Spatial and temporal variation in pollutant concentrations: Implications for exposure and risk assessment of airborne pollutants.

摘要

Environmental exposure and risk assessment rely on estimates of pollutant concentrations in air, water, soil, food and other exposure pathways. These concentrations result from complex patterns of source emissions, transport and fate processes, all of which are temporally and spatially variable. The accuracy and representativeness of concentrations used in typical exposure assessments, which are often point estimates, are uncertain. This dissertation examines the spatial and temporal variation of pollutant concentrations in air, surface runoff and soil, and the effects of this variation on exposure estimates. Transport and fate models are used to evaluate the spatial variation air pollutant concentrations and the temporal variation of pollutant concentrations in soil and runoff. Exposure estimates derived from models that assume constant and variable pollutant transport processes are compared, and extreme value theory is used to examine the probability and magnitude of short-term, high soil pollutant concentrations.; Predicted pollutant levels showed significant spatial and temporal variation, e.g., different types of emission sources produced vastly different spatial patterns and pollutant concentrations. Exposure indicators based on either proximity to a pollution source or large geographic areas poorly represented spatial patterns and pollutant levels. Surface runoff loads resulting from atmospheric deposition of pollutants showed significant temporal variation that depended on site-specific precipitation patterns. Concentrations of soluble pollutants in surface soils also showed significant temporal variation, especially at locations with little precipitation. Models that assumed constant transport processes produced estimates that were neither accurate nor representative, and simulation models using relatively short time steps are recommended. The use of a stochastic model to produce synthetic precipitation data in conjunction with the extreme value analysis allowed the prediction of maximum pollutant concentrations with return periods of up to several hundred years. These methodologies appear broadly applicable to infrequent events that may cause acute exposures and adverse impacts.; This dissertation suggests that the spatial and temporal variation in environmental pollutant levels is significant and not fully recognized in the current exposure assessment practices. The techniques presented in this dissertation can help to evaluate this variation and improve the exposure estimates used in epidemiology and risk assessment.