Home > GFS Ensemble bias-corrected datasets documentation

Ensemble forecasts attempt to address the uncertain nature of NWP.

Both initial condition errors and errors in the formulation of the physics forcing are compounded in NWP non-linear models, degrading the accuracy of forecasts. To account for increase uncertainty from growing instabilities, the NCEP operational global ensemble modeling suite, Global Ensemble Forecast System (GEFS), utilizes the techniques of Bred Vector and Ensemble Transform to generate initial perturbations and Stochastic Total Tendency Perturbation (STTP) scheme to account for model errors.

The current system, effective 2012FEB, produces 80 perturbation forecasts per day, 20 per cycle, at T254L42 resolution for the 8 days, and T190L42 from Day 8 to Day 16. The model output presented in this data base is at one degree latitude by one degree longitude grid, and products on a half by half degree mesh for 0 to 192 hours are under development.

Bias corrected products are produced from the NCEP global ensemble and from the NAEFS ensemble, which includes the NCEP global ensemble in all cycles and the CMC global ensemble at 00Z and 12Z.

Downscaled forecasts on the NDGD are produced from the bias-corrected NAEFS ensemble forecasts from NCEP and CMC.