Why seasonal anomalies use a 1993–2016 hindcast climatology
Seasonal forecast anomalies are measured against each model’s own hindcast climate, not observations. Why the reference period matters when comparing models.
Model climate, not observed climate
Every anomaly on this site is measured against the forecasting system’s own hindcast climatology, not against observations. A hindcast, or reforecast, is the same model run for past years as if it were forecasting them. Averaging those runs gives the model’s expected value for each start month, lead and location. Subtracting it expresses the forecast relative to the model’s own expected climate and accounts for its mean drift with lead. It does not by itself calibrate the forecast against observations.
Why 1993 to 2016
The products shown here use the common C3S reference of 1993–2016. Using the same 24-year window helps align the comparison: a +1 °C anomaly from ECMWF and a +1 °C anomaly from DWD are both departures from the same era. Some centres run longer hindcasts, but only the common period is used here.
Consequences for reading the maps
Temperature anomalies relative to this older baseline can include the warming trend as well as year-to-year variability. Positive anomalies therefore need not indicate an unusual outcome relative to recent decades. The effect varies by location, season and model; zero is not a direct estimate of today’s observed normal. Other C3S products, including some climate indices, use different reference conventions.
The reference period also matters for probabilities: tercile and quintile thresholds are the empirical percentiles of hindcast members pooled across the same 24 years, computed separately for each start month and lead. Changing the reference period can move the thresholds and change the probabilities, although individual values need not change.
What is not corrected
Subtracting a mean does not correct errors in variability, trends, teleconnections or the shape of the forecast distribution. Neither anomaly size nor model-relative thresholds establish reliability. The reference period and member counts used for each layer are recorded in the site’s provenance data.
Sources and method
These references explain the underlying science and provider conventions. Descriptions of this site’s calculations, colours and coverage were checked against its processing code and imported data.