1. Ensemble version | ||||
|---|---|---|---|---|
| Ensemble identifier code | GEPS 7 | GEPS 6 | GEPS 5 | GEPS 4 |
| Short Description | Global ensemble prediction system (GEPS) version 7.0. The sequential Ensemble Kalman Filter (EnKF) is replaced by the Local Ensemble Transform Kalman Filter (LETKF) data assimilation algorithm. The version of the GEM model is upgraded to 5.1 with more advanced physics. The number of vertical levels is increased from 81 to 85 for the data assimilation and from 45 to 85 for the forecast. A Stochastic Parameter Perturbation (SPP) method is introduced to replace the Physics Tendency Perturbation (PTP). GEPS, which has 21 members, is extended to day 32 once a week (Thursday at 00Z). | Global ensemble prediction system with initial conditions generated with the Ensemble Kalman Filter which receives observations that are background-checked and bias-corrected by the ECCC Global Deterministic Prediction System. Different members have different model configuration perturbations (multi-parametrization physics). GEPS, which has 21 members, is extended to day 32 once a week (Thursday at 00Z). | Global ensemble prediction system with initial conditions generated with the Ensemble Kalman Filter which receives observations that are background-checked and bias-corrected by the ECCC Global Deterministic Prediction System. Different members have different model configuration perturbations (multi-parametrization physics). GEPS, which has 21 members, is extended to day 32 once a week (Thursday at 00Z). | Global ensemble prediction system with initial conditions generated with the Ensemble Kalman Filter which receives observations that are background-checked and bias-corrected by the ECCC Global Deterministic Prediction System. Different members have different model configuration perturbations (multi-parametrization physics). GEPS, which has 21 members, is extended to day 32 once a week (Thursday at 00Z). |
| Research or operational | Operational | Operational | Operational | Operational |
| Data time of first forecast run |
| 04/07/2019 | 20/09/2018 | 02/07/2015 |
2. Configuration of the EPS | ||||
| Is the model coupled to an ocean model? | Yes. | Yes. | No, persistent SST anomaly | No, persistent SST anomaly |
| If yes, please describe ocean model briefly including frequency of coupling and any ensemble perturbation applied | The ocean model is NEMO 3.6 at 0.25° horizontal resolution and 50 vertical levels. Initialized with ECCC ocean analysis. There is no ocean initial perturbation. Frequency of coupling is 15 minutes. | The ocean model is NEMO 3.6 at 0.25° horizontal resolution and 50 vertical levels. Initialized with ECCC ocean analysis. There is no ocean initial perturbation. Frequency of coupling is 15 minutes. | ||
| If no, please describe the sea surface temperature boundary conditions (climatology, reanalysis ...) | ||||
| Is the model coupled to a sea Ice model? | Yes. | Yes. | No, sea-ice adjusted to SST | No, sea-ice adjusted to SST |
| If yes, please describe sea-ice model briefly including any ensemble perturbation applied | The CICE 4.0 model is used for the sea-ice component. Initialized with ECCC sea ice analysis. No perturbation in sea ice initial condition. | |||
| Is the model coupled to a wave model? | No | No | No | No |
| If yes, please describe wave model briefly including any ensemble perturbation applied | ||||
| Ocean model | NEMO 3.6 at 0.25° resolution | NEMO 3.6 at 0.25° resolution | ||
| Horizontal resolution of the atmospheric model | Yin-Yang grid at 0.35° uniform resolution (~39 km) | Yin-Yang grid at 0.35° uniform resolution (~39 km) | Yin-Yang grid at 0.35° uniform resolution (~39 km) | 0.45º x 0.45º |
| Number of model levels | 85 | 45 | 45 | 40 |
| Top of model | 0.1 hPa | 0.1 hPa | 0.1 hPa | 2 hPa |
| Type of model levels | hybrid, log-hydrostatic pressure, Charney-Phillips grid | hybrid, log-hydrostatic pressure, Charney-Phillips grid | hybrid, log-hydrostatic pressure, Charney-Phillips grid | hybrid, log-hydrostatic pressure, Charney-Phillips grid |
| Forecast length | maximum 32 days (768 hours) | maximum 32 days (768 hours) | maximum 32 days (768 hours) | maximum 32 days (768 hours) |
