The observation that towns are warmer than the country around them is among the oldest in climatology, and its physical causes are well understood. Built surfaces absorb and store more solar energy than vegetated ones, street canyons trap radiation and slow the loss of heat to the night sky, impervious ground sheds water that would otherwise evaporate, and buildings and vehicles release waste heat (Oke, 1982). The result is an urban heat island, most pronounced in air temperature on calm, clear nights and in surface temperature on sunny days.

For cities in the tropics the phenomenon carries particular weight. Background temperatures are high in every month, so any urban increment is added to conditions already near the limits of comfort, and high humidity restricts the body's ability to lose heat by sweating. Sherwood and Huber (2010) showed that there is an upper bound of wet-bulb temperature beyond which the human body cannot dissipate its own metabolic heat, and the humid tropics lie closer to that bound than any other inhabited region. Yet tropical cities have long been under-represented in urban climate research (Roth, 2007), and guidance on mitigation has largely been imported from the mid-latitudes.

Two measures dominate that guidance: raising the solar reflectance of roofs and increasing tree cover. Both were set out together in an influential paper by Akbari et al. (2001), and both now appear in the heat action plans of many cities. This review considers what the published evidence says about each, with attention to the features of tropical climates and tropical cities that may strengthen or weaken the case.

The article is a narrative review. It draws on the published literature in urban climatology, building physics and urban forestry, including foundational papers, earlier reviews, field studies from tropical cities and numerical modelling. No systematic search was carried out and the selection reflects a judgement about which sources best illuminate the question. Green roofs, cool pavements and water bodies are mentioned only where they help to interpret the evidence on the two measures of interest. The review is concerned with outdoor and indoor heat exposure and does not attempt an economic appraisal.

The urban surface energy balance partitions net radiation and anthropogenic heat among sensible heating of the air, latent heat consumed in evaporation, and storage in the fabric of the city (Oke et al., 2017). Mitigation measures work by changing one of these terms. Taha (1997) identified albedo, evapotranspiration and anthropogenic heat as the three levers available. A reflective roof reduces the net radiation absorbed. A tree intercepts sunlight before it reaches the ground and walls, and diverts part of the energy it absorbs into evaporating water. Each acts on a different surface at a different height.

It is also necessary to say which heat island is meant. The surface heat island seen by satellite thermal sensors is largest by day and reflects the temperature of roofs, roads and tree crowns viewed from above. The canopy-layer heat island measured by thermometers at street level is largest at night (Voogt and Oke, 2003). A measure can sharply reduce one and barely alter the other. Comparison across cities is further complicated by the choice of rural reference, a problem that the local climate zone scheme addresses by classifying measurement sites according to their structure and surface cover (Stewart and Oke, 2012).

Roth (2007) reviewed urban climate research in tropical and subtropical regions and found the body of work small and uneven in quality. Where heat island intensities had been measured they were generally lower than in temperate cities of comparable size, and they varied with the seasonal cycle of rainfall, tending to be larger in the dry season when the surrounding countryside is dry and cools quickly at night. Observations in Singapore, one of the better studied equatorial cities, show a nocturnal heat island that peaks some hours after sunset and differs among land uses and seasons (Chow and Roth, 2006).

Background climate shapes the daytime picture as well. Zhao et al. (2014) found that in humid regions the daytime heat island is strengthened because cities are aerodynamically smoother than the tall vegetation around them and so shed heat to the atmosphere less efficiently. Manoli et al. (2019) explained much of the variation in heat island intensity among cities worldwide by population and mean precipitation. Their analysis implies that in wet tropical climates, where the surroundings already evaporate freely, adding vegetation to the city yields diminishing returns in air temperature, and that measures based on reflectance, shade and ventilation deserve corresponding emphasis. These findings caution against assuming that a result from a dry or temperate city will hold in a humid one.

A cool roof combines high solar reflectance with high thermal emittance, so that it absorbs little sunlight and readily radiates the heat it does absorb. White coatings and membranes are the simplest form. Coloured products that reflect strongly in the near-infrared part of the spectrum are also available.

Trees modify the urban climate through two processes that are often conflated. Shade reduces the solar radiation reaching people, paving and walls. Transpiration converts absorbed energy into latent heat. Their relative importance changes with climate, and that change is the key to judging trees in the tropics.

Set side by side, the two measures have little in common beyond their purpose. Table 1 summarises the contrasts that emerge from the literature reviewed above.

The contrasts point to complementary roles. Cool roofs protect the occupants of the building, especially those on the top floor of low-rise housing, and they do so at once and at low cost. Trees protect people outdoors, on the routes they walk and in the places they wait and trade. Where single-storey buildings stand among mature trees the two interact, since shade on a roof reduces what reflectance has left to do. Figure 1 groups the main effects of each measure and the conditions on which both depend.

A common weakness in the evaluation of both measures is the reliance on air temperature as the outcome. Air temperature is easy to measure and to model, but it captures neither the radiant relief provided by shade nor the indoor conditions beneath a roof. Indices of thermal comfort that combine air temperature, humidity, wind and radiation give a fuller account, and in humid climates they can rank interventions differently from air temperature alone.

Several gaps follow from the review. Field measurements from tropical cities remain scarce compared with model studies, and those that exist are concentrated in a few wealthy locations. Long-term records of the reflectance of roofing materials under humid tropical exposure would allow maintenance needs to be planned realistically. For trees, measurements that separate shade from transpiration, and that follow conditions through the night, are needed to settle how canopy should be arranged where ventilation is scarce. Studies of indoor temperature in unconditioned low-income housing before and after roof treatment would address the population with the most to gain. Finally, almost all modelling treats the two measures separately, and their combined effect in districts of realistic form has received little attention.

Cool roofs and tree canopy both have a sound physical basis and a supporting literature, but neither transfers to the tropics unchanged. Cool roofs lose nothing in a climate without winter and gain relevance where low-rise, lightly roofed housing is common, provided their reflectance is maintained. Trees remain valuable chiefly for their shade, and their arrangement must respect the need for air movement and night-time cooling. Planners in tropical cities are better served by asking where people are exposed to heat, indoors and out, and choosing the measure that acts at that place, than by ranking the two against each other on air temperature.

To be declared by the author(s) before publication.

To be declared by the author(s) before publication.