Day vs. night land surface temperature: which matters more for cooling center siting

If you've pulled land surface temperature data for a heat vulnerability assessment, you've probably noticed the daytime and nighttime passes don't agree with each other. A parking lot and a big-box roof will blow past everything else on the daytime image. Come back at 3am and those same pixels have dropped close to ambient, while the dense rowhouse blocks and older multifamily stock are still running several degrees warm. Both images are real. They're measuring different physics, and for cooling center siting, one of them tells you a lot more than the other.

What each pass shows

Daytime land surface temperature is mostly a story about materials and sun exposure. Asphalt, metal roofing, and bare rooftops absorb shortwave radiation and re-emit it, so daytime LST maps light up industrial parks, big parking fields, and anywhere tree canopy is thin. It's a useful layer for spotting heat sources and figuring out where shade trees or reflective roofing would knock down the surface temperature fastest.

Nighttime LST is a different measurement. By a few hours after sunset, the sun's contribution is gone and what you're looking at is thermal mass: how much heat a block's buildings and pavement absorbed during the day and how slowly they're giving it back. Brick, concrete, dense multifamily construction, and narrow streets with little sky exposure all retain heat and release it slowly overnight. That's why older, denser neighborhoods often read warmer at night than suburban blocks that ran hotter at noon.

For siting, the nighttime pass is the one that lines up with the health outcome you're trying to prevent. Heat-related illness and death track closely with nighttime low temperatures, because that's when a body that's been stressed all day is supposed to get a few hours of recovery. A block that cools to 68°F overnight gives residents without AC a break. A block that only drops to 80°F doesn't, no matter how it looked at 2pm. If you're ranking blocks for where a cooling center or a cooling-focused outreach route would do the most good, nighttime retention is a closer proxy for physiological stress than the daytime hot spots that tend to dominate a first glance at the imagery.

Building that into placement criteria

None of this means daytime LST is useless. It's the right layer for siting shade trees, reflective pavement pilots, and other interventions aimed at reducing the heat a block generates in the first place. But if the question is where to put a cooling center, or which blocks get priority for extended hours and transportation support, nighttime thermal imagery combined with building type and demographic density gets you closer to the answer. A block with brick walk-ups, a high share of residents over 65, and poor overnight cooldown is a different priority than a block that spikes at noon but sits next to tree cover and empties out to single-family homes with central air by evening.

The building type layer matters here because it explains the nighttime signal instead of just confirming it. Thermal mass and retention vary by construction, so cross-referencing night LST against building footprints tells you whether the heat is coming from the structures themselves, which changes what kind of intervention helps versus a block where the problem is really about who's living there without access to AC in the first place.

Seasonal cadence matters too. A single summer night is weather. A pattern that repeats across the hottest weeks of the season, block after block, over multiple years, is what should drive a cooling center budget. One-off heat events don't tell you where to put a permanent asset.

Heat Exposure Mapping layers seasonal nighttime and daytime surface temperature against building type and demographic density, so the blocks that run hot and hold their heat overnight, with the residents least able to ride it out, are the ones that show up first.

If you're building the case for where the next cooling center or shade-tree investment goes, that's the layer worth starting with.