| Run Frequency | once a week (Thursday 00Z) | once a week (Thursday 00Z) | once a week (Thursday 00Z) | once a week (Thursday 00Z) |
| Is there an unperturbed control forecast included? | Yes | Yes | Yes | Yes |
| Number of perturbed ensemble members | 20 | 20 | 20 | 20 |
| Integration time step | 15 Minutes | 15 Minutes | 15 Minutes | 15 Minutes |
3. Initial conditions and perturbations | ||||
| Data assimilation method for control analysis | mean of ENKF | mean of ENKF | mean of ENKF | mean of ENKF |
| Resolution of model used to generate Control Analysis | Yin-Yang grid at 0.35° uniform resolution (~39 km) L85 | Yin-Yang grid at 0.35° uniform resolution (~39 km) L81 | Yin-Yang grid at 0.35° uniform resolution (~39 km) L81 | 0.45º x 0.45º L74 |
| Ensemble initial perturbation strategy | Ensemble Kalman Filter (ENKF) | Ensemble Kalman Filter (ENKF) | Ensemble Kalman Filter (ENKF) | Ensemble Kalman Filter (ENKF) |
| Horizontal and vertical resolution of perturbations | Yin-Yang grid at 0.35° uniform resolution (~39 km) L85 | Yin-Yang grid at 0.35° uniform resolution (~39 km) L45 | Yin-Yang grid at 0.35° uniform resolution (~39 km) L45 | 0.45º x 0.45º L40 |
| Perturbations in +/- pairs | No | No | No | No |
| Initialization of land surface | ||||
| 3.1 What is the land surface model (LSM) and version used in the forecast model, and what are the current/relevant references for the model? | The GEM model a mosaic approach with 4 types land, water, sea ice and glacier is used. For the land part, a version of the ISBA scheme is used (Noilhan and Planton, 1989; Bélair et al. 2003a and 2003b) to simulate soil variables. | The GEM model a mosaic approach with 4 types land, water, sea ice and glacier is used. For the land part, a version of the ISBA scheme is used (Noilhan and Planton, 1989; Bélair et al. 2003a and 2003b) to simulate soil variables. | The GEM model a mosaic approach with 4 types land, water, sea ice and glacier is used. For the land part, a version of the ISBA scheme is used (Noilhan and Planton, 1989; Béir et al. 2003a and 2003b) to simulate soil variables. | The GEM model a mosaic approach with 4 types land, water, sea ice and glacier is used. For the land part, a version of the ISBA scheme is used (Noilhan and Planton, 1989; Bélair et al. 2003a and 2003b) to simulate soil variables. |
| Are there any significant changes/deviations in the operational version of the LSM from the documentation of the LSM? | No | No | No | No |
| 3.2 How is soil moisture initialized in the forecasts? (climatology / realistic / other) | ‘Realistic’ in the sense that a pseudo-analysis is done using a correction of the 1.5m temperature and relative humidity. It is an indirect analysis since a comparison between the trial field (6 hour forecast and observations at the screen level) are used to correct the soil moisture and temperature (for more detail see Bélair et al. 2003a and 2003b). | ‘Realistic’ in the sense that a pseudo-analysis is done using a correction of the 1.5m temperature and relative humidity. It is an indirect analysis since a comparison between the trial field (6 hour forecast and observations at the screen level) are used to correct the soil moisture and temperature (for more detail see Bélair et al. 2003a and 2003b). | ‘Realistic’ in the sense that a pseudo-analysis is done using a correction of the 1.5m temperature and relative humidity. It is an indirect analysis since a comparison between the trial field (6 hour forecast and observations at the screen level) are used to correct the soil moisture and temperature (for more detail see Bélair et al. 2003a and 2003b). | ‘Realistic’ temperature and relative humidity. It is an indirect analysis since a comparison between the trial field (6 hour forecast and observations at the screen level) are used to correct the soil moisture and temperature (for more detail see Bélair et al. 2003a and 2003b). |
