| The spatial variability of airborne contaminants in an international airshed was investigated using geostatistics and air dispersion modeling. Analyses were conducted on contaminant species measured in Detroit and Windsor over a two-week period during September 2008. Ordinary kriging with anisotropy and cokriging of measurements with different scales of support were expected to improve estimates generated with autocorrelated ordinary kriging models. Available emissions and meteorological data were also integrated into an air dispersion model to test the feasibility of predicting spatial variability using emissions inventories and traffic estimates for two of the sampled air pollutants, benzene and nitrogen dioxide (NO2).;Results showed no improvement for models incorporating anisotropy. Cokriging improved correlation coefficients and reduced residual statistics for some pollutant concentration estimates compared to ordinary kriging. This suggests that, under favorable conditions, secondary data can be used to increase the spatial resolution of concentration estimates while improving model agreement with field measurements. |
