Seasonal forecasts predict how temperature and precipitation over the coming months are likely to depart from their long-term average, expressed as probabilities across three categories: below-normal, near-normal, and above-normal. Rather than a single deterministic outlook, the forecast communicates the relative likelihood of each category based on the current state of the ocean and atmosphere – including slowly-evolving drivers such as ENSO – that shift the odds of seasonal climate outcomes.
How the forecast is produced
The models behind the forecast have changed over the life of the product. Choose an era below for the methodology that applies to it.
About C3S+NMME Forecasts Starting in 2026
Beginning in April 2026, the CCSR/IRI probabilistic seasonal climate forecasts of precipitation and near-surface temperature is based on a super-ensemble of re-calibrated forecasts from the U.S. National Oceanographic and Atmospheric Administration (NOAA)’s North American Multi-Model Ensemble Project (NMME) and the European Union’s Copernicus Climate Change Service (C3S).
The NMME ensemble seasonal prediction systems include NOAA’s National Centers for Environmental Prediction, Environment and Climate Change Canada, NOAA/Geophysical Fluid Dynamics Laboratory, NCAR and COLA/University of Miami. The C3S seasonal forecast systems include ECMWF, Météo-France, the German Weather Service (Deutscher Wetterdienst, DWD), and the Euro-Mediterranean Center on Climate Change (Centro Euro-Mediterraneo sui Cambiamenti Climatici, CMCC).
The output from each model is re-calibrated prior to multi-model ensembling to form reliable probability forecasts. The forecasts are presented on a 1-degree latitude-longitude grid.
The climatological base period currently used is 1991-2020.
About NMME Forecasts Starting in April 2017
Starting in April 2017, the IRI probabilistic seasonal climate forecast has been based on a re-calibration of output from the U.S. National Oceanic and Atmospheric Administration (NOAA)'s North American Multi-Model Ensemble (NMME) project. NMME combines coupled ocean-atmosphere prediction systems from several modeling centers – including NOAA's National Centers for Environmental Prediction, Environment and Climate Change Canada, NOAA/GFDL, NASA, NCAR, and COLA/University of Miami. Pooling models from multiple centers reduces the impact of any single model's biases and produces more reliable probabilistic forecasts than any individual model alone.
Each NMME model's output is statistically re-calibrated against observed climatology before the models are combined into a single multi-model forecast, presented on a 1-degree latitude-longitude grid. The climatological base period currently used is 1991-2020.
Methodological Details
The ensemble mean precipitation (or temperature) forecast from each individual model, on an interpolated 1-degree latitude-longitude grid, is used in conjunction with extended logistic regression (ELR; Wilks 2009, Meteorol. Appl. DOI: 10.1002/met.134) to produce probability forecasts from each individual model; these probabilities are then averaged together with equal weight to create a multi-model forecast probability (Vigaud et al. 2017, DOI: 10.1175/MWR-D-17-0092.1). In the case of precipitation, the final probability maps are smoothed spatially with a 9×9 point Gaussian smoother.
The ELR is trained using hindcast ensemble-means and observed precipitation (or temperature) data over the 1991-2020 period. The observed precipitation data is from the CPC-CMAP-URD dataset, while the GHCN-CAMS dataset is used for observed temperatures; both datasets are given on the same 1-degree latitude-longitude grid. Observed tercile-category occurrences are used to train the ELR.
Interpreting the maps
The forecast maps show the probability of the most-likely (tercile) category of precipitation or near-surface temperature for a given target season. White areas indicate grid points where all three categories are roughly equally likely (no strong signal), while grey areas indicate that the near-normal category is dominant.
About the Dry mask: To address the discontinuity in observed precipitation between 0 and non-zero seasonal cumulated amounts, and the fact that GCMs tend to have very few non-rainy days compared to observation due to their tendency to simulate too many rainfall events with intensities overall too low (Goddard et al. 2001), ELR forecast are only produced when at least 10% of the training sample received less than or equal to 1 mm over a 3-month season.
About the colorscale: The “40” category includes probabilities from 37.5 to 42.5. The other categories are designed similarly.
Disclaimer: The IRI seasonal forecast is a research product. Please see the NOAA CPC forecast for the official seasonal forecast over the U.S. Please consult your country's national meteorological service for the official forecast for your country.