Seasonal Forecasts Based on Numerical Weather Prediction Models

Introduction

Since September 1995 Environnement Canada (EC)) has been producing seasonal forecasts of surface air temperature (SAT) anomalies and precipitations (PCPN) anomalies (1-3 month outlooks) for Canada from the Meteorological Service of Canada. The seasonal forecast results from an ensemble of 40 model runs, with 10 run of each of the following models: the Climate version of the Global Environmental Multiscale model (GEM-CLIM), the second generation of the Atmospheric General Circulation Model (AGCM2), the third generation of the Atmospheric General Circulation Model (AGCM3) and the Spectral aux éléments finis (SEF) one.

• The GEM model (Côté et al. 1998) was developed at the Recherche en Prévision Numérique du temps (RPN). This model has a horizontal resolution of 2 degrees with 50 vertical levels.
• The AGCM2 (McFarlane et al. 1992) model, from the, (Canadian Centre for Climate Modelling and Analysis (CCCma), has an horizontal resolution of 625 km (T32) with 10 vertical levels.
• The AGCM3 (Scinocca et al. 2004), also from the CCCma uses an horizontal resolution of 315 (T63) with 32 vertical levels.
• The SEF model, developed at RPN was used in previous studies for global data assimilation and medium-range weather forecasting (Ritchie, 1991; Ritchie and Beaudoin, 1994). It is also a global spectral model, with an horizontal resolution of 210 km (T95) and 27 vertical levels.

All models use the same CMC atmospheric analyses. However, they differ in the way they use the analyzed surface fields. Surface forcing, crucial in controlling the seasonal atmospheric variability, has to be treated carefully. The numerical models used in the EC Seasonal Forecast System are forced by 3 fields: sea surface temperatures (SST), sea ice and snow cover. The treatment of these fields is done in the following way:

• For the GEM-CLIM model: Since no interactive ocean is used, the SST anomalies observed just prior the beginning of the forecast (preceding 30 days) are fixed throughout the forecast period and they are added to the evolving climatology. The sea ice extent analysis is relaxed toward the climatology during the first 15 days of the forecast period. The snow cover is a prognostic variable of the model initialised off from weekly observations.
• For the AGCM2 model: The treatment of SST anomalies is done in the same way as for the GEM model. The sea ice cover is climatological all along the numerical integrations. The snow cover is a prognostic variable of the model initialised off from weekly observations.
• For the AGCM3 model: The treatment of SST anomalies is done in the same way as for the GEM model. The sea ice extent analysis is relaxed toward the climatology during the first 15 days of the forecast period. The snow cover is a prognostic variable of the model initialised off from weekly observations.
• For the SEF model: The treatment of SST anomalies is done in the same way as for the GEM model. The sea ice extent and the snow cover analyses are relaxed toward the climatology during the first 15 days of the forecast period.

The climatic drift in the models is removed using their known climatology (see below for more explanations). The models climatology comes from the 30 year hindcasts of the seasonal Historical Forecasting Project (HFP). The HFP was also essential to estimate the expected forecast skill of the models (see discussion in Derome et al., 2000; Kharin and Zwiers, 2001; Kharin et al., 2001; Plante and Gagnon, 2000). The data from the HFP are available on line from the CCCma Web site.

Surface Air Temperature Forecast Methodology

The surface air temperature forecasts are made in doing first an average of the daily temperature as predicted by the models. The climatologies of the models are then subtracted from the mean forecast seasonal temperatures to derived the forecast anomalies of each model. The anomalies of the four models are then normalized and combined using an arithmetic average. The surface air temperature forecast anomalies are the anomalies of the mean daily temperature measured at the Stevenson screen height (2 metres). Finally the anomalies are divided in three categories (above, near and below the normal).

Precipitation Forecast Methodology

The precipitation forecasts are made using the total accumulated water precipitation over the season. The precipitation predicted by the models is the total liquid and includes all types: snow, rain, ice pellets, etc. The climatology of the models is subtracted from the total precipitation forecast to derive the anomalies. The anomalies of the four models are then combined using a simple normalized average as described for the surface air temperature. Finally the precipitation anomalies are divided in three categories (above, near and below the normal) as is done for the temperature anomaly forecast.

References

Côté, J., S. Gravel, A. Méthot, A. Patoine, M. Roch, and A. Staniforth, 1998: The operational CMC-MRB Global Environmental Multiscale (GEM) model: Part I - Design considerations and formulation, Mon. Wea. Rev., 126, 1373-1395. [paper]

Derome J., G. Brunet, A. Plante, N. Gagnon, G. J. Boer, F. W. Zwiers, S. J. Lambert, J. Sheng, et H. Ritchie, 2001: Seasonal Predictions Based on Two Dynamical Models.Atmos. Ocean., 39, 485-501. [paper] (Requires Acrobat Reader to view)

Kharin, V. V. et F. W. Zwiers, 2001: Skill as a function of time scale in ensemble of seasonal hindcasts. Climate Dynamics, 17, 127-141. [abstract]

Kharin, V.V ., F. W. Zwiers et N. Gagnon, 2001: Skill of seasonal hindcasts as a function of the ensemble size. Climate Dynamics, 17, 835-843. [abstract]

McFarlane, N.A., G.J. Boer, J.-P. Blanchet et M. Lazare. 1992: The Canadian Climate Centre second-generation general circulation model and its equilibrium climate. J. Climate, 5, 1013-1044. [abstract]

Plante A. et N. Gagnon, 2000: Numerical Approach to Seasonal Forecasting. In the "Proceedings of the sixth workshop on operational meteorology", Halifax, Novembre 1999, 162-165. (Requires Acrobat Reader to view)