Assessing annual and seasonal spatial variability of ambient PM10 using linear regression analysis in a United States-Mexico urban sprawl
As air quality issues grow progressively in the consciousness of the global community, stakeholders are pressing for epidemiologic studies of air pollution. The effects of chronic exposure to ambient air pollution remain a difficult challenge due to its substantial small-scale spatial variation. Recent approaches to assess intra-urban exposure have employed the use of proximity-based assessment, interpolation methods, emission-meteorological models, dispersions models, and land-use regression (LUR) models. This thesis assesses the spatial variability of ambient particulate matter (PM10) obtained by five self-governing models for the area of El Paso. Using multiple linear regression analysis five regression-based equations were developed to predict annual and seasonal intra-urban gradient of PM 10. The first four models were limited to seasonal analysis in order to evaluate the sensitivity of PM10 mass concentration to conditions attributable specifically to each seasonal period such as: temperature, human activities, diurnal patterns and atmospheric stability; as well as to explore short-term spatial variability for each season. The fifth and final model was developed to estimate annual individual levels of long-term exposure of ambient PM10. Integrated measurements of 7-day periods were collected as part of a three-year monitoring campaign. The samples were measured employing Dichotomous samplers located at 13 sites spread throughout the study region. Three of these sites were co-located with the Texas Commission on Environmental Quality (TCEQ) monitoring stations and thus correlated with the measurements of TCEQ's Tapered Element Oscillating Microbalance (TEOM) samplers for quality-assurance and quality-control (QA/AC). These concentrations were used to fit a multiple linear regression model to assess the association between logarithmic concentrations of PM10 and surrogate parameters derived using ESRI ArcGIS 9.2 geographic information system. Integrated local spatial covariates regarding land use, traffic-related, population, property value and physical geography were derived from circular buffers with radii from 200 m to 5000 m around the monitors. The validated models explained between 74% to 90% of the variation in PM10 concentrations associated with four variables: zoning, traffic-based, nearest distance to the border area, and population. The PM10 surface area was generated with a 3528 grid points distributed orthogonally at a 610 m isotropic resolution capable of characterizing small-scale variations and capturing the relative impact of the bordering urban areas. Overall, high concentrations were perceived in the fall model which continue during the winter season and lessened during the spring where finally moderates abruptly during the summer season. The five models, including the annual model, concede with a southward pattern where all of the cases identified the downtown core area as a hot spot where the air quality is often under relative stress when compared to other areas. In the case of exposure and habitats, the higher burdens seen southward is partly due to the higher pollutant concentrations found in these areas. PM 10 concentrations, for example, are greatly increased in the downtown core largely because of the human activity in the area; particulate concentrations are raised not only by local emissions but also by emissions from Ciudad Juarez. In fact, 2.3, 84, and 4% of the population of El Paso were found to be exposed to an increase of 10 ìg/m3 of PM10 from spring to summer, summer to fall, and fall to winter season respectively. Developing a health risk of 1 to 10% increase in symptoms related to cough, decline lung function, and asthma attacks. The predictive maps from LUR analysis appear to capture small-area variations in PM10 concentrations assessing short- and long-term exposure. The application of such prediction models showed that a substantial fraction of the variability in PM10 concentrations was explained by land use-zoning variables. This approach results in optimal epidemiologic studies and air-quality management as it reduces the potential for exposure misclassification posed by other modeling methods.
Environmental Health|Environmental science|Environmental engineering
Garcia, Mario Ivan Garcia, "Assessing annual and seasonal spatial variability of ambient PM10 using linear regression analysis in a United States-Mexico urban sprawl" (2010). ETD Collection for University of Texas, El Paso. AAI1477783.