Air pollution exposure during commuting depends on commuting time, transport mode, ambient air pollutant concentrations, and, for motorized transport, infiltration of ambient pollution into vehicles. Daily round-trip motorized transport commuting times in Hong Kong (HK) range from 35 minutes to 180 minutes. Commuting data for most regions are not directly available. A gravity model is used to estimate commuting flows as a function of working population in a home sector, employment opportunities in work sectors, and a deterrence function that accounts for the influence of travel time. Commuting flows include flow of people from their home sectors to their work sectors. Gravity model methodology was applied to Hong Kong to estimate commuting flows. The objectives of this work are: (1) to identify and evaluate alternative parameterizations of gravity model deterrence functions, select a preferred parameterization, and evaluate the model performance; (2) to assess the sensitivity of the commuting flows to the choice of deterrence function parameterization and (3) to demonstrate the application of gravity model estimated commuting flows to the estimation of commuting exposure. Working population data were obtained from the HK 2016 By-census. Employment opportunities data were derived from data reported by the HK Census and Statistics Department and Planning Department. Travel times were obtained from Google Maps based on routes between sector centroids. The U.S. Environmental Protection Agency’s stochastic, microenvironmental, population-based Air Pollutant EXposure (APEX) model was adapted to HK to quantify commuting exposures. Key commuting inputs to APEX include: (1) commuting flows; and (2) the distribution of commute times to all the work sectors from a given sector. The alternative parameterizations of deterrence functions identified and evaluated were: (i) power; (ii) exponential; (iii) combined power and exponential; (iv) gamma; (v) lognormal; (vi) normal and (vii) Weibull. The estimated commuting flows based on these parameterizations were compared to available commuting flows from the HK Travel Characteristics Survey. The estimated commuting flows were sensitive to the choice of deterrence function parameterization. Variability in PM2.5 commuting exposure was analyzed.
