Hydrological forecasts are produced by forcing meteorological forecasts through a hydrological model or conceptual algorithm. In EFAS, meteorological forecasts from different numerical weather prediction systems are used for different forecasting ranges.
| Product | Forecast Name | Provider | EFAS acronym | Maximum lead time | N. of Ensembles | Spatial Resolution |
|---|---|---|---|---|---|---|
| Flash flood/Medium range | COSMO | 5 days | 20 | ~ 7 km | ||
| Flash flood | Opera nowcasting | UPC | Opera | 4 hours | 1 | ~ 2 km |
| Medium range | DWD | Det. DWD | 5 days | 1 | ~ 6.5 km | |
| Medium range | ICON* | DWD | Det. DWD | 7 days | 1 | ~ 13 km |
| Medium range | HRES | ECMWF | ECMWF-HRES | 10 days | 1 | ~ 9 km |
| Medium range | ENS | ECMWF | ECMWF-ENS | 15 days | 51 | ~ 18 km until 26.06.2023 ~ 9 km from 27.06.2023 |
| Extended range | ENS-extended** | ECMWF | ECMWF-ENS-Ext | 6 weeks | 51 (101 from 27.06.2023, but only first 51 used in EFAS) | ~ 36 km |
| Seasonal outlook | SEAS5 | ECMWF | SEAS5 | 8 weeks | 51 | ~ 36 km |
* The deterministic DWD forecast in EFAS is produced by combining the higher resolution ICON-EU model with the lower resolution ICON global model. The global fields are used after 5 days (when ICON-EU is not available) and also used outside of the ICON-EU domain during the first 5 days. This is necessary as the ICON-EU domain (23.5°W–62.5°E, 29.5°N–70.5°N) does not fully cover the EFAS model domain (on the edges of the domain, mainly in the eastern areas near the Caspian Sea). This merging is done by simply copying the ICON global fields (precipitation, temperature and evapotranspiration) over to those areas where the ICON-EU is not available. Please note, on the outer edges of the ICON-EU domain, where the ICON field is used as a substitute, there can be some usually small discontinuities in the forcing fields (as these two models are genuinely two different forecast runs with potentially different weather situations evolving). In extreme cases this might produce some noticeable features in the flood signal, but the likelihood for this to happen is expected to be very small.
** The EFAS sub-seasonal (or extended-range) forecasts changed the way the forcing is used from 12 March 2025 (EFAS version 5.4). Before this date, the EFAS sub-seasonal forecasts only relied on the ENS-extended from day1 to week6 all the way. However, from v5.4, the sub-seasonal forecast simulations combine the use of the ENS-extended with the higher resolution ENS. This means, in the first 15 days of the simulations the higher resolution ENS meteorological forecasts are used instead of the low resolution ENS-extended. The combination of the two forcing data sets are mechanical, i.e. for each ensemble member the high resolution meteorological fields are used as forcing in the first 15 days, simply followed by the low-resolution extended-range meteorological fields from day16 to week6. There is no additional similarity metric applied amongst the members in order to pair them. For example, member 1 in the first 15 days will be followed by member 1 in day16 to week6, member 2 will be followed by with member 2, etc. Although the combination of the different resolution meteorological forcings, joined at day15, can create some discontinuities in the hydrological fluxes from day16, especially over complex orography, these were shown to be limited and thus allow the utilisation of the much higher meteorological resolution in the first 15 days (9 km vs 36 km).