Optimization of Stochastic Total Tendency Perturbation Scheme

A Stochastic Total Tendency Perturbation (STTP) scheme, previously known as Stochastic Perturbation Scheme (SPS), was implemented into the Global Ensemble Forecast System (GEFS) at the National Centers for Environmental Predictions (NCEP) on Feb. 23, 2010. The scheme represents the uncertainty associated with the forecast model itself by perturbing the total tendency of the model for all model variables. To control the size and temporal variation of these perturbations, the current version of STTP employs a simple function with prescribed parameters to determine a coefficient which varies with latitude, season and forecast lead time.

Previous studies demonstrated that STTP can significantly increase ensemble spread, reduce systematic error and improve ensemble based probabilistic forecasts. It is also expected that STTP will continue to contribute to forecast improvement when the NWP model used is improved and the horizontal/vertical resolution is increased. Some adjustments to the STTP parameters will be necessary to maximize its benefit with these changes in the ensemble configuration. The NWP model used in GEFS, namely the Global Forecast System (GFS), was greatly improved with the July 29, 2010 implementation, and will be used in a new GEFS configuration to be implemented in 2011. The upgraded GEFS will be running at a resolution (T254L42) higher than the current setting of T190L28 for the first 192 hours of integration.

This article focuses on the optimization of STTP scheme in the planned GEFS configuration. Extensive experiments will be conducted with the upgraded GFS model and increased resolution to determine the optimal STTP structure and parameters. An optimal setting of STTP parameters will be determined by systematically comparing the results of the experiments.