1. Forecast system version
System name: CMCC-SPS4
First operational forecast run: 1st August, 2025
2. Configuration of the forecast model
Is it a coupled model ? YES
Coupling frequency:
Atmosphere-Ocean: 60 minutes (every second full time-step of atmospheric model)
Atmosphere-Land: 30 minutes (also full time-step of atmospheric model)
Atmosphere-Sea Ice: 30 minutes (also full time-step of atmospheric model)
Detailed documentation:
CMCC technical documentation: Sanna et al. (2025) CMCC Technical Note n.301 DOI: https://doi.org/10.25424/cmcc-dkcv-fs25
2.1 Atmosphere and land surface
| Model | CESM2.3 - CAM6 (Atmosphere) CESM2.3 - CLM 5.1(Land surface) |
|---|---|
| Horizontal resolution and grid | 1/2° lat-lon approx |
| Atmosphere vertical resolution | 83 levels in the vertical |
| Top of atmosphere | 0.01 hPa approx. |
| Soil levels (layers) | 25 Layer 1 (soil): 0-0.020m Layer 2 (soil): 0.020-0.060m Layer 3 (soil): 0.060-0.0120m Layer 4 (soil): 0.0120-0.200m Layer 5 (soil): 0.200-0.320m Layer 6 (soil): 0.320-0.480m Layer 7 (soil): 0.480-0.680m Layer 8 (soil): 0.680-0.920m Layer 9 (soil): 0.920-1.200m Layer 10 (soil): 1.200-1.520m Layer 11 (soil): 1.520-1.880m Layer 12 (soil): 1.880-2.280m Layer 13 (soil): 2.280-2.720m Layer 14 (soil): 2.720-3.260m Layer15 (soil): 3.260-3.900m Layer 16 (soil): 3.900-4.640m Layer 17 (soil): 4.640-5.480m Layer 18 (soil): 5.480-6.420m Layer 19 (soil): 6.420-7.460m Layer 20 (soil): 7.460-8.600m |
| Time step | Main (Physics) Time-step: 30 minutes. “Tracer” Advection Time step: 20 minutes Fluid-Dynamics Time step: 10 minutes |
Detailed documentation:
CAM Model documentation: https://www.cesm.ucar.edu/models/cam
CLM Model documentation: https://www.cesm.ucar.edu/models/clm
CMCC technical documentation: Sanna et al. (2025) CMCC Technical Note n.301 DOI: https://doi.org/10.25424/cmcc-dkcv-fs25
2.2 Ocean and cryosphere
| Ocean model | NEMO v4.2 |
|---|---|
| Horizontal resolution | 1/4° |
| Vertical resolution | 75 levels in the vertical |
| Time step | 20 minutes |
| Sea ice model | CICE6 |
| Sea ice model resolution | 1/4° |
| Sea ice model levels | 8 ice layer + 3 snow layers (5 ice categories) |
| Wave model | NO |
| Wave model resolution | N/A |
Detailed documentation:
Nemo Model documentation: https://sites.nemo-ocean.io/user-guide/
CMCC technical documentation: Sanna et al. (2025) CMCC Technical Note n.301 DOI: https://doi.org/10.25424/cmcc-dkcv-fs25
3. Initialization and initial condition (IC) perturbations
3.1 Atmosphere and land
| Hindcast | Forecast | |
|---|---|---|
| Atmosphere initialization | EDA | EDA |
| Atmosphere IC perturbations | 10 | 10 |
Land Initialization | Forced monthly run from three continuous transient simulations started in January 1960, with atmospheric forcings provided by 3 different EDA analyses | Forced monthly run from three continuous transient simulations started in January 1960, with atmospheric forcings provided by 3 different EDA analyses |
| Land IC perturbations | 3 | 3 |
| Soil moisture initialization | From land initialization | From land initialization |
| Snow initialization | From land initialization | From land initialization |
| Unperturbed control forecast? | NO | NO |
| Horizontal resolution of perturbation | N/A | N/A |
| Perturbations in +/- pairs | NO | NO |
| Data assimilation method for control analysis | N/A | N/A |
Detailed documentation:
For more details on ERA5 and EDA: Hersbach et al. (2020) DOI: https://doi.org/10.1002/qj.3803; Isaksen et al. (2010) DOI: 10.21957/obke4k60
CMCC technical documentation: Sanna et al. (2025) CMCC Technical Note n.301 DOI: https://doi.org/10.25424/cmcc-dkcv-fs25
3.2 Ocean and cryosphere
| Hindcast | Forecast | |
|---|---|---|
| Ocean initialization | C-GLORS Global Ocean 3D-VAR | C-GLORS Global Ocean 3D-VAR |
| Ocean IC perturbations | 4 | 9 |
| Unperturbed control forecast? | YES | YES |
Detailed documentation:
More details on ocean and sea-ice data assimilation: Storto and Masina (2016) DOI: https://doi.org/10.5194/essd-8-679-2016; Cipollone et al. (2023) DOI: https://doi.org/10.5194/egusphere-2023-254
CMCC technical documentation: Sanna et al. (2025) CMCC Technical Note n.301 DOI: https://doi.org/10.25424/cmcc-dkcv-fs25
4. Model uncertainties perturbations:
| Model dynamics perturbations | NO |
|---|---|
| Model physics perturbations | YES (Ocean Model only, only during perturbed data assimilation cycles) |
If there is a control forecast, is it perturbed? | There is no control forecast |
Detailed documentation:
More details on ocean DA perturbations: Storto and Andriopoulos (2021) DOI: https://doi.10.1002/qj.3990
CMCC technical documentation: Sanna et al. (2025) CMCC Technical Note n.301 DOI: https://doi.org/10.25424/cmcc-dkcv-fs25
Boundary conditions - climate forcings
5. Boundary conditions - climate forcings
Most forcing data comes from the CMIP6 protocol.
