How the Sunshine Score works
One number, 0–100, for how good a destination's sunshine is in a given month. I compute it the same way for every place on the atlas, from long-term climate normals — not forecasts.
The formula
For each destination and month I combine three ingredients:
- Warmth (about half the score) — the month's average daytime high. Full marks from 20 °C to 32 °C; nothing below 8 °C; fading above 32 °C until 42 °C scores zero. Pleasant sunshine, not just any sunshine.
- Dryness — the month's precipitation, scaled so 0 mm is perfect and 150 mm or more scores nothing. Wet season is cloud season, so this carries the seasonal sunshine signal.
- Sunniness — the destination's annual sunshine hours, scaled against 3,500 h/yr (about the sunniest places on Earth). Every destination has a figure; a missing one would count as a neutral 2,000 h/yr.
Score = 100 × warmth × (0.5 + 0.5 × (0.55 × dryness + 0.45 × sunniness)). Warmth multiplies the whole score, so a freezing month scores zero no matter how clear the sky.
The score's sunniness is the year's. Each month's own sunshine hours are in every destination's table since November 2026, and they decide which month a page calls the sunniest; the month with the best score is called the best month. The two often differ: a desert's sunniest month is usually too hot to score well.
What's on the atlas
Every destination on Sunshine Atlas has its own airport — that is the selection rule. The atlas maps the world's 3,833 metro-primary airports (one per metropolitan area: London appears once, not five times), so every dot on the map and every ranking entry is a place you can actually fly to. No viewpoints, no unreachable idylls — real, bookable destinations.
Destination types
Every destination has one type, decided by four facts in this order: how many people live there (the population the atlas holds for the place: for most, the town or city itself; for many big cities, their urban area), whether it stands on the sea, its height, and how warm its sea is. "On the sea" means the place stands within 15 km of the coastline on Natural Earth's 1:10m outlines and the sea there is measured (a cell of NOAA's sea surface temperature analysis within 40 km). That leaves out river ports far up an estuary, such as Hamburg and Montréal, and, for want of a measurement, a few fjord towns such as Trondheim and Ushuaia. The three cities whose own limits reach the sea although their centre stands further from it are counted on the sea by name: Los Angeles, Göteborg and Brisbane.
| Type | What it means | Destinations | For example |
|---|---|---|---|
| Beach resort | Under 400,000 people (or none on record), on a sea that is 21 °C or warmer for at least four months of the year. | 832 | Phuket, Punta Cana, Alicante |
| Seaside town | Under 400,000 people (or none on record), on a sea that is 21 °C or warmer for fewer than four months. | 461 | Reykjavík, Bergen, Stavanger |
| Mountain | Under 2 million people, at 900 m or more and at least 30° from the equator, in a climate that is not dry; and six alpine resorts that rule would miss, named one by one: Innsbruck, Bolzano, Grenoble, Chambéry and Queenstown in their valleys, and Montrose below Telluride. | 85 | Calgary, Salt Lake City, Innsbruck |
| High altitude | Under 2 million people, at 1,500 m or more within 30° of the equator. | 97 | Lijiang, Querétaro, Guanajuato |
| Major city | 2 million people or more, away from the sea. | 165 | London, Shanghai, Beijing |
| Major coastal city | 2 million people or more, on the sea. | 89 | New York, Istanbul, Tokyo |
| Coastal city | 400,000 to 2 million people, on the sea. | 212 | Taoyuan, Doha, Dublin |
| Large city | 400,000 to 2 million people, away from the sea. | 501 | Amsterdam, Orlando, Las Vegas |
| Regional city | 100,000 to 400,000 people, away from the sea. | 508 | Bergamo, Chiang Mai, Eindhoven |
| Small town | Under 100,000 people, or none on record, away from the sea. | 883 | Luxembourg, Mineralnye Vody, Billund |
Climates
Every destination's climate is its Köppen-Geiger class, worked out from the atlas's own monthly averages by the rules of Peel, Finlayson and McMahon (2007), and grouped into twelve on the map. A class is decided by the average temperatures of the coldest and the warmest month, how much rain falls in a year, and in which half of the year it falls. Each destination's page names the class itself ("hot desert", "humid subtropical"); the map names the group. Against the WMO 1991–2020 station normals at the 1,369 destinations with a station nearby, the class worked out here is the station's own at 84.4% of them, and the same main group (A to E) at 93.9%. No destination falls in the Polar group at present.
