Monitor-to-monitor temporal correlation of air pollution and weather variables in the North-Central US

Numerous time series studies have reported associations between daily ambie nt concentrations of air pollution and morbidity or mortality. Recent perso nal exposure studies have also reported relatively high longitudinal correl ation between personal exposures to particulate matter (PM) and home outdoo r PM concentrations, lending support to the health effects reported in time series studies. However, the question remains as to how well the temporal fluctuations in the air pollution levels observed at an outdoor monitor rep resent the temporal fluctuations in the population exposures to pollution o f outdoor origins in a city, and how such representativeness affects the si ze and significance of risk estimates. Also, such spatio-temporal correlati ons would vary from pollutant to pollutant, likely influencing their relati ve significance of statistical associations with health outcomes. In this s tudy, we characterized the extent of monitor-to monitor correlation over ti me among multiple monitoring sites for PM less than 10 mum (PM10), gaseous criteria pollutants, and several weather variables in seven central and eas tern contiguous states (IL, IN, MI, OH, PA, WI, and WV) during the study pe riod of 1988-1990. After removing seasonal trends, the monitor-to-monitor t emporal correlation among the air pollution/weather variables within 100-mi le separation distance in these areas could be generally ranked into three groups: (1) temperature, dew point, relative humidity (r>0.9); (2)O-3, PM10 , NO2 (r: 0.8-0.6); and (3) CO, SO2 (r<0.5). Using the subsets for separati on distance less than 100 miles, regression analyses of these monitor-to-mo nitor correlation coefficients were also conducted with explanatory variabl es including separation distance, qualitative (land use, location setting, and monitoring objectives) and quantitative(large and small variance) site characteristics, and region indicators for Air Quality Control Region (AQCR ). The separation distance was a significant predictor of monitor-to-monito r correlation decline especially for PM10 and NO2 (<similar to>0.2 drop ove r 30 miles). Site characteristic variables were, in some cases, significant predictors of monitor-to-monitor correlation, but the magnitude of their i mpacts was not substantial. Regional differences, as examined by AQCR, were in some cases (e.g., in Metropolitan Philadelphia) substantial. In these a reas, the pollutants that had generally poor monitor-to-monitor correlation in the overall seven states data (i.e., for SO2 and CO) showed higher moni tor-to-monitor correlations, comparable with PM10 and O-3, within the AQCR. These results are useful in interpreting some of the past time series epid emiological results. The differences in monitor-to-monitor correlations fou nd across pollutants in this work ( i.e., r similar to0.8 vs. r similar to0 .4) are sufficiently large that they could be a factor in the different pol lutant significance levels reported in the epidemiologic literature. It is recommended that future epidemiological studies collect and incorporate inf ormation on spatial variability among air pollutants in the analysis and in terpretation of their results.