Abstract
Geographical information systems (GIS) offer potentially powerful tools for exposure assessment in support of both air pollution epidemiology and air quality policy. To date, however, most epidemiological applications have relied on relatively simple techniques, such as buffering and distance functions. In part, this reflects the often limited understanding of the underlying exposure pathways or etiology and the nonspecific nature of the hypotheses being examined. In part, also, it reflects lack of awareness or suspicion of the more advanced capabilities for exposure modeling available in GIS. This article outlines and illustrates some of these techniques. After an initial consideration of “traditional” location-based methods, it examines the use of interpolation methods and dynamic modeling techniques, including modeling of time–activity patterns. It then discusses some of the implications of these different approaches for air pollution epidemiology, and raises the questions of which exposure metrics should be used and what level of spatial and temporal accuracy is required to meet the needs of environmental epidemiology.