| Climate | What it means | Köppen classes | Destinations | For example |
|---|---|---|---|---|
| Tropical | Warm all year (the coldest month 18 °C or more on average) and wet in most or all months: rainforest and monsoon climates. | Af, Am | 588 | Singapore, Mumbai, Manila |
| Savanna | Warm all year, with a marked dry season and a wet one. | Aw | 507 | Bangkok, Miami, Ho Chi Minh City |
| Desert | Less than half the rain a place this warm would need to stay green. | BWh, BWk | 314 | Dubai, Las Vegas, Doha |
| Semi-arid | Dry, but with more rain than a desert: steppe and dry grassland. | BSh, BSk | 349 | Beijing, Los Angeles, Denver |
| Mediterranean | Mild winters that bring the rain, and dry summers. | Csa, Csb, Csc | 233 | Istanbul, San Francisco, Barcelona |
| Subtropical | Hot, humid summers (the warmest month 22 °C or more) and mild winters, with no dry summer. | Cfa, Cwa | 512 | New York, Shanghai, Tokyo |
| Oceanic | Mild in every season with rain in every month; no month averages 22 °C. | Cfb, Cfc | 268 | London, Paris, Amsterdam |
| Highland | Cool for its latitude because it stands high: the subtropical highland climates, and any temperate climate without a hot summer, any continental climate and tundra at 1,500 m or more within 35° of the equator. | Cwb, Cwc and others | 71 | Mexico City, Bogota, Johannesburg |
| Continental | Warm or hot summers and winters that freeze (the coldest month 0 °C or below on average). | Dfa, Dfb, Dwa, Dwb, Dsa, Dsb | 544 | Chicago, Moscow, Toronto |
| Subarctic | Long freezing winters and short cool summers: fewer than four months average over 10 °C. | Dfc, Dfd, Dwc, Dwd, Dsc, Dsd | 373 | Krasnoyarsk, Irkutsk, Tromsø |
| Tundra | No month averages over 10 °C, but at least one is above freezing. | ET | 74 | Punta Arenas, Sabetta, Nuuk |
What the score is not
It isn't a weather forecast, and it's deliberately opinionated — my opinion, encoded: a 41 °C August scores low because standing in it is no fun, and a gloriously clear −5 °C ski morning scores zero because this is a sunshine atlas, not a snow one. If you'd weigh things differently, the raw data is free — score it your way.
Data sources
- Day highs and night lows: NASA POWER climatology, January 2001 – December 2020 (MERRA-2 reanalysis, a cell of about 50 km; CC BY 4.0), corrected to what thermometers at the airports measure. A 50 km cell mixes sea, forest and town into a narrower day than an airport's open ground has, and it averages the mountains around a valley. Since November 2026 the correction reads five things about each place and month: the cell's own day-night range over the year, the month's departure from it, the latitude, the season, and how far the cell stands above or below the destination (fitted at 5.6 °C a kilometre for the highs and 5.3 for the lows, where the textbook 6.5 was assumed before). It was fitted on the WMO 1991–2020 climate normals at 1,435 stations and judged by cross-validation by station. The month's mean is the midpoint of its corrected high and low, as the stations define it. Against the normals at the 1,451 stations within 15 km and 150 m of an atlas airport, the highs are now off by 0.97 °C on average and the lows by 1.12 °C, with no bias left in any band of latitude (the raw cells were off by 1.45 and 1.48). What a 50 km cell still cannot see is a valley or a coast under an inversion: Santiago stands 680 m below its cell and reads about 3.5 °C warm, and La Serena, Iquique and Ilo likewise.
- Sunshine hours, for every month: modelled from NASA POWER's clearness index over the same period, through the Ångström–Prescott relation (n/N = (Kt − 0.17) / 0.66, the share of the day's possible sunshine), times each month's true astronomical daylength. Where the midday sun stands low (under 37° in full, fading out by 64°) the share is also held to what the direct beam allows, read from two more NASA POWER fields (the direct beam under the real sky and under a cloudless one). A sunshine recorder counts only the direct beam; the clearness index counts scattered light too, which is plentiful over snow and under Arctic stratus, and on its own it gave the far north more sunshine than Germany. The coefficients were fitted on the WMO normals at 1,702 stations with every band of latitude weighing the same. The year's figure is the twelve months added up.