| Is there horizontal and/or vertical interpolation of initialization data onto the forecast model grid? If so, please give original data resolution(s). | Yes the fields are generated on the global deterministic prediction system (GDPS) grid (about 25 km) while the global ensemble prediction system (GEPS) is different and coarser (about 39 km). An interpolation is done to the coarser grid after a filter was applied. | Yes the fields are generated on the global deterministic prediction system (GDPS) grid (about 25 km) while the global ensemble prediction system (GEPS) is different and coarser (about 39 km). An interpolation is done to the coarser grid after a filter was applied. | Yes the fields are generated on the global deterministic prediction system (GDPS) grid (about 25 km) while the global ensemble prediction system (GEPS) is different and coarser (about 50 km). An interpolation is done to the coarser grid after a filter was applied. | Yes the fields are generated on the global deterministic prediction system (GDPS) grid (about 25 km) while the global ensemble prediction system (GEPS) is different and coarser (about 50 km). An interpolation is done to the coarser grid after a filter was applied. |
| Does the LSM differentiate between liquid and ice content of the soil? | Yes | Yes | Yes | Yes |
| If so, how are each initialized? | Liquid soil moisture (I1) is initialized as explained above. Solid soil moisture (I2) is taken from trial fields. | Liquid soil moisture (I1) is initialized as explained above. Solid soil moisture (I2) is taken from trial fields. | Liquid soil moisture (I1) is initialized as explained above. Solid soil moisture (I2) is taken from trial fields. | Liquid soil moisture (I1) is initialized as explained above. Solid soil moisture (I2) is taken from trial fields. |
| If all model soil layers are not initialized in the same way or from the same source, please describe. | Same approach is used for all layers only the amplitude of the correction is different. | Same approach is used for all layers only the amplitude of the correction is different. | Same approach is used for all layers only the amplitude of the correction is different. | Same approach is used for all layers only the amplitude of the correction is different. |
| 3.3 How is snow initialized in the forecasts? (climatology / realistic / other) | ‘Realistic’, the snow depth analyzed following the Brasnett (1999) method. | ‘Realistic’, the snow depth analyzed following the Brasnett (1999) method. | ‘Realistic’, the snow depth analyzed following the Brasnett (1999) method. | ‘Realistic’, the snow depth analyzed following the Brasnett (1999) method. |
| Is there horizontal and/or vertical interpolation of data onto the forecast model grid? If so, please give original data resolution(s) | Yes the fields are generated on the global deterministic prediction system (GDPS) grid (about 25 km) while the global ensemble prediction system (GEPS) is different and coarser (about 39 km). An interpolation is done to the coarser grid after a filter was applied. | Yes the fields are generated on the global deterministic prediction system (GDPS) grid (about 25 km) while the global ensemble prediction system (GEPS) is different and coarser (about 39 km). An interpolation is done to the coarser grid after a filter was applied. | Yes the fields are generated on the global deterministic prediction system (GDPS) grid (about 25 km) while the global ensemble prediction system (GEPS) is different and coarser (about 50 km). An interpolation is done to the coarser grid after a filter was applied. | Yes the fields are generated on the global deterministic prediction system (GDPS) grid (about 25 km) while the global ensemble prediction system (GEPS) is different and coarser (about 50 km). An interpolation is done to the coarser grid after a filter was applied. |
| Are snow mass, snow depth or both initialized? What about snow age, albedo, or other snow properties? | The LSM is given the analysed snow depth and it compute the snow mass (prognostic variable) with that. | The LSM is given the analysed snow depth and it compute the snow mass (prognostic variable) with that. | The LSM is given the analysed snow depth and it compute the snow mass (prognostic variable) with that. | The LSM is given the analysed snow depth and it compute the snow mass (prognostic variable) with that. |
