| Popis: |
Many studies have reported associations between respiratory symptoms and resident proximity to traffic. However, only a few have documented information about the relationship between traffic volume and air quality in local areas. This study investigates the impact of traffic volume on air quality at different geographical locations in the state of South Carolina using multilevel linear mixed models and Grey Systems. Historical traffic volume and air quality data between 2006 and 2016 are obtained from the South Carolina Department of Transportation (SCDOT) and the United States Environmental Protection Agency (EPA) monitoring stations. The data are used to develop prediction models that relate Air Quality Index (AQI) to traffic volume for selected counties and schools. For the counties, two models are developed, one with Ozone (O3) and one with PM2.5 as the dependent variable. For the schools, only one model is developed, with O3 as the dependent variable. The number of counties and schools studied are limited by the availability of air monitoring stations dedicated to measuring O3 and PM2.5. Several types of models were investigated. They include linear regression model (LM), linear mixed-effect regression model (LMER), Grey Systems (GM), error corrected GM (EGM), Grey Verhulst (GV), error corrected GV (EGV), and LMER + EGM. The LM model produced the least accurate estimate while the LMER + EGM model produced the most accurate estimate (average RMSE is less than 5%). The models’ estimates suggest that air quality in South Carolina will continue to get worse in the coming years due to increasing AADT. An interesting finding of this study is that some counties and schools will have higher levels of O3 or PM2.5 when AADT decreases. This finding suggests that there are other factors, other than AADT, that influence the air quality in these counties and schools. Keywords: Air quality index, Annual average daily traffic, Emissions, Grey systems, Multilevel linear models |