- Rainfall (mm, monthly and annual): a blend of CHELSA V2.1 (Karger et al., WSL; a 1981–2010 climatology at about 1 km; CC0) and the NASA POWER precipitation normals, 0.94 × (0.8 × CHELSA + 0.2 × POWER), since November 2026. Against the WMO 1991–2020 normals at 1,450 stations the blend's monthly error is 12.2 mm against 19.7 mm for POWER alone (117 against 201 mm a year, with no bias): CHELSA resolves the rain shadows and mountain slopes a 50 km cell cannot. The weight and the scale were picked by cross-validation by station.
- Sea temperatures (coastal destinations): NOAA's Optimum Interpolation Sea Surface Temperature analysis, version 2.1, its long-term monthly means for 1991–2020 (a quarter of a degree; public domain). A thirty-year normal like every other figure here; until November 2026 they were a single year (2024) from Open-Meteo's marine service, which stood about a degree warmer. A destination is coastal when the place it is drawn on stands within 15 km of the sea and the analysis has a sea cell within 40 km (see the destination types).
- Climate class: the Köppen-Geiger class of each destination, worked out from the monthly averages above (see the climates).
- Midday UV index: NASA POWER's all-sky surface UV index over the same 2001–2020 period. POWER publishes it as a 24-hour mean (nights included), so it is rescaled to the familiar midday value using the UVI ∝ cos(SZA)2.42 relationship (Allaart et al. 2004) — the noon-to-whole-day ratio is pure solar geometry per latitude and month — then lifted 7% to match the daily-maximum UV index that forecasts report, and adjusted by the WHO's ~10% per 1,000 m altitude rule over the same cell-vs-destination elevation gap. Checked against NASA POWER's own hourly UV series (clouds included) at ten climatically diverse sites × 12 months: mean absolute error 0.27 UV points, no bias, and 92% agreement on the WHO exposure category. Published on the WHO scale: 0–2 low, 3–5 moderate, 6–7 high, 8–10 very high, 11+ extreme.
- Snow per year (≈): annual snowfall depth from NASA POWER's monthly snowfall normals (MERRA-2 PRECSNO, water equivalent) over the same January 2001 – December 2020 window, converted with the standard 10:1 fresh-snow-to-liquid ratio. Months only count in proportion to how cold the destination itself is (full weight when the mean night low is −2 °C or below, nothing above +5 °C; within the tropics a month must actually freeze), and when the ~50 km grid cell sits well above the destination its snowfall is attenuated toward the town — otherwise the cell smears mountain snow onto warm valley cities. Treat it as an honest estimate of typical snowfall, not a station record: valley towns beside big ranges can still read up to about twice their true figure.
- Places, populations and coordinates: © GeoNames and OurAirports contributors, with 240 populations corrected from Wikidata (CC0) in November 2026, most of them small places that had been given the figure of a city nearby. Elevations from the Copernicus DEM via Open-Meteo.
- The period. NASA's service now labels this climatology 2001–2025. Asked for 2001–2020 or for 2001–2025 it returns the same numbers, and they equal the ones the atlas read in July 2026 (all 960 values of a tile compared in November 2026), so the figures here are unchanged and keep the period they were published under, 2001–2020.
- Photography via Wikimedia Commons.
How accurate are the sunshine hours?
I check them against two references: the WMO 1991–2020 climate normals, which give recorder-measured sunshine at 494 stations within 15 km and 150 m of an atlas airport (none in the United States, Canada or India, and leaving out Indonesia's, which record an eight-hour window rather than a whole day), and 362 cities where a national met service publishes a figure. Against the stations the year's figure is off by 160 hours on average, with a bias of +4; a month's by 17 hours, and the sunniest and the least sunny month a station measured are named to within a month at 90% of them. Against the published figures outside the United States: correlation 0.93, a typical city within 5.5% of its figure, and a bias of −29 hours.
The errors are not evenly spread, and it's worth knowing where they bite. A sunshine recorder only counts direct beam above about 120 W/m² and deliberately ignores scattered light; a satellite sees the scattered light too. So wherever a persistent low cloud deck sits under a high sun, the model reads high: Lima, whose famous garúa stratus it misses almost entirely, comes out at 2,159 hours a year against a published 1,230. It reads low in the Gulf, where Dubai is 14% under. This is a well-documented limitation of satellite-derived sunshine rather than anything peculiar to this atlas (Kothe et al. 2017, Remote Sensing 9(5), 429).
