Insights / Urban climate

Why local climate zones need a grid, not a map

Administrative boundaries are the wrong unit for describing urban heat. Here's what happens when you use them anyway, and why the field moved on.

Ask a city where its heat islands are and you'll usually get an answer shaped by whatever boundaries the city already tracks: neighbourhoods, census tracts, postal codes. That's a reasonable place to start, because it's the data everyone already has. It's also, for this specific question, close to the wrong unit of measurement.

Local Climate Zone classification, the framework Stewart and Oke formalised in 2012 and that WUDAPT has since standardised globally, exists to describe urban form at the scale that actually drives local temperature: building density, height, surface cover, in patches of roughly one to a few hundred metres. That's a deliberate choice, not an approximation. Heat behaves locally. A dense cluster of mid-rise buildings a few blocks from an open park can sit in a meaningfully different thermal regime, even if both fall inside the same neighbourhood boundary on a map.

Administrative units erase exactly that variation. A neighbourhood with a handful of towers scattered among mostly low-rise housing will average out to "mixed" or "low-rise" once you aggregate to the boundary, because the towers are a small share of the total area. The result reads as a smooth, plausible-looking map. It also hides the one thing you were trying to find: the specific blocks where the built form is doing something different from its surroundings.

A fixed grid doesn't have this failure mode, because it isn't drawn around anything. A 100 to 250 metre cell doesn't care whether it happens to straddle two neighbourhoods or sit entirely inside one. It just describes what's physically there. That's the whole point of the WUDAPT Level 0 methodology: classify the grid, not the map that happens to already exist for an unrelated administrative purpose.

None of this means administrative boundaries are useless. They're often the right unit for policy, budget, and reporting, because that's how cities are actually governed. The mismatch only bites when you use a governance boundary to answer a physical question it was never drawn to answer. The fix isn't a better boundary. It's using a grid for the physical question and a boundary for the governance one, and being honest about which one you're looking at.

See a live grid-based classification across our pilot region, or get in touch if you want to know how this generalises to your own city.