A temperature anomaly is the difference between an observed temperature and a specified reference average. It is not an absolute outdoor temperature. Before comparing two climate charts, check their region, period, variable, dataset, units and baseline. A different zero line can make two valid charts show different anomaly values for the same observation.
An original two-baseline example
Suppose a hypothetical region's annual mean is 15.2°C. Reference period A has a mean of 14.0°C; reference period B has a mean of 14.4°C.
Anomaly relative to A = 15.2 − 14.0 = +1.2°C
Anomaly relative to B = 15.2 − 14.4 = +0.8°C
The observation did not change. The reference changed by 0.4°C, shifting the anomaly by the same amount. These numbers are a teaching example, not measured global temperatures or an estimate of a pre-industrial baseline.
| Year | Mean temperature | Anomaly vs A | Anomaly vs B |
|---|---|---|---|
| 1 | 14.6°C | +0.6°C | +0.2°C |
| 2 | 14.9°C | +0.9°C | +0.5°C |
| 3 | 15.2°C | +1.2°C | +0.8°C |
Why a constant baseline change does not change the trend
Subtracting a different constant from every value shifts all points vertically. In the example, the difference from year 1 to year 3 is 0.6°C under either baseline. That algebra explains rebasing one consistent series; it does not establish that all differences between independently produced datasets are just baseline offsets.
Spatial coverage, land/ocean treatment, measurement adjustments and aggregation can also differ. NASA's FAQ explains its use of anomalies and why simply averaging available station temperatures is not an adequate global analysis.
Read a climate chart in this order
- Variable: surface air temperature, sea-surface temperature or another quantity?
- Area: global, hemispheric, national or a single station?
- Time: monthly, annual or a rolling average?
- Reference: which baseline period and how is the anomaly calculated?
- Uncertainty/version: what does the dataset document about coverage and revisions?
The largest anomaly is not necessarily the warmest absolute month
A month can be unusually warm relative to its own normal while still being cooler in absolute temperature than another season. Compare like with like. A local cold day also measures a different scope from a global annual average. Neither observation should be silently substituted for the other.
How to rebase responsibly
For a consistent series, calculate the mean anomaly over the new reference period and subtract it from each anomaly. Verify that the required months/years and geographic definitions align. Do not apply an assumed universal offset between every dataset, especially for a reference period the dataset does not cover. State the source, version and method so another reader can reproduce the change.
The arithmetic examples here were independently checked. No current global-temperature record or new climate finding is claimed. For the difference between a relative percentage and a temperature threshold in everyday weather, compare the dew-point explanation.