Urban heat

Urban heat islands depend on the medium being measured and on the comparison area. Arnfield’s review describes measurements ranging from individual surfaces and streets to neighbourhoods, and explains why different sensing techniques describe different thermal environments. A single temperature difference should therefore be accompanied by the spatial scale, measurement method and reference location used to obtain it 1.

Peng and colleagues examine satellite-derived surface temperatures across 419 large cities. They report an annual average daytime urban–suburban difference of 1.5 ± 1.2°C and a nighttime difference of 1.1 ± 0.5°C. Daytime and nighttime intensities were not correlated across cities. These are surface measurements and cannot be read as the temperature difference in the air people breathe 2.

Zhou and colleagues study 32 Chinese cities using satellite observations from 2003 to 2012. Their estimated heat-island footprint, including the urban area, averages 2.3 times the urban extent during the day and 3.9 times at night. They also report substantial variation across places and times. The finding makes the choice of rural reference an important part of interpreting a city’s heat-island estimate 3.

Hsu and colleagues combine summertime surface-temperature and census-tract data for 175 large urbanized areas in the continental United States. In all but six, the average person of color lives in a tract with higher surface heat-island intensity than the average non-Hispanic white resident. They find a similar disparity when comparing households below the poverty line with those above twice that threshold. This describes unequal exposure in the study population, not an estimate of heat-related deaths 4.

Li and colleagues’ global greenness study distinguishes the cooling capacity of a city from the benefit experienced by an average resident. Across approximately 500 large cities, the resident-level daytime surface-cooling benefit averages 2.2°C in the Global South and 3.4°C in the Global North. The study attributes the gap to differences in green-space quantity and quality shaped by socioeconomic and natural conditions. These are surface-cooling measures, not indoor temperatures 5.

Li and Bou-Zeid combine observations and modelling to examine heat waves and urban heat islands together. They find that heat waves can increase the urban–rural temperature difference as well as raise background temperatures. Their analysis connects the additional urban warming to limited surface moisture and lower wind speeds. This supports evaluating hot-weather conditions alongside the urban form, rather than treating the two effects as independent 6.

Gunawardena and colleagues distinguish cooling within the urban canopy from effects higher in the atmosphere. Their synthesis finds that vegetation’s evaporation and transpiration matter for canopy-layer cooling, while surface roughness also changes convection. The size, distribution and geometry of green space influence how cooling spreads. The review also warns that water bodies may contribute to warming at night under some conditions, despite useful daytime cooling 7.

Venter and colleagues compare satellite surface temperatures with crowdsourced weather-station air temperatures in 342 European urban clusters during the 2019 heat wave. Mean surface heat-island intensity was 1.45°C, compared with 0.26°C for the canopy-layer air measure. Their warning concerns the use of surface measurements as a proxy for human heat exposure. It does not imply that every satellite-based finding or health-risk estimate is overstated by the same factor 8.

Streutker’s Houston study used early-morning satellite observations and found that the spatial extent of its modelled surface heat island was independent of its magnitude and rural temperature. Zhou and colleagues examine how the footprint declines toward rural areas across Chinese cities. These studies use different places, designs and questions. Their results should not be presented as a direct scientific disagreement merely because one discusses extent and the other intensity 93.

Li and colleagues model green and reflective roofs over the Baltimore–Washington region during a June 2008 heat wave. Both approaches reduce surface heat-island intensity more than the near-surface air-temperature difference. Green roofs with ample soil moisture perform comparably to cool roofs with an albedo of 0.7 in that model. Irrigation can provide additional cooling when unmanaged soil moisture is low, making water availability a material condition of the result 10.

Kumar and colleagues review 202 studies covering 51 kinds of green, blue and grey infrastructure. Their synthesis identifies substantial cooling potential but uneven research coverage across intervention types. Gardens, wetlands, walls, trees and other approaches are represented by different studies and contexts. The review supports examining co-benefits and local conditions alongside temperature effects; it does not turn every reported category average into a reliable forecast for a new site 11.

Rawat and Singh review cool-roof performance across climate zones and report building energy savings ranging from 15% to 35.7% in the literature they surveyed. These are energy-use outcomes from that review, not an estimated saving for all buildings. A local appraisal still needs the building’s starting condition and climatic context, and should keep energy consumption distinct from roof-surface and outdoor-air temperature measures 12.

Wu and colleagues test a spectrally selective textile designed to limit heat gained from surrounding urban surfaces. In simulated outdoor urban conditions, the vertically placed fabric was 2.3°C cooler than a solar-reflecting broadband emitter. That comparator is another radiative-cooling material, not conventional clothing. The result is a material-temperature comparison; the abstract does not establish a corresponding reduction in body temperature, illness or city-wide heat 13.

Park and colleagues combine satellite imagery, gridded air temperatures and deep-learning local-climate-zone mapping across 157 South Korean cities. Their map achieves overall classification accuracy of 0.78. The study illustrates a route toward comparing urban form with canopy-level thermal conditions across a country. Classification accuracy is not a cooling effect, and the map’s role is to improve assessment rather than demonstrate the success of an intervention 14.