Assessment of Personal Exposure to Air Pollution Based on Trajectory Data
MetadataShow full metadata
Air pollution has been among the biggest environmental risks to human health. Exposure assessment to air pollution is essentially a procedure to quantify the degree to which people get exposed to hazardous air pollution. Exposure assessment is also a critical step in health-related studies exploring the relationship between personal exposure to environmental stressors and adverse health outcomes. Given the critical role of exposure assessment, it is important to accurately quantify and characterize personal exposure in geographic space and time.
For years numerous exposure assessment methods have been developed with respect to a wide spectrum of air pollutants. Of all the methods, the most commonly used one is to use a representative geographic unit as the surrogate location to estimate the potential impact from hazardous air pollution from differing sources on that location. The representative unit is one person’s home location in most cases. Such studies, however, have failed to recognize the significance of both the dynamics of human activities and the variation of air pollution in geographic space and time.
It is believed that personal exposure is essentially a function of space and time as an individual’s time-activity patterns and intensities of air pollutant in question vary over space and time. It is therefore imperative to account for the spatiotemporal dynamics of both in exposure assessment. To this end, the goal of this study is to account for the spatiotemporal dynamics of both human time-activity patterns and air pollution for assessing personal exposure. More specifically this dissertation aims to achieve three objectives as summarized below.
First, in light of the deficiency of existing home-based exposure assessment methods, this study proposes an innovative trajectory-based model for assessing personal exposure to ambient air pollution. This model provides a computational framework for assessing personal exposure when trajectories, documenting human spatiotemporal activities, are modeled into a series of tours, microenvironments (MEs), and visits. A set of individual-level trajectories was simulated to test the performance of the proposed model, in conjunction with one-day air pollution (PM2.5) data in Beijing, China. The results from the test demonstrated that the trajectory-based model is capable of capturing the spatiotemporal variation of personal exposure, thus providing more accurate, detailed and enriched information to better understand personal exposure. The findings indicate that there is considerable variation in intra-microenvironment and inter-microenvironment exposure, which identified the importance of distinguishing between different MEs. Moreover, this study tested the proposed model using an empirical dataset.
Second, little is known about the difference between the estimated exposure based on home locations only and that considering the locations of all human activities. To fill this gap, this study aims to test whether the exposure calculated from the home-based method is statistically significantly different from the exposure estimated by the newly developed trajectory-based model. A Dataset containing 4,000 individual-level one-day trajectories (Dataset 1) was simulated to test the aforementioned hypothesis. The exposure estimates in comparison are the average hourly exposure over a 24-hour period from two exposure assessment methods. The 4,000 trajectories were split into another two subsets (Datasets 2, 3) according to the difference between home-based exposure estimates and trajectory-based exposure estimates. The Wilcoxon Signed-rank test was used to evaluate whether the difference between the two models is significant. The results show that the statistically significant difference was found only in Dataset 3. The same test was also applied to a set of empirical trajectories. The significant difference exists in the results from the empirical data. The mixed results suggest that additional research is needed to verify the difference between the two exposure assessment methods.
Third, little research has taken into consideration of hourly traffic variation and human activities simultaneously in a model for assessing personal exposure to traffic emissions. To fill this gap, this study develops a new trajectory-based model to quantify personal exposure to traffic emissions. The hourly share of daily traffic volume of each roadway in the study area was estimated by calculating the traffic allocation factors (TAFs) of each roadway. Next, the hourly traffic emission surfaces were built using the hourly shares and a kernel density algorithm. A 3-D cube representing the spatiotemporal distribution of traffic emission was constructed, which overlaid the simulated individual-level trajectory data for assessing personal exposure to traffic emissions. The results showed that people’s time-activity patterns (e.g., where an individual lives/works, where an individual travels) were significant factors in exposure assessment. This study suggests that people’s time activities and hourly variation of traffic emission should be simultaneously addressed when assessing personal exposure to traffic emissions.
To sum up, this study has devoted a large effort in quantifying and characterizing personal exposure in geographic space and time. A few of contributions to the knowledge of exposure science are listed as follows. First, this study contributes two exposure assessment models in characterizing personal spatiotemporal exposure using trajectory data. One is developed for assessing personal exposure to ambient air pollution, and the other one is for assessing personal exposure to traffic emissions. Second, this study demonstrates the intra- and inter-microenvironment variation of personal exposure and reveals the significance of people’s time-activity patterns in exposure assessment. Third, this study investigates the difference in exposure estimates between conventional home-based methods considering home locations only and trajectory-based methods accounting for the locations of all activities. The mixed findings from Wilcoxon Signed-rank tests suggest more research is needed to explore how personal exposure varies with time-activity patterns. All these contributions will have important implications in exposure science, environment science, and epidemiology.