| 3.4 How is soil temperature initialized in the forecasts? (climatology / realistic / other) Realistic | Realistic, with the pseudo-analysis method of Bélair et al. 2003 a and b like the soil moisture | Realistic, with the pseudo-analysis method of Bélair et al. 2003 a and b like the soil moisture | Realistic with the pseudo-analysis method of Bélair et al. 2003 a and b like the soil moisture | with the pseudo-analysis method of Bélair et al. 2003 a and b like the soil moisture |
| Is the soil temperature initialized consistently with soil moisture (frozen soil water where soil temperature ?0°C) and snow cover (top layer soil temperature ?0°C under snow)? | Yes, some coherency checks are done in the LSM. | Yes, some coherency checks are done in the LSM. | Yes, some coherency checks are done in the LSM. | Yes, some coherency checks are done in the LSM. |
| Is there horizontal and/or vertical interpolation of data onto the forecast model grid? If so, please give original data resolution(s) | Yes the fields are generated on the global deterministic prediction system (GDPS) grid (about 25 km) while the global ensemble prediction system (GEPS) is different and coarser (about 39 km). An interpolation is done to the coarser grid after a filter was applied. | Yes the fields are generated on the global deterministic prediction system (GDPS) grid (about 25 km) while the global ensemble prediction system (GEPS) is different and coarser (about 39 km). An interpolation is done to the coarser grid after a filter was applied. | Yes the fields are generated on the global deterministic prediction system (GDPS) grid (about 25 km) while the global ensemble prediction system (GEPS) is different and coarser (about 50 km). An interpolation is done to the coarser grid after a filter was applied. | Yes the fields are generated on the global deterministic prediction system (GDPS) grid (about 25 km) while the global ensemble prediction system (GEPS) is different and coarser (about 50 km). An interpolation is done to the coarser grid after a filter was applied. |
| If all model soil layers are not initialized in the same way or from the same source, please describe. | Not the case. | Not the case. | Not the case. | Not the case. |
| 3.5 How are time-varying vegetation properties represented in the LSM? Is phenology predicted by the LSM? If so, how is it initialized? | From monthly values form a look-up table interpolated daily following the values of Appendix A of Giard and Bazile (2000) paper. | From monthly values form a look-up table interpolated daily following the values of Appendix A of Giard and Bazile (2000) paper. | From monthly values form a look-up table interpolated daily following the values of Appendix A of Giard and Bazile (2000) paper. | From monthly values form a look-up table interpolated daily following the values of Appendix A of Giard and Bazile (2000) paper. |
| 3.6 What is the source of soil properties (texture, porosity, conductivity, etc.) used by the LSM? | Agriculture Canada for the Canadian region, USDA for the USA and FAO for the rest of the world. | Agriculture Canada for the Canadian region, USDA for the USA and FAO for the rest of the world. | Agriculture Canada for the Canadian region, USDA for the USA and FAO for the rest of the world. | Agriculture Canada for the Canadian region, USDA for the USA and FAO for the rest of the world. |
| 3.7 If the initialization of the LSM for re-forecasts deviates from the procedure for forecasts, please describe the differences. | In the reforecast the land surface scheme was run for 30 years forced with ERA-interim re-analysis fields (screen-level temperature and dew-point as well as precipitation amount) at the resolution of the GEPS (39 km). For more details, see Gagnon et al. 2014 and Lin et al. 2016. | In the reforecast the land surface scheme was run for 30 years forced with ERA-interim re-analysis fields (screen-level temperature and dew-point as well as precipitation amount) at the resolution of the GEPS (39 km). For more details, see Gagnon et al. 2014 and Lin et al. 2016. | In the reforecast the land surface scheme was run for 30 years forced with ERA-interim re-analysis fields (screen-level temperature and dew-point as well as precipitation amount) at the resolution of the GEPS (39 km). For more details, see Gagnon et al. 2014 and Lin et al. 2016. | In the reforecast the land surface scheme was run for 30 years forced with ERA-interim re-analysis fields (screen-level temperature and dew-point as well as precipitation amount) at the resolution of the GEPS (50 km). For more details, see Gagnon et al. 2014 and Lin et al. 2016. |
