The urban heat island, explained: why your block is getting hotter every year

Your block is getting hotter for two reasons, and at neighborhood scale the local one is the larger of the two. Pavement and roofs replaced shade and soil. On one afternoon, the gap between the hottest and coolest parts of a single US city now reaches 17°F, against about 2.7°F of global warming since the mid-1800s.

That local gap is the urban heat island (the extra heat a built-up area holds compared with the land around it). It is the part a city, a developer, or a property owner can change on a ten-year horizon.

Why the global temperature record leaves out your block

The global temperature record cannot show the heat on your block because the people who build that record subtract the urban effect on purpose. Where a city has grown up around a weather station, that station's long-term trend is replaced with the trend from rural stations nearby. NASA (the US National Aeronautics and Space Administration, which keeps one of the global records) compares each urban station with rural stations up to 1,000 kilometers away. Its own estimate puts the urban contribution to the global average at a hundredth of a degree. The urban share is that small because rural stations cover most of the land surface the record draws on. The neighborhood record is built the opposite way, by volunteers who drive sensor-equipped cars through a city on one of its hottest days. NOAA (the National Oceanic and Atmospheric Administration) has supported those campaigns in more than 70 communities between 2017 and 2023, with campaigns continuing each summer since. Their block-by-block results show the gap the global record removes.

Each of the two records was built to answer a different question, and the two questions have little overlap. The global record measures whether the planet is warming. The neighborhood record measures why one Houston neighborhood ran 14°F hotter than another at the same hour on a campaign day in 2024. That second question decides whether a building's tenants sleep and what its cooling bill runs. The second record rarely enters the public argument about warming, because that argument is conducted in the units of the first. I would say that is the main reason the heat island stays underweighted in planning: the number people argue about is the one that has had the urban effect subtracted out of it.

What makes one block hotter than the next

One block runs hotter than the next because of what covers it, and three features of that covering explain much of the difference. Dark roofs and pavement absorb sunlight all day. The trees and soil that used to cool the air are gone. The heat stored in those surfaces is released after sunset and keeps the night warm. The EPA (the US Environmental Protection Agency) puts the effect across US cities at 1 to 7°F by day and 2 to 5°F by night. The night figure matters most for health. Daytime heat can be escaped indoors, while a night that stays warm prevents the body from cooling down and recovering. That pattern contributes to the 702 heat deaths a year the Centers for Disease Control counted on average from 2004 to 2018. NOAA has ranked heat as the leading weather-related killer in the US for three decades.

Which blocks run hottest was largely settled by earlier decisions about where to pave and where to plant. Climate Central (a climate research group) puts nearly 34 million people, across the 65 largest US cities it analyzed, in neighborhoods at least 8°F hotter because of the built environment. A Columbia University analysis found neighborhoods redlined in the 1930s (graded as hazardous for mortgage lending in the federal surveys of that decade) running 6.5°F hotter on average than the top-graded neighborhoods of the same cities. A block's heat level was set by its lending and land-use history long before the global trend added anything to it.

Three different temperatures get called "the heat island" in public debate, and a project ends up paying for whichever one it chose to measure. Surface temperature is what a roof or a road reaches, and the EPA notes conventional roofing can run 66°F above the surrounding air on a warm day. Air temperature is what a person breathes at head height, and it is the quantity the day and night figures above describe. Radiant temperature is the heat load that reaches a body from the sun and from the hot surfaces around it. In a dry, sunny city it governs how a pedestrian feels far more than the air reading does.

