Reading environmental data
What is a climate normal?
A climate normal is the average of a weather variable over a fixed multi-decade window, used as the reference that departures are measured against. The World Meteorological Organization's standard window is thirty years, currently 1991–2020. It is a description of a recent past, not a target and not a healthy value — and because the window moves every decade, a warming climate quietly raises the bar that “above normal” is judged by.
Why it has to be a window, and why the window moves
A single year is weather. A normal is the thing that makes one year comparable to another.
Weather varies enormously year to year for reasons that have nothing to do with any trend. To say a day was unusually hot you need to know what usual is at that place on that date, and thirty years is the convention: long enough to average out year-to-year noise, short enough to describe a climate people actually live in.
The window is recalculated every ten years. 1961–1990 gave way to 1971–2000, then 1981–2010, and 1991–2020 is current. Each update replaces a cooler decade with a warmer one, so the normal rises. The consequence is uncomfortable and worth stating plainly: the same weather is less anomalous against a newer normal, and a run of “near normal” summers can conceal a climate that has moved.
It also has to be per place and per date. A departure measured against somebody else's normal is meaningless — Chennai's normal and Jodhpur's are about ten degrees apart, and a threshold that ignored the difference would declare an event in one city every summer and never in the other.
- Normal
- The multi-decade mean for a place and a date. Not an average of everything, and not a target.
- Departure
- How far a reading ran from its normal, in the reading's own units. The quantity most Indian heat criteria are built on.
- Baseline
- The window a normal is computed over. State it or a departure cannot be checked.
- Anomaly
- The same idea as departure, and the word climate science usually uses for it.
The reference this site uses, and what it is used for
Stated because a departure without its baseline is not a checkable claim.
Normal window
1991-2020
WMO standard climate normal period
Heatwave departure — what the normal is used for
4.5 °C above normal
India Meteorological Department
Severe heatwave departure
6.5 °C above normal
India Meteorological Department
Plains floor — the absolute gate the departure sits behind
40 °C
India Meteorological Department
Legal limit on a departure
None
A normal is a reference, not a standard
The method matters as much as the window. This site computes each station's normal per calendar day as the mean daily maximum over a plus-or-minus seven-day window across the thirty years — about 450 samples for each day of the year — because a single calendar date across thirty years is only thirty numbers, and thirty numbers make a noisy reference.
There is nothing to break here. A normal regulates nothing; it is the yardstick a criterion is written against, and a criterion is not a health threshold either.
How to read an “above normal” claim
Four questions, and most published anomalies answer none of them.
Which window? 1991–2020 and 1961–1990 give materially different anomalies for the same reading. Climate science often uses a pre-industrial or 1850–1900 baseline for exactly this reason; meteorology uses the recent thirty.
Computed how? A daily normal, a monthly normal and a seasonal normal are three references. A figure that says “above normal” without saying above what period's normal is not checkable.
At which place? Normals are per station. A regional average of normals is a fourth thing again.
Above normal, or above a limit? They are different claims and only one of them can be broken. Rainfall in Delhi has a normal and no legal threshold at all, so “above normal” there is a comparison and never a breach.
What this number cannot tell you
- A normal cannot tell you what is safe. It describes what has recently been usual, which in a changing climate is a moving and increasingly poor proxy for that.
- It cannot be compared across baselines. Two anomalies computed on different windows are different quantities.
- It cannot be built where there is no long record. A station with ten years has no normal, and borrowing a neighbour's introduces the error the per-station rule exists to avoid.
- It says nothing about extremes. A normal is a mean, and the tail of a distribution can move while the mean does not.
Where this comes from
Every figure above is read out of one of these, at the address printed beside it.
- WMO Climatological Normals World Meteorological Organization · The thirty-year convention and the 1991–2020 period
- Heat wave criteria India Meteorological Department · The departure thresholds a normal is used for
- ERA5 reanalysis archive ECMWF, served by Open-Meteo · The series this site computes its per-station normals from
Reuse freely — CC BY 4.0. The grant covers this page. Each source above keeps its own terms, which is why every one of them is named.
Next
The reading this explains, the explanations next to it, and the one place to stand in it.