4. Model uncertainties perturbations | ||||
| Is model physics perturbed? If yes, briefly describe methods | Model uncertainty is now mainly sampled using the Stochastic Parameter Perturbation (SPP) scheme in which uncertain parameters and processes are perturbed with evolving random fields (Markov sequences) with specified spatial and temporal correlations. | multi-parameterization, stochastic perturbations, stochastic kinetic energy backscattering scheme | multi-parameterization, stochastic perturbations, stochastic kinetic energy backscattering scheme | multi-parameterization, stochastic perturbations, stochastic kinetic energy backscattering scheme |
| Do all ensemble members use exactly the same model version? | Yes | Yes | Yes | Yes |
| Is model dynamics perturbed? | No | No | No | No |
| Are the above model perturbations applied to the control forecast? | No | No | No | No |
5. Surface boundary perturbations | ||||
| Perturbations to sea surface temperature? | No | No | No | No |
| Perturbation to soil moisture? | No | No | No | No |
| Perturbation to surface stress or roughness? | No | No | No | No |
| Any other surface perturbation? | No | No | ||
| Are the above surface perturbations applied to the Control forecast? | N/A | N/A | N/A | N/A |
| Additional comments | ||||
6. Other details of the models | ||||
| Description of model grid | Arakawa-C grid | Arakawa-C grid | Arakawa-C grid | Arakawa-C grid |
| What kind of large scale dynamics is used? | Implicit semi-Lagrangian | Implicit semi-Lagrangian | Implicit semi-Lagrangian | Implicit semi-Lagrangian |
| What kind of boundary layer parameterization is used? | 1.5 order closure E-L | 1.5 order closure E-L | 1.5 order closure E-L | 1.5 order closure E-L |
| What kind of convective parameterization is used? | Deep convection: Updated version of Kain and Fritsch scheme (Kain and Fritsch, 1990 and 1993); Shallow convection: A non-precipitating mass-flux scheme based on Bechtold (2001). | Kain-Fritsch + Kuo-type | Kain-Fritsch + Kuo-type | Kain-Fritsch + Kuo-type |
| What kind of large-scale precipitation scheme is used? | Sundqvist et al (1989) | Sundqvist et al (1989) | Sundqvist et al (1989) | Sundqvist et al (1989) |
| What cloud scheme is used? | Sundqvist + Kain-Fritsch + Kuo-type | Sundqvist + Kain-Fritsch + Kuo-type | Sundqvist + Kain-Fritsch + Kuo-type | Sundqvist + Kain-Fritsch + Kuo-type |
| What kind of land-surface scheme is used? | ISBA | ISBA | ISBA | ISBA |
| How is radiation parametrized? | Li and Barker (2005) | Li and Barker (2005) | Li and Barker (2005) | Li and Barker (2005) |
| Other relevant details? | ||||
7. Re-forecast configuration | ||||
| Number of years covered | 20 past years (2001-2020) | 20 past years (1998-2017) | 20 past years (1998~2017) | 20 past years (1995~2014) |
| Produced on the fly or fix re-forecasts? | On the fly | On the fly | On the fly | On the fly |
| Frequency | Produced on the fly once a week to calibrate the Thursday 00Z real-time forecasts. The re-forecasts consist of a 4-member ensemble starting the same day and month as the Thursday real-time forecasts for the 20 years of 2001-2020. | Produced on the fly once a week to calibrate the Thursday 00Z real-time forecasts. The re-forecasts consist of a 4-member ensemble starting the same day and month as the Thursday real-time forecasts for the 20 years of 1998-2017. | Produced on the fly once a week to calibrate the Thursday 00Z real-time forecasts. The re-forecasts consist of a 4-member ensemble starting the same day and month as the Thursday real-time forecasts for the 20 years of 1998-2017. | Produced on the fly once a week to calibrate the Thursday 00Z real-time forecasts. The re-forecasts consist of a 4-member ensemble starting the same day and month as the Thursday real-time forecasts for the 20 years of 1995-2014. |
