Geographical Information System Time-Activity and Population Exposure Modeling

Abstract

PP-30-191 Background/Aims: Although it is increasingly recognized that environmental concentrations do not necessarily equate to individual exposure, concentrations at place of residence are often used to approximate exposure. However, individual exposure can vary, depending on circumstances such as the amount of time spent indoors and outdoors (Jarup 2004). Furthermore, environmental pollutants and exposures vary spatially and temporally with, for example, different levels during day and night (Briggs 2005). A better approximation of exposure can be obtained through time-activity modeling, using a mixture of individual level data and statistical (aggregated) data from time-activity surveys. Methods: A model has been programmed in a geographical information system (ESRI ArcGIS) using network analysis tools. The model can use geocoded data for start and end locations of each trip (X/Y), start and end times of each trip, trip mode, travel speed, route, and destination type. Using these data, the model is able to reproduce the time-activity sequence. However, in many cases much, if not all, of these data will not be known and so the model can probabilistically impute, on the basis of available statistical information, any of these data. These statistical data comprise generalized distributions for the study population (eg, national surveys). Results: Using data collected as part of a study to analyze the effect of traffic-related air pollution on the journey to school (Walker et al, 2009), model results have been validated. Results using 31 time-activity diaries, including actual routes taken by school children (captured using global positioning system technology), showed the model to be a reasonable predictor of routes. Conclusion: This research demonstrates the difference obtained in exposure values from fixed home location in comparison with modeled spatio-temporal data. These results allow some quantification of exposure misclassification that may be introduced into small-scale spatial epidemiological analysis.