The far north reads true since November 2026. Until then the model gave the Arctic more sunshine than the mid-latitudes (Tromsø 1,805 hours against a published 1,265, Iqaluit 1,985 against 1,477), because snow and Arctic stratus fill the sky with scattered light that a recorder ignores, and 35 destinations beyond 70° had no figure at all. Held to the direct beam under a low sun, the stations beyond 60° are now off by 87 hours on average (184 before, almost all of it too high): Tromsø reads 1,110 hours against a published 1,265, Iqaluit 1,426 against 1,477 and Reykjavík 1,296 against 1,326.
The United States is the largest disagreement in the whole check, and it is one-sided. Against the 52 US cities that publish a recorder-measured figure the model runs 343 hours low, and every one of them comes out below its published number; outside the US the same model runs 29 hours low on average. The miss is about the same in the dry West and the humid East (Phoenix −431, Cleveland −438, Boston −480, Miami −611). That shape points at the reference data rather than at the model, because the same climates outside the US do not show it: cloud physics does not stop at a border. The published US sunshine figures descend largely from NOAA's historical sunshine network, which was already down to roughly 160 stations by the early 1980s and is no longer maintained as a current product, so this is very likely a 2001–2020 satellite climatology being scored against a mostly pre-1980 recorder climatology, on instruments known to over-read. The WMO stations the coefficients are fitted to include none in the US, so the US figures are the least checked here. I have not resolved it. Until I do, treat US sunshine hours as carrying roughly a 350-hour uncertainty on top of everything else here, and prefer the station figure where one exists.
How accurate are the temperatures and the climates?
The day highs are off by 0.97 °C and the night lows by 1.12 °C on average against the WMO 1991–2020 normals at about 1,450 stations near an atlas airport, and the climate class worked out from the atlas's averages is the station's own at 84.4% of the 1,369 destinations with a station measuring both temperature and rain. Two places where the check cannot reach are worth naming. The equatorial highlands with no station in the WMO file read warm in their cool season: Nairobi's July averages about 20 °C here against 17.7 °C at its airport, which is enough to make it a savanna rather than the highland climate its airport measures. And valleys and coasts under an inversion read warm, as the sources above say. Two faults in the reference itself are left out of every check: the normals file holds Guatemala City's temperatures under Puerto Barrios and the reverse, and Indonesia's sunshine stations record an eight-hour window.
Gridded normals also can't capture microclimates, and a single cell can average a coastline together with a mountain. Treat the scores as a comparison tool and the tables as long-term averages rather than guarantees.
Want the numbers themselves? Download the full dataset — every destination and month as CSV/JSON, free under CC BY 4.0.
Questions and answers
How is the Sunshine Score calculated?
The Sunshine Score is a 0–100 rating combining three things from long-term climate normals: daytime warmth (full marks around 20–32°C), dryness (0 mm of rain is perfect) and annual sunshine hours. Score = 100 × warmth × (0.5 + 0.5 × (0.55 × dryness + 0.45 × sunniness)). Warmth multiplies the whole score, so a freezing month scores zero however clear the sky. It is a climate average, not a forecast.
How accurate are Sunshine Atlas sunshine hours?
Checked against two references, the WMO 1991–2020 climate normals at 494 stations near an atlas airport and 362 cities where a national met service publishes a recorder-measured figure, the modelled annual sunshine hours are off by 160 hours a year on average against the stations, and outside the United States a typical city lands within 5.5% of its published figure (correlation 0.93). They are modelled from NASA POWER satellite clearness (2001–2020), held to the direct beam where the sun stands low, with coefficients fitted to those measurements. Where a persistent low cloud deck sits under a high sun the model reads high (Lima). The United States is the largest one-sided disagreement: all 52 US cities with a published figure read low, by 343 hours on average, most likely because the US reference data is largely a pre-1980 recorder climatology; treat US sunshine hours as carrying roughly a 350-hour uncertainty.
What data does Sunshine Atlas use?
Temperature, sunshine and midday UV index come from NASA POWER climate normals (2001–2020, CC BY 4.0), with the day highs, night lows and sunshine hours corrected to measured station normals; rainfall blends CHELSA V2.1 (CC0) with NASA POWER; sea temperatures are NOAA OISST v2.1 normals for 1991–2020 (public domain); places, populations and coordinates come from GeoNames and OurAirports, with populations corrected from Wikidata. The full dataset for every destination is free to download under CC BY 4.0 at sunshineatlas.com/data/.