Comparative evaluation of methods for estimating potential human exposure to ozone: photochemical modeling and ambient monitoring

Abstract

Photochemical modeling and ambient monitoring of ground-level ozone concentrations provide two alternative/complementary methods for calculating potential population exposure estimates. A comparative evaluation of these methods was undertaken over a study area comprised of the entire state of New Jersey and neighboring parts of Delaware, Maryland, Pennsylvania, and New York. Kriging, a geostatistical interpolation technique, was used for the interpolation of hourly ozone data from 38 air quality monitoring stations operating within the study area, to derive concentration fields for the entire domain. The Urban Airshed Model (UAM-IV), a comprehensive photochemical grid-based model, was then used to calculate the same concentrations from emissions and meteorology inputs. Concentration fields, thus developed, were linked with corresponding population data to calculate potential population exposure estimates to outdoor ozone (Ep.o). The adequacy of kriging as an interpolation technique was evaluated by comparing Ep.o estimates derived via photochemical UAM modeling with those calculated by using concentrations obtained from kriging UAM-calculated values at the locations of the monitoring stations. In general, UAM was found to predict higher Ep.o compared to those derived by kriging observations. In order to test the robustness of the interpolation methodology with respect to assumptions of statistical correlation, two different semivariogram models, spherical and exponential, were used for kriging. Application of the different semivariograms yielded almost identical Ep.o patterns.