WeatherWhether monthly climatology, 2001–2025
The derived dataset contains twelve calendar-month averages for four climate variables: mean temperature, total precipitation, wet-day frequency and mean cloud cover. WeatherWhether calculated these values from CRU TS 4.10, produced by the Climatic Research Unit at the University of East Anglia.
Coverage is global land excluding Antarctica. Oceans, some small islands and other missing cells have no values. The grid spacing is 0.5°, approximately 55 km north to south; east–west distances vary with latitude. Country outlines provide geographic context and do not determine the climate cells.
| Variable | Unit | Definition |
|---|---|---|
| Mean temperature | °C | Monthly average air temperature, including day and night. |
| Total precipitation | mm/month | Monthly precipitation total; rain and snow are not separated. |
| Wet-day frequency | days/month | Days receiving at least 0.1 mm of precipitation. |
| Mean cloud cover | % | Average fraction of the sky covered by cloud. |
How are the monthly averages calculated?
For each grid cell, WeatherWhether takes the arithmetic mean of the same calendar month across all 25 years. January is the mean of 25 Januaries, February of 25 Februaries, and so on. Each variable uses all 300 source months; no years are excluded.
Each of the 67,420 valid land cells has all 25 values for each calendar month. Source archive checksums, timestamps, grid coordinates and sample counts were checked during processing. The native grid is retained without spatial resampling or interpolation into missing cells.
Web values are rounded to 0.1 units, with a maximum rounding error of 0.05. That is an encoding precision, not a measure of the original climate estimates’ accuracy. These 25-year averages are not an official 30-year climate normal.
How is “Best time to visit” calculated?
Weather-fit is a 0–100 preference score calculated from the four monthly climate means. The model is named weather-fit-v1. It is a product heuristic, not a validated tourism index or a probability of good weather.
Each factor receives a match between 0 and 1. The factors are combined with a weighted geometric mean, so a weak factor reduces the overall result.
score = round(100 × TwT × PwP × WwW × CwC)
- Temperature (T): full credit within the chosen monthly-mean band: cool 10–20°C, mild 18–26°C or warm 24–30°C. Outside the band, the match falls linearly to zero at a distance of 15°C from the nearest band edge.
- Precipitation (P):
1 / (1 + monthly_mm × 30 / month_days / 100). Less precipitation gives a higher match. - Wet days (W):
max(0, 1 − wet_days / month_days). - Cloud cover (C):
1 − 0.8 × cloud_percentage / 100.
Month lengths use 28.24 days for February, accounting for the six leap years in 2001–2025. Other months use their usual lengths. The exponents are the weights below divided by 100.
| Preference | Temp. | Precip. | Wet days | Cloud |
|---|---|---|---|---|
| Balanced | 40 | 25 | 25 | 10 |
| Less rain | 30 | 30 | 30 | 10 |
| Clearer skies | 35 | 20 | 20 | 25 |
The default is mild temperatures with balanced weights. Precipitation and wet days are related; their combined contribution intentionally favors drier trips.
Scores of 85–100 mean great fit, 70–84 good fit, 50–69 mixed fit and 0–49 poor fit. The location profile lists up to three highest-scoring months with a score of at least 70; ties follow calendar order. Missing climate inputs produce no score. CSV exports record the model version and selected preferences.
What are the limits of the map?
- Historical averages, not a forecast. Individual years and travel dates can differ. The score does not estimate year-to-year reliability.
- Regional scale. A city search represents the surrounding grid cell. Elevation, coastlines and local microclimates may differ substantially within it.
- Missing measurements. Sunshine hours, sunny/cloudy-day counts and separate snowfall totals are not included. Cloud percentage cannot supply those measurements.
- Incomplete travel context. The score excludes humidity, wind, daily extremes, sea conditions, activities, crowds and travel safety.
- Source uncertainty. Station availability and estimation uncertainty vary. The atlas does not display the source station-count or uncertainty layers.
Place-name search uses Photon and OpenStreetMap. A search sends the entered place name to Photon when you submit it. It does not send device geolocation. Search changes the selected map cell, not the underlying climate data.
Download the monthly climate data
The twelve monthly files total about 4.08 MB compressed. Each file contains all four climate grids. See the metadata index for file sizes, SHA-256 checksums, units and provenance, and the licence and encoding notes before reuse.
After gzip decompression, each file holds four consecutive 360×720 little-endian unsigned 16-bit grids in temperature, precipitation, wet-days and cloud-cover order. Rows run north to south and columns west to east. The missing sentinel is 65535. Decode temperature as value / 10 − 100 and other variables as value / 10. The weather-fit score is calculated separately and is not a fifth downloaded grid.
For a simpler table, select a place on the map and download its monthly CSV.
Sources, attribution and reuse
Original climate data: Climatic Research Unit, University of East Anglia, CRU TS 4.10. WeatherWhether produced the calendar-month aggregation and web encoding.
Scientific reference: Harris I, Osborn TJ, Jones P and Lister D (2020). Version 4 of the CRU TS Monthly High-Resolution Gridded Multivariate Climate Dataset. Scientific Data 7, 109. doi:10.1038/s41597-020-0453-3.
Derived database licence: Open Database License 1.0, with the Database Contents License. Preserve attribution and applicable share-alike obligations. These data licences do not assign a licence to the application code.
Geographic context: country outlines from Natural Earth (public domain); place names from Photon using OpenStreetMap data, © OpenStreetMap contributors. Country boundaries are context, not a legal determination.