Urban Heat Island Mapping

Replace fields and trees with concrete, asphalt and metal and you change the surface energy balance. Dark, dense, sealed materials absorb solar radiation through the day and release it slowly into the night; buildings trap heat in street canyons; and paving that sheds rainwater leaves little moisture for the evaporative cooling that vegetation and soil provide. The result is the urban heat island (UHI): built-up areas measurably warmer than their rural or vegetated surroundings, with the peak over the densest, most sealed fabric and cool dips over parks and water.

Two distinctions matter. First, surface UHI (skin temperature, what a satellite measures) is not the same as air UHI (what a thermometer at head height feels), though they are related. Second, UHI is a relative statement - city warmer than its surroundings - so any UHI map depends on where you draw the rural baseline. Say which baseline you used.

Optical bands record reflected sunlight; a thermal infrared (TIR) band records emitted heat - the longwave radiation a surface gives off because of its own temperature. Landsat carries such thermal bands, and they are the free workhorse for city-scale land-surface temperature (LST). The catch is that the thermal band is coarser than the optical bands, so LST is a neighbourhood-scale signal, not a per-building reading.

Getting from raw thermal data to LST is a short, honest chain. You convert the band's recorded values to brightness temperature (the temperature a perfect emitter would need to give off that much radiation), then correct for emissivity - how efficiently a real surface radiates heat, which differs between water, vegetation and concrete. Emissivity is commonly estimated from NDVI (greener pixels get a vegetation emissivity), which is why the last lesson feeds directly into this one. The output is a temperature raster you can map, contour and compare.

A heat map on its own is striking but shallow - it shows where it is hot, not why. The explanation usually lies in two layers you already know how to make. Imperviousness - the share of ground sealed by roofs, roads and paving - tracks LST closely: the more sealed a neighbourhood, the hotter it tends to run. NDVI works the other way: greener, moister surfaces run cooler through evaporative cooling and shade.

Overlay LST with imperviousness and NDVI and the causal story becomes visible and defensible: this hot ridge is the sealed industrial belt; that cool notch is the lake and its park. You can quantify it - mean LST per land-cover class, or a plot of temperature against percent-sealed - and turn a dramatic map into an argument for specific, located interventions. Global built-up products such as the GHSL can supply imperviousness where you lack a local layer.

Thermal remote sensing rewards humility. LST is a surface skin temperature at the satellite's overpass moment (often late morning), not the 3 p.m. air temperature a resident endures, and not a night-time value unless you use a night scene. Its coarse resolution means you read districts and corridors, not individual plots. Clouds block thermal sensing entirely, and a single date can be unrepresentative - averaging several clear summer scenes gives a steadier picture.

Used within those limits, LST is invaluable. It reveals heat corridors and hotspots at a scale no ground survey could cover, flags the neighbourhoods most exposed during heatwaves, and provides a baseline to test whether cool roofs, tree cover or a new park actually lowered surface temperature. In Indian cities facing rising heat stress, that city-wide, repeatable view is exactly the evidence heat-action planning needs.