Three thermometers, one number
WBGT was developed for the US military in the 1950s after heat casualties in recruit training, and it has survived because it captures something the heat index cannot: the difference between standing in shade and standing in sun.
| Thermometer | Responds to | Outdoors | Indoors / shade |
|---|---|---|---|
| Natural wet bulb | Humidity, wind, and sun falling on a wet wick left in the open | 0.7 | 0.7 |
| Black globe, 150 mm | Radiant heat: sun above, hot surfaces below | 0.2 | 0.3 |
| Dry bulb | Shaded air temperature | 0.1 | — |
The weighting is not arbitrary: humidity dominates human heat stress, so the wet bulb gets most of the weight. But the globe's 0.2 is what makes WBGT useful, because the black globe can run 30–40°C above the air temperature in strong sun.
Note the natural wet bulb. It is not the shaded, ventilated wet bulb from the wet bulb page — it sits out in the weather, so sun and ambient wind both act on it, and it reads higher.
Why standards use it
Because it answers the question a supervisor actually has. Not "how hot does it feel" but "how much can someone do here before their core temperature climbs". That maps onto:
- Occupational limits. ACGIH threshold limit values and NIOSH recommended alert limits are tabulated as WBGT against workload, which is what the work/rest calculator implements.
- Athletic flag systems. White, green, yellow, red and black thresholds on WBGT, used across US military and school sport.
- Japanese law. Since 1 June 2025 Japan requires workplace heat measures at WBGT 28°C or air temperature 31°C for work of over one continuous hour or four hours a day — the only one of these three that is a statutory number rather than guidance. The Japanese page covers it in detail.
The honest part: radiation is not reported
Here is the constraint that shapes every app and website in this space, including this one.
WBGT needs radiation. No weather service publishes solar irradiance in its standard feeds — not Apple's weather data, not the ordinary national feeds. So the globe term cannot be measured from weather data; it has to be modelled.
The model this site uses, and the one in the app, works like this:
- Clear-sky global irradiance for the sun's actual elevation, via the Haurwitz relation.
- Attenuated for cloud cover by the Kasten and Czeplak relation.
- Split into direct beam and diffuse sky by the Erbs correlation.
- Downward longwave from Brutsaert's clear-sky emissivity with a cloud correction.
- Wick and globe energy balances solved in the analytic form of Liljegren's model published by Kong and Huber, which is reported to sit within 1°C of the full iterative model in 99% of cases.
Two consequences follow, and they are stated on every page that shows a WBGT:
- Every value is an estimate. The accuracy figure above is about the mathematics, not the inputs. A modelled sky is not a measured one.
- It reads low in hot, dry, still conditions — by up to 2°C. That is documented behaviour of the analytic form, and it happens to be the desert-afternoon case. In that region the number is a floor, and the calculator says so when you are in it.
If you need a value that stands up as a record, buy the instrument. App or WBGT meter is the version of that comparison written without a sales motive.
Reading a WBGT figure
- Check whether it is indoor or outdoor weighted. They are different numbers for the same conditions.
- Check the workload it is being compared against. 28°C is unremarkable for light work and past the limit for heavy work.
- Check whether it assumes acclimatisation. Most published tables do, and unacclimatised limits are several degrees lower.
- Check whether it was measured or modelled, and if modelled, whether the conditions are hot, dry and still.