Ambient air quality
Conventional inventories can show broad trends, but they often smooth out the stop-start dynamics that make urban traffic so polluting in the first place.
Direct tailpipe measurement, meanwhile, can produce highly detailed results but is hard to expand across an entire city.
A new study from researchers at MIT and partner institutions points towards a different model: using existing urban data streams to estimate traffic emissions in near-real time, at the level of individual roads and even individual hours.
Applied in Manhattan, the framework drew on traffic-camera imagery and mobile-phone location data to reconstruct how vehicles moved through the road network, then combined that with emissions modelling to produce hyperlocal estimates of carbon monoxide, carbon dioxide, nitrogen oxides and PM2.5.
What makes the study especially relevant for environmental monitoring professionals is not just the sophistication of the model, but the fact that it relies largely on infrastructure cities already have.
The researchers’ argument is that high-resolution emissions intelligence does not necessarily require a vast new hardware roll-out. In some cases, it may come from integrating existing camera assets, mobility data, vehicle classification, and emissions factors far more effectively than cities do now.
The results suggest that many cities may be underestimating the scale and unevenness of traffic pollution. In the Manhattan case, the new estimates were reported as 23 to 49% higher than US EPA figures, while also correlating well with data from the city’s relatively sparse ambient air quality sensor network.
One reason is that traditional methods often fail to capture what happens around traffic signals. The study found that ignoring fine-grained inputs such as signal timing, speed variation and fleet heterogeneity can introduce uncertainties ranging from -49% to +25%, and that omitting traffic-light effects can underestimate some pollutants by about 50%.
For monitoring professionals, that is a significant warning. If the underlying traffic model is too simplified, the policy conclusions drawn from it may be wrong as well.
That matters because cities are increasingly expected not just to declare climate measures, but to prove that they work. The researchers used the same framework to assess Manhattan’s congestion-pricing scheme, which began in January 2025 in the Congestion Relief Zone south of and including 60th Street.
Eight weeks after implementation, traffic volumes were down by about 10%, while emissions fell by 16% to 22%, depending on pollutant. Just as important, the reductions were not evenly distributed. The biggest gains appeared on higher-order roads and near major entry points, while some areas outside the zone saw smaller benefits or modest increases.
That kind of spatial granularity is exactly what city authorities need if they want to distinguish between a policy that works on average and one that works equitably.
For the monitoring sector, the implication is clear: urban transport policy is becoming a verification problem. It is no longer enough to publish an annual emissions inventory and assume that the broad direction of travel is understood. Cities will increasingly want systems that can show what changed, where it changed, when it changed and who benefited.
That creates a larger role for integrated platforms combining roadside sensing, camera analytics, traffic-modelling software, meteorological context and ambient air quality measurements.
It also strengthens the case for strategically placed reference-grade monitors and low-cost sensor nodes that can ground-truth modelled emissions hotspots and detect spill-over effects in neighbourhoods outside a charging zone or low-emission area.
The study itself notes that the new estimates aligned with sparse air quality sensor data, which points to a future in which modelling and measurement are used together rather than treated as alternatives.
If cities begin to view transport decarbonisation as something that must be monitored almost operationally, they may start buying differently. Instead of commissioning one-off inventories or isolated pilot studies, authorities may look for interoperable data ecosystems that connect cameras, mobility data, emissions models and air-quality monitoring into a continuous policy-assessment workflow.
Vendors able to support calibration, validation, sensor siting and data fusion could benefit as cities try to convert big-data experimentation into routine decision support. The technology challenge, in other words, is shifting from simple detection to defensible attribution: demonstrating that a traffic intervention produced a measurable environmental outcome, not just a plausible one.
The equity dimension may prove just as important as the climate one. New York’s own congestion-pricing programme has already been linked to targeted air-quality improvement measures in the Bronx, including replacement of diesel transport refrigeration units, electric truck-charging infrastructure, school air filtration and other mitigation projects.
That underlines a broader point. Hyperlocal emissions mapping is valuable not only because it can show where pollution falls, but because it can reveal where benefits fail to arrive. For environmental monitoring professionals, that could make future urban networks less focused on citywide averages and more focused on exposure, distribution and environmental justice.
In that sense, the Manhattan study is best understood not as a substitute for conventional air quality monitoring, but as a sign that the market is changing around it. Cities still need trusted measurements.
They still need defensible instrumentation and calibrated networks. But they also need faster feedback loops than traditional inventories can provide. The more climate policy moves into the street, the more monitoring will have to do the same.
IET 36.2 Mar/Apr 2026
Product directory
.jpg)
