Assessing Dynamic Exposure to Air Pollution - Space-Time Integration in Geography and GIScience - page 291

categories in each subgroup (e.g., age was modeled by three variables: young,
middle, and old age). A regression coefficient of 0.18 and an R 2 value of 0.03 were
obtained for the regression model, suggesting that the independent variables only
accounts for 3 % of the dynamic exposures of the individuals in the subsample.
Among the independent variables, only income has a significant but weak influence
on individual exposure to air pollution (p D 0.03).

16.6 Discussion and Conclusion

This study explored a method for integrating dynamic pollution concentrations and
detailed space-time activity patterns of individuals in air pollution assessment. It
implemented two notions of exposure to air pollution: static exposure evaluated at an
individual's residential location, and dynamic exposure assessed based on various
locations that individual travels to throughout the day. Results from the analysis
do not support the expectation that individual exposures as evaluated by the static
and dynamic measures differ significantly. In addition, gender, education, and age
do not seem to have any influence on people's dynamic exposure to air pollution.
In sum, the analysis did not yield a clear or consistent pattern of statistically
significant differences between static and dynamic exposures, as well as among
various population subgroups using the dynamic measure. In other words, results
from this study do not seem to be supportive of the argument that taking people's
daily movement into account in exposure assessment is crucial for the accuracy of
such assessment.
These results, however, need to be interpreted in the specific context of the study
area and limitations in the data available for modeling the geographic patterns
of air pollution. First, instead of suggesting that people's daily movement does
not affect their exposure to air pollution in significant ways, the results are likely
caused by the specific geography of urban opportunities in the study area and the
activity-travel patterns of the survey participants. As shown in Fig. 16.3 , the highest
levels of pollution concentrations are found mainly in the southeast of downtown
Columbus and in the west at the intersection of two major highways. But the highest
concentrations of urban opportunities (e.g., shopping and recreational facilities) are
located largely in northern suburbs where pollution concentrations were low. Thus
even as survey participants undertook activities and traveled outside of their home,
their activities locations would tend not to coincide geographically with areas of
highest pollution concentrations.
Further, the Gaussian formulation of the dispersion model used in the study is
such that pollution concentrations are subject to a strong distance decay effect,
where the highest levels of particulate matter concentrations are located close to the
sources but quickly decline as distance increases from these pollution sources. Thus,
unless an individual's activities or trips were located close to these pollution sources,
the levels of air pollution they experienced might not be high enough for revealing
significant differences between their static and dynamic exposures. Because of these