Detecting improvement in ambient air toxics: An application to ambient benzene measurements in Houston, Texas

Traditional regulatory methods for evaluating air toxics have several limitations. Two common methods rely either on self-reported industrial emissions from the Toxics Release Inventory or a single summary statistic such as the average or arithmetic mean. A novel statistical approach for detecting overall long term improvement in ambient air quality is demonstrated using measurements of the air toxic benzene evaluated over five years in Houston, Texas. Through trends of seven key statistical measures, long term improvements were detected at more monitors than would have been found using traditional methods while lack of improvement is highlighted at other monitors. This new approach includes analysis of high and low end concentrations, as well as central tendency, evaluated at specific air toxic human health risk thresholds.