Estimation of spatially variable atmospheric concentrations deduced from regional mass balance models

Multimedia box models of contaminant fate in a regional environment generate estimates of average atmospheric concentrations, ie. the total mass of substance in a defined volume. In this study the issue of estimating the distribution of concentrations, ie. spatial variability, consistent with the estimated average, is addressed. Simulations show that the distribution depends on the number and location of emission points, the local atmospheric dispersion or diffusion characteristics and the rate of chemical removal by reaction and irreversible deposition. The form of a correlating equation is suggested to relate these quantities to the relative standard deviation and the spread factor in the Weibull distribution, the latter being preferred for this analysis. Analysis of monitoring data for volatile organic chemicals (VOCs) in Ontario suggests that the value of the Weibull spread factor is predictable (with error limits), thus enabling average concentrations to be expressed also as approximate spatial distributions. It is also shown that spatial distributions differ depending on whether they are expressed on a sampling site, area, or population basis. Implications for analysis of data generated by multimedia box models to yield distributions and hence estimates of fractions of the environment exceeding specified exposure concentrations are discussed.