Why Cities Cook: The Urban Heat Island

AEveryone who has crossed from a park into a city street on a still summer night has felt the difference, and the textbook account of why it exists is not seriously disputed: brick, asphalt and concrete absorb radiation through the day and give it back after dark, while narrow streets trap the outgoing heat and a shortage of vegetation removes the cooling that evaporation would otherwise supply. What is far less secure is the number attached to the effect. An urban heat island is defined as a difference, and a difference has two ends, so every published value depends on the choice of a rural station against which the city is measured. Textbooks and press releases nonetheless treat the resulting figure as though it were a property of the city alone, in the way that a building has a height. The distinction is not pedantic. Between 1995 and 2005 the quoted intensity for large European cities drifted upwards in the literature without any obvious change in the cities themselves, which ought to have prompted more suspicion than it did.

BThat suspicion is the starting point for Priya Venkataraman, a climatologist at the Halden Institute for Urban Climate, who contends that a large share of the reported warming is an artefact of where the thermometer outside the city happens to stand. Reference stations, she points out, are usually sited on managed farmland, and irrigated crops cool the air above them by evaporating water that unmanaged ground would never release, so the baseline is not merely rural but abnormally cool. Her group reprocessed records from sixty-eight cities gathered between 2011 and 2019, discarding every reference site under irrigation and substituting unwatered grassland at a comparable elevation. The median night-time intensity fell from 3.2 to 1.9 degrees Celsius. The correction was largest in dry regions, where watered fields contrast most sharply with the land around them, and almost negligible in the humid tropics, where vegetation outside the city needs no help. Venkataraman is explicit that she is not denying the phenomenon, only the size of it. Her quarrel is with the arithmetic rather than with the physics.

CDaniel Achterberg, who runs the Brenner Laboratory for Boundary-Layer Studies, is unconvinced that a change of reference site can account for something he regards as thermodynamically obvious. Cities burn fuel, and the heat released by vehicles, boilers and cooling machinery has to go somewhere; his instruments on a ninety-metre tower in a dense commercial district recorded an average release of 32 watts per square metre through the winter of 2016, rising above 80 watts on the coldest weekday mornings. On his account the island is manufactured rather than merely stored. Venkataraman does not challenge the tower measurements, and concedes that on winter nights in a compact centre such releases may dominate, but insists that the figure cannot be carried across to summer, to suburbs, or to cities whose energy goes on cooling rather than heating. Achterberg's evidence, in fairness, comes from one district of one city in a single season, and a tower samples only the air that happens to pass it. The disagreement is therefore one about scope.

DA second confusion has done more practical damage than the first. Satellites do not measure the temperature of the air; they measure the temperature of whatever surface they can see, and on a July afternoon a dark roof may sit twenty-five degrees above the air a metre above it, while a shaded pavement sits below it. The two quantities are correlated, loosely, but they peak at different hours and are not interchangeable in the way that colourful satellite maps encourage. Air temperature differences between city and countryside are greatest a few hours after sunset, because open land cools quickly once the sun has gone while masonry continues to release what it took in during the day; surface differences, by contrast, are greatest at midday. The error is systematic rather than random. A council choosing which districts to treat from a midday satellite image will therefore rank its neighbourhoods by roof colour rather than by the night-time warmth that keeps residents from sleeping, and the two rankings, as several municipal audits since 2018 have found, frequently disagree.

EWhat follows for policy depends on which hours are being defended. Reflective roofing lowers the surface temperature of a building and cuts the heat entering it, which is why its measured benefit is concentrated in the middle of the day and why it performs best in dry, sunny climates; it does comparatively little to the night-time air that matters most for health, because a pale roof stores less heat but also gives up what it has more quickly. Street trees work differently, cooling by shade and by transpiration, the loss of water through their leaves, but only where the soil holds enough moisture for that loss to continue, which in the driest weeks of a heatwave is exactly when it does not. Shade, however, is not free. A programme in one Mediterranean city planted eleven thousand trees between 2013 and 2017 and had to irrigate a third of them through every August, at a running cost that quietly overtook the planting bill. Excess deaths in a heatwave are governed less by afternoon maxima than by night-time minima, which suggests that the two measures are not competitors so much as answers to different questions, and that a city buying only one of them has not really chosen.