STATISTICAL APPROACHES IN WIND SECTOR ANALYSES FOR ASSESSING LOCAL SOURCE IMPACTS

Nonparametric statistical methods were used as part of a wind sector a nalysis assessment for the purpose of investigating local source impac ts. Linear-angular rank correlations were employed to test for the pre sence of association between pollutant tracer concentrations and wind direction for a local area dominated by a single emission source (a bi omedical waste combustor). Testing for the presence of these associati ons in ambient data from a local area likely to include multiple pollu tant sources (including a resource recovery facility) was accomplished using the nonparametric Kruskal-Wallis test. Modified pollutant wind rose plots were used to qualitatively investigate the nature of the as sociations detected using these statistical tests, focusing on the sou rces of interest known to emit the pollutant tracers-of-opportunity. T he statistical methods presented provide a quantitative basis for the assessment of source impacts based on observed associations between wi nd direction and pollutant concentrations.