Date: May 2003
Creator: Dohanich, Francis Albert
Description: On-road remote sensing has the ability to operate in real-time, and under real world conditions, making it an ideal candidate for detecting gross polluters on major freeways and thoroughfares. In this study, remote sensing was employed to detect carbon monoxide (CO), hydrocarbons (HC), and nitrogen oxide (NO). On-road remote sensing data taken from measurements performed in six states, (Arizona, California, Colorado, Illinois, Texas, and Utah) were cleaned and analyzed. Data mining and exploration were first undertaken in order to search for relationships among variables such as make, year, engine type, vehicle weight, and location. Descriptive statistics were obtained for the three pollutants of interest. The data were found to have non-normal distributions. Applied transformations were ineffective, and nonparametric tests were applied. Due to the extremely large sample size of the dataset (508,617 records), nonparametric tests resulted in "p" values that demonstrated "significance." The general linear model was selected due to its ability to handle data with non-normal distributions. The general linear model was run on each pollutant with output producing descriptive statistics, profile plots, between-subjects effects, and estimated marginal means. Due to insufficient data within certain cells, results were not obtained for gross vehicle weight and engine type. The "year" ...
Contributing Partner: UNT Libraries