PM_2.5 Population Exposure in New Delhi Using a Probabilistic Simulation Framework

Abstract

This paper presents a Geographical Information System (GIS) based probabilistic simulation framework to estimate PM2.5 population exposure in New Delhi, India. The framework integrates PM2.5 output from spatiotemporal LUR models and trip distribution data using a Gravity model based on zonal data for population, employment and enrollment in educational institutions. Time-activity patterns were derived from a survey of randomly sampled individuals (n = 1012) and in-vehicle exposure was estimated using microenvironmental monitoring data based on field measurements. We simulated population exposure for three different scenarios to capture stay-at-home populations (Scenario 1), working population exposed to near-road concentrations during commutes (Scenario 2), and the working population exposed to on-road concentrations during commutes (Scenario 3). Simulated annual average levels of PM2.5 exposure across the entire city were very high, and particularly severe in the winter months: ∼200 μg m(-3) in November, roughly four times higher compared to the lower levels in the monsoon season. Mean annual exposures ranged from 109 μg m(-3) (IQR: 97-120 μg m(-3)) for Scenario 1, to 121 μg m(-3) (IQR: 110-131 μg m(-3)), and 125 μg m(-3) (IQR: 114-136 μ gm(-3)) for Scenarios 2 and 3 respectively. Ignoring the effects of mobility causes the average annual PM2.5 population exposure to be underestimated by only 11%.

Integrating travel behavior with land use regression to estimate dynamic air pollution exposure in Hong Kong

“…higher compared to static estimates in Dhondt et al (2012) and de Nazelle et al (2013),respectively. In another Asian city setting, Saraswat et al (2016) found ignoring effects of mobility led to an underestimation of annual PM2.5 population exposure by about 11% in New Delhi. When work locations were considered in addition to residential, exposures to NOX and NO2 were found to increase by 5 -10 ppb (Shafran-Nathan et al, 2017).…”

“…In other studies assessing dynamic air pollution exposure, mobility or time-activity data were derived from transport and activity-based simulation models (Setton et al, 2011;Dhondt et al, 2012), GPS (Dons et al, 2011), mobile-based tracking (de Nazelle et al, 2013), travel smartcard (Smith et al, 2016), travel surveys (Saraswat et al, 2016) or cellular network information (Dewulf et al, 2016;Nyhan et al, 2016). These were combined with air pollution modeling to assess personal and population exposure to pollutants.…”

“…Recent studies have used travel surveys (Saraswat et al, 2016), activity-based simulations (Setton et al, 2008;Dhondt et al, 2012), GPS or mobile-based tracking (Dons et al, 2011;de Nazelle et al, 2013), travel smartcard (Smith et al, 2016) or cellular network data (Dewulf et al, 2016) to derive dynamic exposure estimates. These approaches have facilitated detailed spatio-temporal analysis of individual travel behaviors.…”

“…These approaches have facilitated detailed spatio-temporal analysis of individual travel behaviors. A number of studies found static estimates underestimated exposure levels (Dhondt et al, 2012;de Nazelle et al, 2013;Dewulf et al, 2016;Nyhan et al, 2016;Saraswat et al, 2016). Simulations based on travel survey and air pollution modeling data found integrating mobility can affect exposure estimates by as much as 30% (Marshall et al, 2006).…”

A review of factors impacting exposure to PM2.5, ultrafine particles and black carbon in Asian transport microenvironments

“…Differences in population demographics may also be important. A recent exposure modeling study for PM 2.5 in New Delhi focused on population mobility and transport-related exposures using a probabilistic framework that included a land use regression component and a zonal model to simulate home and work locations and commuting between them (Saraswat et al, 2016).…”