Temperature	What is measured	What it drives	A measured figure
Surface	The material itself: roof, road, parking lot	Roof/pavement life, stormwater warmth, how much heat releases at night	Conventional roofing up to 66°F above air (EPA)
Air	The air at head height	Sleep, cooling load, the readings in heat-survey campaigns	1-7°F by day, 2-5°F by night across US cities (EPA)
Radiant	Sun plus heat from surrounding surfaces, on a body	Pedestrian comfort and safety in sunny, dry climates	Reflective pavement raised radiant load on pedestrians by 5.5% (Arizona State University, ASU)

Phoenix measured the difference between those three temperatures on its own streets, in a pilot the city ran with ASU. The reflective pavement coating cut noon surface temperature by 12.0°F against untreated asphalt and lowered nighttime air at six feet by half a degree. Radiant load on a pedestrian rose by 5.5%, because the sunlight the road no longer absorbed was reflected upward onto anyone walking on it. A contract written to a surface-temperature target can be met in full while the people on the surface get hotter, because whichever metric the specification names is the one the contractor delivers.

The same study recorded a decay that first-year figures never show, and it changes what a buyer is buying. The coating's reflectivity fell from 33-38% at application to 19-30% ten months later. Untreated asphalt held steady at 12% over the same period. A city or an owner reading that result therefore has a contracting problem as much as a materials problem. A product specified at installation and a product performing in its second summer are two different purchases. Which of the three temperatures a project measures, and on what date after installation, is where our built environment advisory work usually begins.

Built Environment Advisory

Is the heat island getting worse every year?

The heat island in US cities is growing in the surface record, and the growth is measured from satellites. The US Geological Survey (USGS) tracked 50 cities by satellite from 1985 to 2020. It found the urban-rural surface gap widening by 0.58°F per decade, with no significant trend in the surrounding land. That local rate exceeds the 0.37°F per decade at which global air temperature rose over the same 1985 to 2020 window, on our trend fit to NOAA's global land and ocean series. The two are different quantities, satellite surface readings on one side and ground-level air on the other, and adding them would be a mistake. The comparison supports a narrower claim: the land under US cities is heating faster than the land around it, and the global increase adds to that. A parcel on the growing edge of a city therefore collects two increments at once, the local increment (the extra warming from the built surfaces on and around the parcel) and the global increment (the warming that arrives everywhere regardless of what is built). A parcel in a built-out core, where the surfaces stopped changing decades ago, collects mainly the global one.

Part of the widening arises outside the city, on the farmland around it, and that part is easy to misread. A 2021 study in the journal Environmental Research Letters found rural land greening faster than urban land in 189 of 228 megacities worldwide, with the effect strongest where rapid urbanization met cropland intensification (farming the same land harder, which leaves it greener). That divergence explained about a quarter of the surface heat island's growth from 2000 to 2019. Just as an example: a block can register a stronger heat island because the fields outside town got greener while the block stayed as it was. On the parcel itself, canopy share (the fraction under tree cover) and impervious share (the fraction under pavement or roof) are the two figures worth tracking, because those move when someone builds or plants and at no other time.

The heat island grows every year as a decadal trend, and no single summer is guaranteed to run hotter than the one before it... weather sets the order from one August to the next. A cooler summer will follow a hotter one more than once in the coming decade without changing the decadal trend. As I told Al Jazeera in July, in the context of data centers, the current siting and cooling model was designed for average conditions, and average conditions are disappearing. The EPA expects the effect to strengthen as urban areas grow larger and denser. A satellite analysis of 5 million urban grid cells found the global surface heat island intensifying by 0.021°C a year (about 0.04°F) from 2003 to 2018. Both figures describe averages over many years, and neither guarantees that next August will run hotter than this one.

Figure 1. On one hot day, the measured air-temperature gap between a city's hottest and coolest neighborhoods ran 17°F in Richmond, Washington, and Baltimore (2017 and 2018), 14°F in Houston (2024), 13.2°F in Columbus (2022, evening reading), and more than 12°F in Bloomington, Indiana (2024). Divided by the 2.66°F (1.48°C) of global warming in the 2023 to 2025 average against 1850 to 1900, those gaps come to 4.5 to 6.4 times the global figure. A gap between two places at one hour and a change over time are different measurements, so the multiple compares their size rather than adding them. The lower panel sets the USGS surface heat island trend for 50 cities, 0.58°F per decade from 1985 to 2020, beside the 0.37°F per decade global air trend fit to the same 1985 to 2020 window. The two measure different things (satellite surface temperature and ground-level air) and are shown side by side, each on its own line. Campaign figures are single-day snapshots that include topography (Reno's 2024 campaign, left out of this figure, folded a few hundred feet of elevation into a 23°F reading). Analysis: Gasilov Group, our compilation of the campaign results linked in this post and our arithmetic on the ratios, with Celsius converted to Fahrenheit at 1.8 per degree; current as of August 2026.

