Climate Data Methodology

Which Baseline Is In Use

Most temperature anomalies on this site are quoted against the 1961–1990 average, the WMO standard normal used for cross-region comparison. Where a source publishes against a different baseline, the source-native figure is also shown for verification.

• NOAA US states, climate regions, and continents are reported by NOAA against 1901–2000; we re-baseline them to 1961–1990 for the rankings table and roll-ups, and surface the NOAA-native value as a secondary verification figure.
• Met Office HadUK regional series use 1991–2020; the UK pages quote both that and the 1961–1990 figure.
• Paris Agreement 1.5 °C / 2.0 °C tracker is quoted against 1850–1900 pre-industrial, per IPCC convention; this is a global-only metric and is not applied to regions.
• NSIDC sea ice uses NSIDC's 1991–2020 climatology for percent-of-normal extent.
• North America and South America continental land anomalies are 4BYO aggregates built from the country snapshots, because NOAA does not publish a standalone land series for these continents. They are labelled “agg” in the roll-ups.

The Two-Baseline Model

Climate normals are arbitrary. Different agencies use different 30-year reference periods, and switching baseline only shifts every value by a constant — the trend is identical. To make the site both internally consistent and easy to verify against upstream sources, we publish two figures wherever they differ:

1. A comparison baseline of 1961–1990 on maps, rankings, and roll-ups. This is the WMO standard normal and the same baseline OWID, the Met Office Hadley Centre, and most academic literature use, so numbers are directly comparable across regions.
2. The source-native baseline shown alongside for verification. NOAA US states, climate regions, and continents use 1901–2000; the Met Office UK pages use 1991–2020; sea-ice extent uses NSIDC's 1991–2020 climatology; the Paris 1.5 °C / 2 °C tracker uses 1850–1900 pre-industrial.

Re-baselining is a deterministic linear shift: we take each calendar month's mean over 1961–1990, subtract it from the source-native anomaly, and the result is the same observation re-expressed against the comparison period. No smoothing, interpolation, or homogenisation is added on our side.

Data Source Timeline (Record Start)

Each bar shows the year the upstream record begins through to the latest available month. Bars are colour-coded by data family. The vertical guideline marks the start of the 1961–1990 comparison window.

When Monthly Data Lands

Each bar shows the calendar day of the month the upstream source publishes the previous month's data, and when 4BYO runs its snapshot rebuild. The snapshot is timed to fall after the slowest source has refreshed so every page is current as of the same reference month.

When Annual Data Lands

Datasets that publish only once a year — country precipitation, electricity mix, CO₂ emissions, IEA energy outlook and the WMO/NOAA annual State-of-the-Climate reports. Bars show the typical month of release; data covers the previous calendar year.

Source Inventory

How Rankings Are Computed

• For each region we read the latest complete month from the per-region snapshot and compute the 1-month, 3-month rolling, and 12-month rolling anomaly against 1961–1990.
• Rows are sorted globally; the rank shown in the table is the position across all regions of the selected sort window, not within the active filter.
• Roll-ups by group use NOAA-authoritative continent and 9-region US series where they exist. North and South America are 4BYO aggregates from country snapshots because NOAA does not publish standalone land series for them — these rows are flagged “agg”.
• Movers (climbers / fallers) compare the current snapshot against the previous month's archived snapshot in rankings-previous.json.

Cross-Source Caveats

• Country and US-state series come from different upstream pipelines (OWID/HadCRUT5 versus NOAA statewide). Their absolute anomalies are not directly comparable to the kelvin, but trends are.
• UK regions use Met Office HadUK-Grid (1991–2020 native), so the league table re-baselines to 1961–1990 for cross-region comparison; UK detail pages still quote both figures.
• The Paris Agreement 1.5 °C / 2 °C tracker is intentionally global-only. Pre-industrial (1850–1900) regional baselines are not robust because most regions lack reliable instrumental coverage that far back.
• Subnational series outside the US (Canadian provinces, Australian states, etc.) are not published by NOAA. They are deferred to a Phase 2 build that will pull from each national meteorological service (ECCC, BoM, DWD, JMA, …).

Further Reading

• Climate rankings & league table — every region we track, sortable by 1m / 3m / 12m anomaly.
• Global climate update — Paris tracker, continental bars, ENSO state, GHG, sea ice.
• Climate updates — country, US state and UK region detail pages.
• Climate, explained — plain-English guide to the science behind the data.