How are climate normals calculated?
Climate normals are reference statistics describing longer-term temperature, precipitation, and related conditions at a place. The World Meteorological Organization uses the most recent thirty-year standard period ending in a year ending with zero; the widely adopted current period is 1991–2020. One day being above its normal does not by itself establish a climate-change trend.
Reference period and calculation steps
1. Why use thirty years? Weather changes substantially from day to day and between seasons. Treating one hot summer or snowy winter as typical would mislead. A thirty-year window combines variation across many years and gives observing stations a common basis for comparison. WMO supports updating standard reference periods every decade: the shift from 1981–2010 to 1991–2020 provides a baseline closer to recent conditions. Research into long-term change may separately retain a fixed older baseline. These are different questions, so always identify which period a chart uses rather than comparing unexplained numbers. 2. How are daily or monthly values derived? For one station, observations referring to the same calendar day or month across years are checked for quality first. A monthly mean temperature is derived from valid daily observations and then the values of that month over the reference years. For a monthly precipitation normal, the totals for that month across the years are averaged; this does not state how often an exceptionally intense shower occurs. Missing records, relocation, a changed instrument, or a different observation time can affect comparability. The publishing weather service should document its treatment of these issues and any homogenization. 3. What does an average conceal? The same monthly mean could describe consistently mild days in one place or a month with strong hot and cold extremes elsewhere. Read high and low values, wet-day frequency, percentiles, and extreme-event measures separately. A few storms can account for much of a monthly rainfall total. A normal is not a prediction of the weather on that date: it is the statistical center of a past reference period. Elevation, coastal exposure, observation schedule, and record completeness also matter when comparing two towns.
Anomalies, comparisons, and limits
4. How is an anomaly calculated? Subtract the reference-period normal for the same place and month from the observed monthly temperature. If the observed value is 20 °C and the normal is 18 °C, the anomaly is +2 °C; the anomaly is a difference, not the absolute temperature. Precipitation anomalies can be reported in millimeters or as a percentage of normal. Percentages can become especially large in a season when normal rainfall is tiny. Before comparing anomaly colors across organizations, check the baseline years, station coverage, units, and map legend. Different baselines can produce different-looking deviations from identical observations. 5. Does a new normal erase warming? Moving the reference to 1991–2020 can make the temperature anomaly for the same recent day look smaller than it would against a cooler older period. This does not reverse the physical warming; the comparison line has moved. For long-term trends, examine consistent raw or quality-adjusted time series rather than splicing incompatible normals. Station relocation, urban development, and instrument changes can influence local records, so consult the national service’s methods and uncertainty. A countrywide average can also conceal differences between mountains, inland basins, and coasts. 6. How should it be used in everyday decisions? When reading a city’s November normal, identify the station, reference years, and measured variable. Normals offer background for clothing, agriculture, or energy planning, but they do not replace a short-range forecast for a particular day. Follow current official bulletins for storms, heat, and flooding. A month with a low average rainfall can still contain a severe downpour. A responsible explanation says explicitly that a normal is a reference statistic and that extremes and local conditions require separate evidence.