Assessing the Case for a Citywide Heat Sensor Network with NYC Office of Technology & Innovation
Written by Stacey Li
Heat is New York City’s deadliest weather hazard
Heat contributes to roughly 500 deaths in New York City each year, making it the deadliest weather-related hazard the city faces (NYC Health Department). Most of those deaths are never recorded as "heat-related" because heat usually kills indirectly by worsening an existing condition like heart disease (2026 NYC Heat-Related Mortality Report)..
Heat also strains the city's infrastructure. It pushes the power grid toward record demand (Con Edison), and can even warp steel rail, forcing trains to slow down (The Independent).
And extreme heat is only going to become more frequent. NYC is projected to see 17 days a year above 95°F by 2030 and 27 by 2040, compared with just four historically (Con Edison).
Asking city agencies what they need
As heat becomes an increasingly urgent risk, the New York City Office of Technology and Innovation’s Smart Cities and IoT team is exploring whether a citywide heat sensor network could be valuable to city services and operations. As a Siegel PiTech PhD Impact Fellow placed with the team, I set out to 1) map the current landscape of urban heat monitoring and 2) assess NYC agencies’ needs for enhanced heat monitoring solutions.
I began with a literature review to capture existing heat monitoring and response systems, both within NYC and internationally. From there, I interviewed staff across NYC government agencies and academic partners in two rolling phases, guided by a series of interview questions, to gauge interest in a citywide heat sensor network and gather requirements for what it would need to do.
New York City doesn't lack heat data. Between the NYC Department of Health and Mental Hygiene’s Heat Vulnerability Index and the NYC Micronet (a network of street-level weather stations), there are datasets city agencies can draw on. The question is whether more detailed data would functionally change what agencies do.
What we heard
These were the four things that stood out:
1. Interest in new data is not universal
While our interviews revealed that city agencies were interested in additional heat data, there was no consensus on how much, or what kind.
Some agencies want more real-time coverage of meteorological data. Others have collected heat data (it’s published on NYC Open Data), and it's not clear to them that collecting more would change how they operate.
2. Being open to new data isn't the same as being able to act on it
Some agencies said, in principle, they would welcome more granular heat data but had no immediate use for it. Their levers, mostly zoning and slower-moving policy changes, don't operate on the timescale or geography that street-level sensors are built for. A block-by-block temperature reading doesn't help much if your fastest tool moves at the pace of a zoning amendment.
Even where the timescale fits, capacity often doesn't. Several teams told us they lack the in-house expertise or bandwidth to get real use out of raw heat data. Handing an agency more numbers doesn't help if staff don’t have the time to analyze them. Analysis capacity turns out to be as limiting a factor as data availability itself.
3. More data doesn't automatically mean more funding for interventions
Agencies described repeated, unsuccessful efforts to use heat and demographic data to apply for Federal Emergency Management Agency (FEMA) grants to fund air conditioning and heat pump retrofits (E&E News by Politico). FEMA's benefit-cost formula requires tying heat data to a quantified loss of life or property at a specific site (FEMA), a bar that hyperlocal sensor readings don't clear on their own.
For other agencies, the bottleneck isn’t data at all. It’s funding for the interventions themselves.
4. Agencies want a way to measure before and after
This was a consistent ask across conversations: temporary sensor deployments that capture baseline and post-intervention readings at the same site.
Several city heat interventions, such as cool pavement pilots and alternative building materials, are upcoming or already underway. In principle, these interventions need before/after data to show how effective they are. Nobody currently has a repeatable, deployable method for before/after comparison.
Taken together, these findings point toward a specific kind of gap: a lack of infrastructure that matches how city agencies work. Some need continuous coverage; others need a way to test one location before and after an intervention. Some can process raw data; others want it interpreted for them. Below are three recommendations, each targeting one part of the mismatch.
Recommendation 1: Build on Micronet's existing infrastructure
The NYC Micronet is the New York City part of the New York State Mesonet, a statewide network of weather stations that provides hyperlocal data on temperature, humidity, and other conditions. Right now, Micronet's data primarily supports utility operations for Con Edison and feeds into broader weather forecasting, including flood prediction during major storms like Hurricane Ida in 2021 (University at Albany). Several agencies we spoke with wanted more real-time meteorological coverage than currently exists. This existing infrastructure offers a practical foundation for expanding heat data collection. Con Edison funds and sites Micronet's stations while SUNY Albany owns, operates, and quality-controls them. The university brings years of technical and scientific groundwork to this kind of network. Extending what already works is a faster way to increase data collection than standing up a parallel network.
Recommendation 2: Pair the stationary network with a library of sensor types agencies can draw from as needed
Agencies didn't agree on how much data they needed or what kind, and one reason is that no single sensor type serves every use case. Mobile sensors capture detail across many locations but only a snapshot in time; stationary sensors offer continuous data but need far more units to match that same coverage. Establishing a shared collection of mobile or temporary sensors agencies can deploy to complement existing stationary networks would let the City meet a wider range of agency needs without duplicating infrastructure.
Recommendation 3: Offer more than one way to receive the data
As several teams told us, capacity to interpret heat data varies too much for a single format to work for everyone. This effort should support two tiers: raw data and Application Programming Interface (API) access for agencies with existing analytical capacity, and ready-to-use summaries for agencies without it.
None of this closes every gap we learned about, especially funding for interventions, which sits outside the control of a citywide heat monitoring effort. It does, however, target the gap we can close: giving agencies a shared, extensible way to measure whether their interventions are working, instead of asking them to act on data they don't have the tools or the mandate to use.
Reflection
My background is in HCI and system design, and most of my research starts the same way: understand what people need, then design something toward it. This fellowship put that approach into a new context, designing for a city government, where the first question was whether the thing should be built at all, and for whom. That required a different kind of research rigor up front. I interviewed stakeholders across agencies, synthesized conflicting needs, and scoped a vague question into something actionable.
Government also introduced constraints I hadn’t designed around before, from considering whether a location makes sense scientifically and whether the agency managing that infrastructure would allow a sensor to be mounted there, to knowing exactly what interventions are needed but being unable to prove it in the format grant formulas require.
That gap between what research can show and what institutions can act on sharpened my sense of what designing for real deployment means. OTI will use these recommendations to evaluate options for expanding the city’s heat monitoring infrastructure. Getting to see that handoff from research to policy recommendation to potential city action grounded the work in a way I didn’t expect.
Stacey Li
Ph.D. Student, Information Science, Cornell University