Data sources & licences

This site is built entirely on open data and open-source software. Thank you to everyone who collected, shared and maintains it. Here is exactly what we use and the licence each part comes with.

Data version

Beck et al. (2023) Köppen-Geiger maps, Figshare version 2 (doi:10.6084/m9.figshare.21789074.v2); GeoNames cities15000 (34,157 classified cities); Natural Earth 10m countries.

Last refreshed

2026-10-12

Software

The map is drawn with MapLibre GL JS and PMTiles, and the site is built with Astro. These are open-source projects under their own licences.

Using our images and embeds

Images you download from the map and the embed code both carry the credits above. Please leave them in place.

How to cite the climate maps

Beck, H. E., McVicar, T. R., Vergopolan, N., Berg, A., Lutsko, N. J., Dufour, A., Zeng, Z., Jiang, X., van Dijk, A. I. J. M. and Miralles, D. G. (2023). High-resolution (1 km) Köppen-Geiger maps for 1901–2099 based on constrained CMIP6 projections. Scientific Data 10, 724. doi:10.1038/s41597-023-02549-6

Article: doi:10.1038/s41597-023-02549-6. Data: doi:10.6084/m9.figshare.21789074.v2 (file koppen_geiger_tif.zip, SHA-256 bb84453d4541f1a0…). Licence: CC BY 4.0. We changed the data by drawing each map as map tiles and by working out areas, shares and changes; the original files are not altered.

The five maps and the scenarios

• 1901 to 1930. Past (reconstructed).
• 1991 to 2020. Recent observations (the current map).
• 2071 to 2099, SSP1-2.6 (low emissions). Model projection.
• 2071 to 2099, SSP2-4.5 (middle of the road). Model projection.
• 2071 to 2099, SSP5-8.5 (high emissions). Model projection.

The future maps are model projections, not forecasts. The three scenarios are shared socioeconomic pathways: SSP1-2.6 (low emissions), SSP2-4.5 (middle of the road, our default future) and SSP5-8.5 (high emissions). They describe different possible amounts of future greenhouse gas emissions. Which one the world follows is not known. We name the period and scenario wherever a future figure appears.

How we use the data

The dataset has a grid cell about 1 km across. We draw each of the five maps as paletted map tiles (up to zoom level 8). Because every class has its own exact colour, the class under a tap is read back from the tile pixel under it, so the card and the map always agree. A tap near a class boundary can differ from the cell next door, because tile pixels are finer than the 1 km cells. Cities that sit on an unclassified coastal cell take the nearest classified cell within 6 cells.

Areas and shares of land use spherical-earth areas of the 1 km cells. "Excluding Antarctica" removes the Natural Earth Antarctica outline. Country figures use Natural Earth outlines, with all its choices for disputed areas. Spherical-earth km2 of classified 1 km cells. Percent of land = share of all classified land cells in the map; the 'excl_antarctica' figures remove cells inside the Natural Earth Antarctica polygon.

Accessibility of the colours

The map uses the standard Köppen-Geiger palette from the dataset, because it is the one people recognise. That palette is not colour-blind safe: some neighbouring classes (for example some reds and greens, and some blues) are hard to tell apart. For that reason the key, the tap card and the pages always give the class code and name in text. We plan to add a colour-blind-friendly palette option.

Köppen and the classes

Köppen-Geiger is a classification of long-term average monthly temperature and rainfall, not of weather. This dataset uses the class rules in Table 1 of the paper, adapted from Peel et al. (2007), with 0 °C as the boundary between temperate (C) and cold (D) climates.