Poster Paper: Towards an in silico Experimental Platform for Air Quality: Houston, TX As a Case Study

Saturday, November 5, 2016
Columbia Ballroom (Washington Hilton)

*Names in bold indicate Presenter

Bianica Pires1, Gizem Korkmaz1, Katherine Ensor2, David Higdon1, Sallie Ann Keller1, Bryan Lewis1 and Aaron Schroeder1, (1)Virginia Polytechnic Institute and State University, (2)Rice University

It has been demonstrated that localized specific exposures to ozone can dramatically increase health risks for cardiac events and asthma. Many studies, however, use 12- or 24-hour activity summaries. Aggregating time to daily periods misses important details such as variations in ozone levels across physical space and time, which can significantly impact individual and population exposure levels and potentially mask important health effects that could be translated into life saving behavioral and policy changes. For example, it was found that an increase of 20 parts per billion (ppb) in ozone over a period of one to three hours is associated with a 4.4% increased risk of having an out-of-hospital cardiac arrest, for which 90% of cases result in death (Ensor et al., 2013). We seek to estimate exposure levels at a higher granularity by coupling a spatiotemporal air quality model of ozone concentration levels with a synthetic information model of the Houston Metropolitan Area. The synthetic population includes socio-demographically relevant activity sequences and geo-spatially mapped locations for these activities, thus we estimate the movements of each individual in the population and their location second-by-second. This population contains 4.9 million individuals, grouped into 1.8 million households, who perform activities that occur in 1.2 million physical locations. We then match the resolution of time intervals obtained from the 47 monitors that measure ozone across Houston. While traditional approaches often aggregate the population, activities, or concentration levels of the pollutant across space and/or time, this research utilizes high performance computing and statistical learning tools to maintain the granularity of the data, allowing specific exposure levels to be attached to the synthetic individuals. Furthermore, the heterogeneous exposure levels of the population across time are more accurately reflected, allowing for increased sensitivity to detecting the variation of exposure across the population. Several scenarios of the model were run at different levels of resolution, one in which individuals were assumed to stay home all day. While average hourly exposures to ozone across the population were similar across the scenarios, when we maintain the granularity of the data, the variation of exposure could reach an increase of 20 ppb over a short period of time, which could be particularly important if experienced by sensitive populations. This results in varying levels of exposure among individuals in the same zip code, neighborhood, block, and even household depending on their activity patterns throughout the day.

Ensor, K.B., Raun, L.H., and Persse, D. (2013). A case-crossover analysis of out-of-hospital cardiac arrest and air pollution. Circulation, AHA-113.