Daimler Trucks NA presents new Class 8 Freightliner Cascadia with fuel efficiency improved up to 8%
ORNL licenses rare earth magnet recycling process to Momentum Technologies

MIT-led study suggests mobile-phone data provide a deeper picture of pollution exposure in urban settings

A study led by MIT researchers, focused on New York City, suggests that using mobile-phone data to track people’s movement provides an even deeper picture of exposure to pollution in urban settings than by studying air-quality levels in fixed places. Their open-access paper is published in the ACS journal Environmental Science & Technology.

Previous environmental epidemiological studies quantifying the health impacts of population exposure to have not considered spatially- and temporally-varying populations. The new study—the first of its kind—measured population activity patterns representing several million people to evaluate population-weighted exposure to air pollution on a city-wide scale. Mobile and wireless devices yield information about where and when people are present; the researchers were able to determine collective activity patterns using counts of connections to the cellular network.

Spatial and temporal human mobility patterns can be characterized based on mobile phone trace data and techniques to extract mobility information from mobile phone traces have progressed in recent times. Data sets that can be used for human mobility analyses include geographically and time-referenced Call Detail Records (CDRs) or counts of connections to the cellular network. These types of data sets may include millions of anonymized records of mobile phone and wireless device usage and can yield detailed information about the activity patterns of large populations, especially where mobile phone and wireless device penetration rates are high relative to the population. Studies have shown that mobile phone trace data can also represent individual mobility patterns and demonstrate advantages over traditional travel surveys used in human mobility studies, which are limited in terms of low response numbers, spatiotemporal scales, and limited update frequencies.

[The] aim of this study was to quantify population-weighted exposure to air pollution by combining extensive population activity patterns and air pollution measurements.

—Nyhan et al.

The study, based on 121 days of data from 2013, broke New York City into 71 districts. Population-weighted exposure was calculated as a function of air pollution concentration in a district and the proportion of the total population of NYC exposed in that district.

The team assumed that the hourly air quality level and the percentage of the total population present within each district were uniformly distributed. Each district had a geospatial centroid coordinate and hourly PM2.5 parameter levels to be inferred or already associated if having an air quality monitoring station located in it. (New York has an extensive monitoring network that measures air quality in 155 locations.)

The researchers compared two scenarios of population-weighted exposure:

  1. Air pollution exposure weighted by population activity counts deciphered using extensive mobile device usage records—referred to as “Active Population Exposure”.

  2. air pollution exposure weighted by assuming people were always located at their home location, using a Census-defined spatial population distribution—referred to as “Home Population Exposure”.

PM2.5 concentration data was obtained from the New York City Community Air Survey (NYCCAS). Spatial distribution maps of air quality parameter levels were developed using the spatial interpolation technique of inverse-distance-weighting, as this had been used effectively in studies quantifying pollution exposures.

Geographically and time-referenced mobile traffic data were used to quantify hourly percentages of the total population of NYC present in each of the 71 districts of the city throughout the study period.

Population-weighted PM2.5 exposures were computed for each district for both population scenarios of Active Population Exposure and Home Population Exposure.

Map of mean population-weighted exposure (PWE) to PM2.5 per district for the Home Population Exposure scenario (top-left), the Active Population Exposure scenario (top-right), and the relative difference between these two scenarios (lower). Units are μg/m3 × percent of population present/district. Credit: ACS, Nyhan et al. Click to enlarge.

The team found that in 68 of the NYC districts, exposure levels to particulate matter were significantly different when the daily movement of 8.5 million people was accounted for.

Specifically, the flow of people into parts of midtown Manhattan, and some parts of Brooklyn and Queens close to Manhattan, appeared to increase aggregate exposure to PM in those areas. Meanwhile, the daytime movement of people away from Staten Island actually lowered overall exposure levels in that borough.

The traditional way to look at pollution is to have a few measurement stations and use those to look at pollution levels. But that’s sensitive to where the [measuring] stations are. If you want to quantify exposure, you also need to know where people are.

—Carlo Ratti, director of MIT’s Senseable City Lab

The researchers believe the method in the study can be applied broadly and create new levels of detail in an important realm of urban and environmental analysis.

Up to now, much of our understanding of the impact of air pollution on population health has been based on the relationship between air quality and mortality and/or morbidity rates, in a population which is assumed to be at their home location all the time. Accounting for the movements of people will improve our understanding of this relationship. The findings will be important for future population health assessments.

—Marguerite Nyhan, a researcher at Harvard’s T.H. Chan School of Public Health, who led the study as a postdoctoral researcher at the Senseable City Lab

The result, Ratti notes, is effectively “two different maps” representing exposure to PM, one showing the exposure that a static, home-based population would have, and the other showing the actual exposure levels given the dynamics of urban mobility.

By analyzing the issue in this form, the researchers believe they have demonstrated a new way for city leaders, health officials, and urban planners to obtain data on pollution levels and analyze their policy options.

The study, the researchers suggest, also underscores the significance of analyzing transportation systems in cities. After all, while some of the PM pollution may come from fixed industrial sources, some of it also comes from automobiles. Studies like this one could help planners identify key locations for low emissions zones, congestion charging, and other tools cities have begun using in an attempt to reduce aggregate exposure among people.

The technology firm Ericsson provided data for use in the study and provides support for the Senseable City Lab.


  • Marguerite Nyhan, Sebastian Grauwin, Rex Britter, Bruce Misstear, Aonghus McNabola, Francine Laden, Steven R. H. Barrett, and Carlo Ratti (2016) ““Exposure Track”—The Impact of Mobile-Device-Based Mobility Patterns on Quantifying Population Exposure to Air Pollution” Environmental Science & Technology doi: 10.1021/acs.est.6b02385



It seems reasonable: they know where and when people are so they can correlate this with pollution (or site pollution monitors where lots of people are) and get a more accurate measure than a blanket approach.

Idea: would it be a good idea to have large fans in crowded areas to blow the pollution away (at rush hour, for instance)?


Measuring pollution intensity with mobile phones could be a very low cost way of doing it..

Prevention is the best way to reduce pollution. Moving is around is a quick local fix?

The comments to this entry are closed.