An evaluation of three empirical air-to-leaf models for polychlorinated dibenzo-p-dioxins and dibenzofurans

Three empirical air-to-leaf models for estimating grass concentrations of p olychlorinated dibenzo-p-dioxins and dibenzofurans (abbreviated dioxins and furans) from air concentrations of these compounds are described and teste d against two field data sets. All are empirical in that they are founded o n simplistic bioconcentration and related approaches which rely on field da ta for their parameterization. One of the models, identified as the EPA Mod el, partitions the total air concentration into vapor and particle phases, and separately models the impact of both. A second model addresses only the vapor phase: grass concentrations are modeled as a function of vapor depos ition. For the third model, it is assumed that the grass plants "scavenge" a fixed Volume of air of dioxins, and hence grass concentrations are modele d as a simple product of total air concentration and a constant scavenging coefficient. Field data from two sites, a rural and an industrial site in t he United Kingdom, included concurrent measurements of dioxins in air and f ield grass, and dioxin and furan depositions, for one 6-week sampling perio d. Principal findings include: (1) the EPA Model underpredicted grass conce ntrations at the rural field site by a factor of 2, while the Scavenging Mo del underpredicted grass concentrations by a factor of 3.8, and the Vapor D eposition Model significantly underpredicted grass concentrations (by a fac tor greater than 10), (2) the presence of high soil concentrations for some of the dioxins and furans at the industrial sire appears to have caused hi gher grass concentrations and confounded the air-to-plant modeling exercise , (3) the Scavenging Model could be calibrated to the data set; however, a key premise of this model - that vapor and particle phase dioxins equally i mpact the plants, is nor supported by the field data, (4) measured depositi ons are highly correlated to but systematically lower than modeled depositi ons, which could be due ro modeling assumptions or a systematic measurement bias. (C) 2000 Elsevier Science Ltd. All rights reserved.