A TRAJECTORY-CLUSTERING-CORRELATION METHODOLOGY FOR EXAMINING THE LONG-RANGE TRANSPORT OF AIR-POLLUTANTS

We present a robust methodology for examining the relationship between synoptic-scale atmospheric transport patterns and pollutant concentra tion levels observed at a site. Our approach entails calculating a lar ge number of back-trajectories from the observational site over a long period of time and subjecting them to cluster analysis. The short-ter m component (weather-related variations) of the pollutant concentratio n time-series data is segregated according to the back-trajectory clus ters. Non-parametric statistics are then used to test for ''significan t'' differences in the chemical composition of pollutant data associat ed with each cluster. Additional information about the spatial and tem poral scales of pollutant transport is obtained from the time-lagged i nter-site correlation analysis of ozone for a specific cluster. To ill ustrate the application of this methodology, we examined 5 yr long tim e-series data of ozone concentrations measured at Whiteface Mountain, NY, Cliffside Park, NJ, and Quabbin Summit, MA. The results provide ev idence of ozone transport to these sites, revealing the spatial and te mporal scales involved in the transport. (C) 1998 Elsevier Science Ltd . All rights reserved.