| Greenhouse gases | Up to 2014, CMIP6 historical values of CO2, CH4, N2O, CFC11 and CFC12 from Meinshausen et al. (2017). From 2015 onwards, greenhouse gas forcings follow the SSP5-8.5 scenario from Meinshausen et al. (2020). |
|---|---|
| Ozone | Ozone values come from a simulation conducted with WACCM and forced with historical values up to 2014, with the SSP5-8.5 scenario from 2015 onwards. |
| Tropospheric aerosols | Emissions come from historical data up to 2014 and from CMIP6 for the following years (reference Hoesly et al., GMD, 2017). Data, provided to the model at monthly frequency, are advected through the module MAM4 along the simulation. |
| Volcanic aerosols | Volcanic stratospheric aerosols are the official CMIP6 forcings provided by Thomason et al (2018). |
| Solar forcing | SSI (Reconstructed spectral solar irradiance) from data are constructed as arithmetic mean of source data (NRLSSI2+SATIRE-TS)/2. from CMIP6 annual mean forcings provided by Matthes et al (2017). However, several modifications of source data were performed before averaging: sub-annual variability has been added for periods before 1882 and after 2014; Inhomogeneity in one source dataset at 11 Aug 1878 has been removed; Sampling artefacts in one source dataset have been removed beyond 10,000nm. Moreover, Total Solar irradiance does not come from source data but is the integral over SSI between 10 and 10,000nm. |
6. Forecast system and hindcasts
| Forecast frequency | Monthly |
|---|---|
| Forecast ensemble size | 50 |
| Hindcast years | 1993-2022 |
| Hindcast ensemble size | 30 |
| On-the-fly or static hindcast set? | static |
7. Other relevant information
Initialization strategy
The 10 atmospheric perturbed initial conditions, the 3 land perturbed initial conditions and the 9 (4 in hindcast) ocean initial conditions (8/3 perturbed plus the unperturbed in forecast/hindcast mode respectively) are combined to yield 270 (120 in hindcast mode) possible perturbed forecast system initial conditions. From this set of combinations, 50 initial conditions uniquely defined (30 in hindcast mode) are randomly chosen, to produce the Forecast Ensemble, which runs in burst mode and starts on the 1st of each calendar month.
Interpolation details
CMCC-SPS4 rests on the GCM CMCC-CM3, which produces netCDF outputs at different resolutions for the different components (Atmoshere and Land share the same horizontal resolution of about 0.5 degrees, while Ocean and Sea-Ice have an approximate horizontal resolution of 0.25 degrees). 3-dimensional fields, namely temperature, geopotential height, humidity and horizontal wind components are archived only at the requested pressure levels, as direct model output, following a vertical interpolation performed by the model itself, which applies the algorithm by Trenberth et al. 1993. All operations such as time sampling, time averaging, horizontal regridding are carried out with ncl/cdo procedures, specifically developed to meet the C3S requirements in terms of output frequency (6-hourly, 12-hourly or daily) and horizontal resolution (1X1 degree). The space interpolations are performed by means of bilinear/mass-conserving interpolation algorithms, according to the different variables involved.
Detailed documentation:
CMCC technical documentation: Sanna et al. (2025) CMCC Technical Note n.301 DOI: https://doi.org/10.25424/cmcc-dkcv-fs25
8. Where to find more information
Sanna, A., A. Borrelli, P. Athanasiadis, S. Materia, A. Storto, S. Tibaldi, S. Gualdi, 2017: CMCC-SPS3: The CMCC Seasonal Prediction System 3. Centro Euro-Mediterraneo sui Cambiamenti Climatici. CMCC Tech. Note RP0285, 61pp. https://www.cmcc.it/publications/rp0285-cmcc-sps3-the-cmcc-seasonal-prediction-system-3
Gualdi, S., A. Sanna, A. Borrelli, A. Cantelli, M. del Mar Chaves Montero, S. Tibaldi, 2020: The new CMCC Operational Seasonal Prediction System SPS3.5. Centro Euro-Mediterraneo sui Cambiamenti Climatici. CMCC Tech. Note RP0288, 26pp. DOI: https://doi.org/10.25424/CMCC/SPS3.5
Sanna, A., S. Gualdi, M. Benassi, A. Borrelli, D. Peano, M. Hashemi Devin, A. Cipollone and S. Tibaldi, 2025: The new CMCC Seasonal Prediction System SPS4. CMCC Tech. Note n. 301, 31 pp. DOI: https://doi.org/10.25424/cmcc-dkcv-fs25
Trenberth, Berry, Buia, 1993: Vertical Interpolation and Truncation of Model-Coordinate Data, NCAR/TN-396.