What an owner or a city can change, and how the market has started to price it

An owner or a city can change the local increment through shade, surface reflectivity, and ground left unpaved, and the market has begun to price the result. Every property listed on Redfin (the home-listings site) now carries a heat-risk score from First Street (a climate-risk data firm) built from surface temperature and land cover. At least one large landlord now reviews heat stress across its whole portfolio at board level each year.

A portfolio owner should track three figures per parcel, canopy share, impervious share, and roof reflectivity, read on the same summer date each year (e.g., the first week of August). The campaign surveys cities rely on are a single-day snapshot that then stands in for years of planning on the presumption that the hottest neighborhoods stay the hottest. A parcel record kept on the same date each year is the check on that presumption.

Shade outperforms coating wherever pedestrians are the concern, and the Phoenix result shows why. The EPA reports shaded surfaces 20 to 45°F cooler than the peak of unshaded materials. A review of 308 studies found urban forests 3.0°F cooler on average than nearby built-up areas. Contracts for reflective roofs and pavements should specify tested reflectivity at twelve months, given the decay Phoenix measured. Cooling budgets belong on the night minimum (the lowest overnight air temperature), since that is the reading that separates a warm summer from a dangerous one.

Insurers and lenders have not yet priced heat as a named peril, and valuations will move once they do. Heat produces no claim event, and a peril with no claim event gets priced through operating cost and coverage terms. Nobody files a heat claim the way they file a flood claim. Because of that, heat shows up in the electricity bill and in the terms of what a policy will cover.

The Urban Land Institute reported in July 2026 that JBG SMITH Properties (the large landlord referred to above) has seen energy costs rise as much as 30% in parts of its portfolio and has written heat into its enterprise risk review. As I told Realtor.com in May, Phoenix buyers do not check Lake Mead levels before making an offer, while insurers, lenders and appraisers have started to, and I expect heat to follow the same sequence I described for water there: a development constraint now, a pricing signal within five years.

The consequence for an appraiser is that a building whose cooling bill rises faster than its rent roll has already lost value, whether or not any policy names heat as a peril.

Miami-Dade County's official heat season was declared to run from May 1 to October 31 each year. The county's stated instrument for shrinking its own island is a canopy target of 30% by 2030. A canopy target is a shade target, and shade produced the largest surface effect in the EPA figures above.

If you are not sure where to start, the Built Environment Strategy Diagnostic, a three-week read, is the entry point. If you already know what you need, get in touch and we can scope it.

Get in Touch

Arif Gasilov is partner for Natural Resources & Built Environment at Gasilov Group, where he works on energy regulatory analysis, water governance, utility rate cases, and the resource footprint of large development. He has been quoted on heat, water, and infrastructure by Al Jazeera, Reuters, Newsweek, Forbes, and Realtor.com. Meet our partners →

Frequently Asked Questions (FAQ):

Can a small town or a suburb have a heat island? 

Yes; the EPA notes heat islands form in small or large cities, in suburban areas, in northern or southern climates, and in any season, wherever paved and roofed surfaces replace vegetation. Intensity scales with size and density, and humid cities in the eastern United States show the largest urban-to-rural differences.

How much does a heat island add to electricity demand? 

The EPA cites increases in air-conditioning demand of 1 to 9% for each 2°F rise in temperature, with the largest increases in countries where most buildings are air conditioned, such as the United States. Peak demand comes on the hottest afternoons, when the heat island and a regional heat wave coincide.