| Ensemble size | 4 members | 4 members | 4 members | 4 members |
| Initial conditions | ERA5 + Land surface initial condition from an off-line run of the surface prediction system (SPS) cycle driven by near-surface atmospheric ERA5 reanalysis and its associated precipitation. ORAS5 ocean initial conditions (0.25 degree). The sea ice is initialized with digitized sea ice charts from the Canadian Ice Service (CIS) and the HadISST2.2. | ERA interim + Land surface initial condition from an off-line run of the surface prediction system (SPS) cycle driven by near-surface atmospheric ERA-interim reanalysis and its associated precipitation. ORAS5 ocean initial conditions (0.25 degree). The sea ice is initialized with digitized sea ice charts from the Canadian Ice Service (CIS) and the HadISST2.2. | ERA interim + Land surface initial condition from an off-line run of the surface prediction system (SPS) cycle driven by near-surface atmospheric ERA-interim reanalysis and its associated precipitation | ERA interim + Land surface initial condition from an off-line run of the surface prediction system (SPS) cycle driven by near-surface atmospheric ERA-interim reanalysis and its associated precipitation |
| Is the model physics and resolution the same as for the real-time forecasts | Yes | Yes | Yes | Yes |
| If not, what are the differences | ||||
| Is the ensemble generation the same as for real-time forecasts? | No | No | No | No |
| If not, what are the differences | homogenous and isotropic perturbations using the algorithm from the ENKF | homogenous and isotropic perturbations using the algorithm from the ENKF | homogenous and isotropic perturbations using the algorithm from the ENKF | homogenous and isotropic perturbations using the algorithm from the ENKF |
8. References
- Bechtold, P., E. Bazile, F. Guichard, P. Mascart and E. Richard, 2001: A mass-flux convection scheme for regional and global models. Q. J. R. Meteorol. Soc., 127, 869-886.
- Bélair, S., L.-P. Crevier, J. Mailhot, B. Bilodeau, and Y. Delage, 2003a: Operational implementation of the ISBA land surface scheme in the Canadian regional weather forecast model. Part I: Warm season results. J. Hydromet., 4, 352-370.
- Bélair, S., R. Brown, J. Mailhot, B. Bilodeau, and L.-P. Crevier, 2003b: Operational implementation of the ISBA land surface scheme in the Canadian regional weather forecast model. Part II: Cold season results. J. Hydromet., 4, 371-386.
- Brasnett, B. 1999: A global analysis of Snow Depth for Numerical Weather Prediction. J. Appl. Meteor., 38, 726-740.
- Gagnon, N., and Co-authors, 2014: Improvements to the Global Ensemble Prediction System (GEPS) reforecast system from version 3.1.0 to version 4.0.0. Canadian Meteorological Centre Technical Note. [Available on request from Environment Canada, Centre Météorologique Canadien, division du développement, 2121 route Transcanadienne, 4e étage, Dorval, Québec, H9P1J3 or via the following web site: http://collaboration.cmc.ec.gc.ca/cmc/CMOI/product_guide/docs/changes_e.html#20141118_geps_4.0.0
- Giard D. and Bazile E. 2000: Implementation of a new assimilation scheme for soil and surface variables in a global NWP model. Mon. Wea. Rev. 128, 997-1015.
Kain, J. S. and J. M. Fritsch, 1990: A one-dimensional entraining / detraining plume model and its application in convective parameterization. J. Atmos. Sci., 47, 2784-2802.
Kain, J. S. and J. M. Fritsch, 1993: Convective parameterization for mesoscale models: The Kain-Fritsch scheme. The representation of cumulus convection in numerical models. Meteor. Monogr., 27, Amer. Meteor. Soc., 165-170.
- Lin H, N. Gagnon, S. Beauregard, R. Muncaster, M. Markovic, B. Denis and M. Charron, 2016: GEPS based Monthly Prediction at the Canadian Meteorological Centre. Mon. Wea. Rev., DOI: http://dx.doi.org/10.1175/MWR-D-16-0138.1
- Noilhan, J. and S. Planton, 1989: A simple parameterization of land surface processes for meteorological models. Mon. Wea. Rev., 117, 536-549