis, plume analysis, and geographic information science. Using a combination of the U.S. Environmental Protection Agency's (EPA) Toxic Release Inventory (TRI), a prison facilities database, and available census data, plume modeling and risk assessment were performed for North Carolina, Pennsylvania, and Texas. The chemicals evaluated are benzene, toluene, ethylbenzene, and xylene (BTEX), known respiratory irritants. The Areal Locations of Hazardous Atmospheres (ALOHA) plume dispersion model was used in this analysis. Risk assessment was performed using an R coded regression analysis evaluating socio-demographic variables, health variables, and prison and TRI facilities to output an air quality value based on the EPA's Air Quality Index (AQI). Using the air quality index that accounts for both physical and socioeconomic characteristics allowed the results to be compared across states and minimized the risk of bias from urban-rural divides. Location analysis was completed using a combination of multinomial regression and probability analysis to assess the relationship between the location of prisons and the location of other LULUs. The results of this analysis were inconclusive however it provided insight into the relationship between income and placement of both prisons and TRI facilities. A Poisson distribution was performed to evaluate the likelihood of TRI facilities being placed in counties with and without prisons as well. This analysis indicated that counties with prisons have a higher probability of receiving TRI facilities. Counties in Texas, North Carolina, and Pennsylvania were selected using a linear regression analysis to assess the relationship between socioeconomic factors and the annual AQI for each county. Income, the percent of housing that is renter occupied, the percent of the population employed, and the percent of asthma related Medicare expenses are the socioeconomic factors most related to air quality. In Texas, the highest modeled AQI was present in the county with a prison facility while in Pennsylvania and North Carolina, the non-prison county yielded the highest modeled AQI. The difference between these modeled AQIs was small for both Pennsylvania and North Carolina; however, in Texas the difference between values was approximately 50 AQI points. Plume analysis was performed using the combined stack and fugitive air emissions for TRI facilities as the emissions source. The plume dispersion models indicated that the BTEX facilities considered are unlikely to pose a serious health risk to the populations in the surrounding area with few exceptions. Plumes were only able to be generated for facilities with the greatest emissions due to limitations in the model's short-range accuracy. This indicates that while the toxins being considered are respiratory irritants, they are unlikely to strongly influence the health of the local populations or prison inmates through direct emissions. The largest dispersion radius calculated was approximately 61 yards indicating that the probability of adverse exposure to toxins from target facilities is minimal.