Methods
Last updated 27 October 2026
Every figure on this site comes from one of four sources. This page says which, how often it is refreshed, and what is known about its accuracy.
Sources
| Source | Used for | Refreshed |
|---|---|---|
| ARPANSA | Measured UV at 13 Australian cities | every 5 minutes |
| US EPA | Forecast UV for 50 US cities | every hour |
| Copernicus (CAMS) | Forecast curves, tomorrow, sun protection times | twice a day |
| Natural Earth | Map outlines | Fixed |
Measured and forecast
A measured reading comes from an instrument. Australian city pages show ARPANSA measurements, taken at the station named on each page and republished unaltered.
Everything else is a forecast: today's curve, tomorrow's peak, the sun protection times, and every US figure. The EPA publishes no measured feed, so US pages are forecast-only and say so.
A measured reading more than 20 minutes old is never presented as the level right now. When a station goes quiet the page says the live reading is delayed rather than showing a stale number as current.
How accurate the forecast is
We compared the forecast against what ARPANSA actually measured, across all 13 Australian stations in October 2025 and January 2025 — 3,516 paired daylight hours.
| Measure | Result |
|---|---|
| Typical error | 1.2 UV index points |
| Within 1.0 of measured | 57% of hours |
| Average bias | +0.39 (slightly high) |
Accuracy is best in southern cities and worst in the tropics during the build-up to the wet season, where afternoon storm cloud is hard for any model to place. Almost all the large errors are cloud being mis-forecast, not UV being mis-calculated.
The forecast model reads about 7% high against these instruments. We publish it uncorrected: the offset varies from station to station by more than its own size, so a single correction factor would make several cities worse rather than better.
Sun protection times
These are the hours the UV index is forecast to reach 3 or above, and they are calculated from the clear-sky forecast — the level if no cloud were present.
That is deliberate. Cloud can break at any time, and the dangerous error is the one that tells you to protect yourself too late. The times are therefore conservative, and the page says so wherever they appear. Where the forecast does not cover a city's whole day, no times are shown at all rather than a boundary being presented as a crossing.
This time of year
Each city page says how high UV usually gets there in the current month: the typical daily peak, and how often the day reached 3 or more (when sun protection is needed), 8 or more (Very High) and 11 or more (Extreme). The two countries' figures come from different kinds of record, and each page says which.
Australia uses ARPANSA's own measurements, published at one-minute resolution on data.gov.au, over the last ten complete years. A day's peak is its highest five-minute average, which follows the five-minutely readings shown live and is not thrown by a single minute's spike at the edge of a cloud. A day counts only if the instrument covered the middle of it, from 10 am to 2 pm.
The United States has no long, freely licensed record of measured UV for its cities, so the figures come from NOAA's daily UV Index bulletin: the noon forecast for about 58 cities, issued the day before and archived by NOAA's Climate Prediction Center since 1994. They are forecasts, with cloud taken into account, not measurements. A city within 60 km of a bulletin city uses that city's forecasts and says so; a city further than that from any has no figures, rather than another city's.
A month is published only with at least 60 valid days behind it — about two years' worth. The counts use the same rounding as the bands below.
Bands
Readings are shown exactly as published, to one decimal place. The band — Low, Moderate, High, Very High, Extreme — comes from the reading rounded to the nearest whole number, so 2.4 is Low and 2.5 is Moderate. The bands are the World Health Organization's.
Times and time zones
Every time on the site is local to the city it describes. Queensland, the Northern Territory and Western Australia do not observe daylight saving while New South Wales, Victoria, South Australia, Tasmania and the ACT do, so a single Australian clock would be wrong for roughly half the country for half the year.
Credits and licences
- UV observations courtesy of ARPANSA. © Commonwealth of Australia as represented by the Australian Radiation Protection and Nuclear Safety Agency. ARPANSA disclaimer.
- US forecasts: United States Environmental Protection Agency.
- "This time of year" in Australia: calculated by uvatlas from ARPANSA's historical UV measurements on data.gov.au, licensed under Creative Commons Attribution 2.5 Australia.
- "This time of year" in the United States: the NOAA/EPA UV Index bulletins archived by NOAA's Climate Prediction Center.
- Forecast curves and sun protection times contain modified Copernicus Atmosphere Monitoring Service information. Neither the European Commission nor ECMWF is responsible for any use of this information.
- Made with Natural Earth.
Nothing on this site is medical advice. See the terms.