Section 2 The ECMWF Integrated Forecasting System - IFS

Integrated Forecasting System - IFS

The ECMWF Integrated Forecasting System (IFS) consists of several components coupled together in various different ways:

an atmospheric model which is run at various resolutions appropriate to the forecast length:
- medium range ensemble (ENS) forecast (Day 0 to Day 15).
- Ensemble Control Forecast (ex-HRES) (Day 0 to Day 15). HRES and Ensemble Control Forecast (ex-HRES) are scientifically, structurally and computationally identical. The Ensemble Control Forecast (ex-HRES) output runs on the schedule of HRES in Cy48 and earlier. It runs before the medium range ensemble starts. It is labelled HRES for the convenience of users but the name will be withdrawn in a future update cycle.
- sub-seasonal range forecast (Day 0 to Day 46).
- seasonal forecast (month 0 to month 7 or month 0 to month 13).
an ocean wave model (ECWAM).
an ocean model (NEMO) including a sea ice model, the Louvain-la-Neuve Sea Ice Model (LIM2).
a land surface model (HTESSEL) including a lake model (FLake).
a data analysis system (4D-VAR).
perturbation techniques for generation of the ensembles.

These are outlined in Fig2-1, Fig 2-2 and Fig2-3. The models and programs run co-operatively to produce forecasts from analysis time out to days, weeks or months ahead.

Fig2-1: ECMWF Integrated Forecasting System (IFS) Illustrates interactions between components of the IFS (observation assimilation and post-processing not shown)

Fig2-2: Exchange of physical quantities between the atmospheric, ocean wave and ocean models. All the models exchange information and additionally the Wave model gains from the Ice model information on ice cover.

Exchange of physical quantities between the atmospheric, ocean wave and ocean models.

The Atmospheric models give:

to the Wave model, information on:
- air density, ice cover, and surface wind and gusts.
to the Ocean model, information on:
- solar energy fluxes and mechanical fluxes.

The Wave model gives:

to the Atmospheric models, information on:
- surface roughness (associated with forecast waves).
to the Ocean model, information on:
- stress, drift and turbulent energy.

The Ocean model gives:

to the Atmospheric model, information on:
- sea-surface temperature, surface energy fluxes, currents.

The Ice Model gives:

to the Atmospheric model, information on:
- ice cover.
to the Wave model, information on:
- ice cover.

Fig2-3: ECMWF Coupling sequence and exchange of physical quantities between the atmospheric, ocean wave and ocean models.

Atmospheric Models are coupled with the Wave Model (ECWAM) and the Ocean Model (NEMO) because:

the ocean is the lower boundary for the atmosphere for a large part of the earth.
the lower boundary gives important feedback to the atmosphere if accurately modelled (especially important for long-range predictions like monthly or seasonal forecasting and for features such as El Nino-Southern Oscillation, ENSO).
the ocean state (including sea‐ice) can change on a daily timescale and these variations can be important in certain situations during the forecast evolution. Important impacts of the modelled state of the ocean upon the evolution include:
- drag from the waves on the atmosphere (can help prevent over-deepening of lows as bigger waves impart more drag),
- ice variations and extent (gives more information on boundary temperatures, can hinder the propagation of ocean waves, reduced drag over ice sheets can allow increased winds),
- storm or hurricane feed back to or from the ocean (potential for upwelling of colder water induced by passage of a major storm or depression over a temperature-stratified ocean).

Atmospheric Models are coupled with HTESSEL and FLake because:

the land is the lower boundary for the atmosphere for a less extensive area than the sea but it has a complex orography and exhibits far more temporal and spatial variability in the characteristics of energy storage and exchange.
the lower boundary gives important feedback to the atmosphere if accurately modelled (important for short and long-range forecasting).
the state of the landmasses (including soil moisture, cover and depth of snow, and lake ice) can change on a daily timescale and these variations can be important in certain situations during the forecast evolution. Important aspects of the modelled state of the land that can affect the atmospheric evolution include:
- rainfall absorption and soil moisture (give more information on boundary moisture flux),
- ice variations on sea and lakes (gives more information on albedo and boundary heat and moisture fluxes),
- land/lake temperature variations for evaluation of heat fluxes,
- snow variations and extent (gives more information on albedo and boundary heat and moisture fluxes).

The coupled model configuration is used with:

the Medium Range ensemble forecast and Ensemble Control Forecast (ex-HRES) forecast (Day 0 to Day 15) twice daily based on 00UTC and 12UTC data.
the Sub-seasonal range forecast (Day 0 to Day 46) once daily based on 00UTC data.
the Seasonal forecast (Month 0 to Month 7 once each month, Month 0 to Month 13 once each quarter) on first day of the month based on 00UTC data.

More information is given within the Users Guide on the structures of the Global Atmospheric Model, the Ocean Wave Model, and the Dynamic Ocean Model and regarding the model resolution currently used within the IFS.

Real-time forecasts are initialized from the operational analysis using 4D-Var. Re-forecasts are initialized from ERA5, except for soil initial conditions (soil temperature, soil moisture, snow initial conditions) which are provided by an offline soil reanalysis. Oceanic models are initialized from the real-time suite, NEMOVAR.

(FUG Associated with Cy49r1)

Space shortcuts

Page tree

Integrated Forecasting System - IFS