Thermal Comfort (UTCI) Map — Summer-Afternoon Heat in the Browser
"Can anyone sit in this plaza in July?" is asked more and more often in site analysis, and answering it has meant specialist software — SOLWEIG, ENVI-met. The Thermal Comfort sheet answers it at concept-design level. Most of the ingredients are things the other sheets already compute; nothing new is downloaded.
This is an explanation of what the map calculates and, more importantly, which way it can be wrong. Do not skip the bias section. Thermal comfort says "you can sit here", and misreading it costs more than misreading a wind plate.
What UTCI is
UTCI — the Universal Thermal Climate Index — folds air temperature, humidity, wind and radiation into one felt temperature. It is defined as the air temperature of a reference environment (calm, shaded, 50% humidity) that would produce the same physiological response, and it is expressed in °C. It came out of an international research action (COST 730) and is widely used in urban climate work.
The stress categories are defined names and we use them verbatim:
| UTCI | Category |
|---|---|
| 48 – 79 °F (9 – 26 °C) | no thermal stress |
| 79 – 90 °F (26 – 32 °C) | moderate heat stress |
| 90 – 100 °F (32 – 38 °C) | strong heat stress |
| 100 – 115 °F (38 – 46 °C) | very strong heat stress |
| above 115 °F (46 °C) | extreme heat stress |
The cold side (slight, moderate, strong, very strong and extreme cold stress below 48 °F) is defined the same way and appears in winter runs. The map is labeled in °F; the index itself is defined in °C.
Ingredients — three of four were already there
| Input | From |
|---|---|
| Air temperature, humidity, solar radiation | The EPW weather file the wind rose uses. It carries all 8,760 hours of dry-bulb, humidity, irradiance and cloud cover for the nearest TMYx station; we fold them by month and hour. |
| Wind speed | The Wind sheet's simulation. If the same site and season has already been computed, it is reused. |
| Shade and sky view | Shadows cast on the building height map the wind solver uses and on the terrain, plus the sky-view factor at each point. The tree mask is added on top where it exists. |
| Ground cover | Green-space polygons ∪ detected vegetation. Grass runs cooler than asphalt (albedo 0.20 vs 0.12). |
From these, mean radiant temperature is computed at every point and combined with air temperature, humidity and wind into UTCI. The UTCI formula is a sixth-order polynomial with 210 terms; transcribing it by hand is a guaranteed error, so it was extracted by script from a validated open implementation, and nine published verification cases are run in the code every time (tolerance 0.1 °C).
You choose 10:00, 14:00 or 18:00 and spring, summer, autumn, winter or annual. With a wind result in hand the map takes a few seconds; without one, the wind simulation has to run first, 30 s to 3 min depending on radius.
Tree shade — the 26 ft assumption
The first version painted parks the same color as the streets. A wooded hillside site in Seoul came out 69% green space and 4.3% shade: no buildings, so no shadows, and trees cast none. This is not a small error. The ground-cover effect of grass is worth about 3 °F of UTCI; canopy shade is worth 14–27 °F. The coolness of a park is mostly its shade.
So detected trees are given an assumed height of 26 ft and cast shadows like buildings. Tree shade is not total: 15% of direct sun is allowed through (transmittance 0.15), and sky view under a canopy is capped at 0.30.
| Summer, 14:00 | Shade before trees | Shade after | Of which trees |
|---|---|---|---|
| Dense downtown, 1,000 ft | 27.1% | 33.8% | 8.6% |
| Wooded hillside | 4.3% | 57.8% | 53.9% |
Twenty-six feet (8 m) is an assumption. Imagery tells us where trees are, not how tall. If the real trees are larger the place is cooler than the map; if they are young, hotter. The on-screen warning says so permanently.
One technical note: the moment canopy is added to the obstacle heights, a person standing under a tree acquires an eye height of 26 ft and reads as "sunlit". The observer height had to be passed separately at ground level. Without that step, adding trees changes nothing.
Known biases
The first two both push the map warmer than reality (less cold in winter).
- Pedestrian-height wind comes out too weak. The wind simulation gives weaker wind near the ground than reality. On the Seoul and Tokyo sample sites, 81–89% of cells sit at the lowest wind the UTCI formula accepts (1.1 mph), so the cooling effect of wind is almost absent. Open, windy places can read warmer than reality (less cold in winter) → warm bias.
- Three trees in ten are missed (Korea; about one in seven in New York, Los Angeles and Boston, one in nine in San Francisco). Missed trees cast no shade → warm bias.
- Terrain is coarse. Wind, shade and sky view include terrain, but elevation (USGS 3DEP) is sampled every 15–30 m depending on the radius, so retaining walls, steps and small mounds are missing. Slope direction does not change ground temperature.
- Surface properties are textbook values. Albedo and emissivity of asphalt, concrete and grass were not measured. This is also why the tool does not produce a separate surface-temperature map: a surface energy balance needs thermal inertia, soil moisture and evapotranspiration, which we do not have.
- Buildings with no height are missing, exactly as in the wind model.
How to read it
As the on-screen warning says — do not quote the numbers; compare places. "UTCI 95.2 °F at this point" is not a value this tool can vouch for. "This alley is in shade and reads green, that plaza is in sun and reads red", "the color changes where the street-tree row breaks" — those are judgements about form, and the map supports them well.
Changing hour and season is the point. The same downtown site is red at 14:00 in summer (sun altitude 66°, 27% shade) and green at 18:00 (altitude 20°, 83% shade). If you are designing a plaza, overlay this with when people will actually be there.
What this is not
Not a thermal environment assessment. A real one runs a validated model over many days and hours, reports statistics, and puts trees in from a survey with species and heights. This is the step before that — a tool for reading the structure of outdoor heat at concept stage. That is why the output image always carries "Tree shade is included, but canopy height is unknown and assumed 26 ft" and "Compare places, do not quote the numbers".