Methodology

Methodology and data sources

We only publish numbers from primary scientific sources — national agencies, the Copernicus programme and peer-reviewed assessments — and we link each one to where it came from.

Principles

• Primary sources only. Data comes straight from the agency that measures it, not from secondary websites.
• Automatic updates. Daily datasets are fetched every day and monthly ones as soon as they are released. Each page shows the date of the latest observation.
• Interpretation from the scientific literature. “What it means” sections summarise the IPCC Sixth Assessment Report, WMO and peer-reviewed updates, with section references.
• Open data. Every chart has a CSV download. Our processed data and charts are free to reuse under CC BY 4.0; the original licence of each source also applies.

Baselines

“Pre-industrial” means 1850–1900, as used by the IPCC. Copernicus ERA5 data are published relative to 1991–2020; we add Copernicus' standard offset of 0.88 °C. This offset is defined for annual means; for single days it is accurate to about ±0.1 °C. NASA GISTEMP starts in 1880, so we express it relative to 1880–1900.

Sources, licences and latest data

City and country pages

City figures come from global 0.25° (~25 km) grids published by the World Bank Climate Change Knowledge Portal: the Copernicus ERA5 reanalysis for the past (1950–2025) and the bias-corrected, downscaled CMIP6 multi-model ensemble (NASA NEX-GDDP-CMIP6) for the future. We take the grid cell containing each city (or the nearest valid cell on coasts).

“2050” means the 2040–2059 average. Projections use the delta method: the observed ERA5 climate of 1995–2014 plus the models' projected change between 1995–2014 and 2040–2059. We show the ensemble median and the 10th–90th percentile range, for an intermediate (SSP2-4.5) and a very high (SSP5-8.5) emissions scenario.

The “analog city” is the large city (population over one million, at least 500 km away) whose 1991–2020 climate is closest to a city's projected 2050 climate, measured across average, daily-high, daily-low and hottest-day temperatures, rainfall, hot days, tropical nights and frost days. Among near-equal matches we pick the largest city. When no city is close we say so. The idea follows Bastin et al. (2019), PLOS ONE.

Country temperatures are area-weighted national averages from the same sources; emissions are territorial CO₂ from fossil fuels and industry (Global Carbon Budget via Our World in Data). City list: GeoNames.

Sea level by city

Local sea level projections come from the IPCC AR6 sea level dataset (Garner et al. 2021, Zenodo, CC BY 4.0), the same data as the NASA Sea Level Projection Tool. We use the nearest tide gauge within 60 km of the city; for a few coastal cities without one we use the nearest grid point, which does not include local subsidence. Values are relative sea level change (ocean rise plus vertical land motion) versus 1995–2014, medium confidence: the median and the likely (17th–83rd percentile) range.

Your lifetime

The birth-year page compares five-year averages of global temperature (Copernicus ERA5 from 1940, NASA GISTEMP before) around the chosen year with the latest five years, and uses NOAA CO₂, NSIDC September sea ice and NOAA satellite sea level where records exist.

Extreme heat map

The extreme heat map counts days a year with a heat index of 39 °C or more (NOAA/NWS formula, the start of the NWS "Danger" category) and shows where the mean annual temperature reaches 29 °C. Both use the same delta method as the city pages — ERA5 1995–2014 plus the CMIP6 ensemble median change for 2040–2059 and 2080–2099 from the World Bank CCKP 0.25° grids — and are drawn in the Equal Earth projection over Natural Earth 110m land. Land shares are area-weighted and exclude Antarctica. Built by pipeline/fetch_heat.py.

Known limitations

• The carbon budget is a straight-line extrapolation at constant emissions; real emissions vary year to year and the budget itself has wide uncertainty.
• Satellite sea level data are published with a delay of several months.
• The latest days of ERA5 temperature data are preliminary and may change slightly.
• A 25 km grid cell smooths out local effects such as urban heat islands, coasts and mountains; city values can differ from a local weather station.

Corrections

Found an error? See About & corrections. We log every correction there.