Establishing a Dynamic Exposure Assessment with an Activity-based Modeling Approach: Methodology and Results for the Dutch Case Study

Abstract

In large-scale population exposure assessments, exposure monitoring studies are not applicable and air quality monitoring does not cover the entire study area. In these cases, air quality modeling approaches are often combined with residential information to perform indirect population exposure assessments. However, this static approach is often the subject of controversy since the exposure is likely to be underestimated in (urban) areas with a large concentration of work activities. Unfortunately, attempts at more advanced dynamic exposure assessments, certainly for large study areas or large time scales, are -to the best of our knowledge - very scarce. The main objective of this research project is to develop a population exposure model for the Dutch population taking into account population and concentration dynamics in the entire study area, and using, for the first time, an activity-based modelling approach for this purpose. More specifically, this project also focuses on the evaluation of spatio-temporal differences between this new dynamic approach and the conventional static exposure assesment. Methodology & results: The first step in the exposure modeling procedure concerned the simulation of people's travel patterns in the Dutch study area. In this first step a comprehensive population dynamics model, the activity-based model ALBATROSS, was used to simulate activity-travel patterns for individuals within the population (approximately 10 million agents). The activity-based model relies on a set of decision rules to simulate activity-travel schedules and takes into account spatial and temporal interrelationships between simulated trips. This activity-based modeling step resulted in hourly population distributions over 4000 population zones. A validation of the predicted travel behavior displayed only slight differences with reported travel values (less than 10%). Next, hourly exposure concentrations were modeled with the AURORA Eulerian grid model, using meteorological, geographical and emission input data. By using a nested procedure the concentrations were simulated for the entire study area with grid cell sizes up to one square kilometer. Considering the size of the study area (± 40,000 km2) and the intended time scale (six months of modeling on 1 h-time resolution), this modeling step was very time-consuming. In a last step, the exposure to atmospheric pollutants was estimated using GIS overlay techniques in ArcView. Due to the importance to health and/or traffic purposes, only the emissions from PM10, PM2.5 and NOX were considered for exposure analysis. Both hourly exposure estimates per population zone as integrated exposure indices were calculated for the Dutch population for the assigned time period. By comparing these results in a GIS environment to the exposure results from the static exposure method the differences between both approaches were evaluated. Conclusion: The results of this study indicate the importance of using a dynamic exposure approach instead of a conventional static exposure method. Further, this research also demonstrates the possibility of using an activity-based model for exposure analyses. Considering the characteristics of this population dynamics model, further research will allow for impact analyses of several traffic control and emission reduction measures on the exposure of people.