S2S global sub-seasonal to seasonal ensemble reforecasts from multiple NWP and research centres since 2015

The Sub-seasonal To Seasonal dataset (S2S) consists of global ensemble real-time forecasts and reforecasts from thirteen numerical weather prediction (NWP) and research centres.
S2S project behind the dataset started in 2013 as a joint initiative of the World Weather Research Programme (WWRP) and the World Climate Research Programme (WCRP).
The goal of S2S project was to improve sub-seasonal forecast skill through combining multiple forecasting systems, enable multi-model evaluations and enhance knowledge sharing between operational centres.
Although S2S project ended at the end of 2023, data archiving continues and is currently undergoing another five-year phase (2024–2028).
There are two related datasets in ECDS:
- real-time forecasts (s2s-forecasts)
- reforecasts complementing the real-time data (s2s-reforecasts)
There are two types of the reforecasts - fixed and on-the-fly depending on the model and its current configuration.
The S2S reforecasts are covering varying periods since 1981.
The typical forecast length is from 34 to 65 days (the same as in the real-time forecasts) with ensemble sizes varying from 4 to 33.
S2S variables are available on surface, soil, oceanic, potential temperature and selected pressure levels.
Please consult the Documentation section for details, including data sources, model configurations and limitations.
The data discovery and sample downloads can be done either via ECDS forms (under Download tab) or alternatively via dedicated MARS S2S catalogue (MARS stands for ECMWF's Meteorological Archival and Retrieval System).
(S2S picture used - courtesy of Mio Matsueda, S2S Museum)