Review of Recent Advances in Dynamical Extended Range Forecasting for the Extratropics

Abstract

Within current definitions, Dynamical Extended Range Forecasting (DERF) refers to numerical forecasts performed beyond the medium range until all dynamical influence of the initial conditions has been lost. The numerical approach to extended range predictions was formulated more than 20 years ago by Smagorinsky (1969) under the name “Deterministic Extended Range Forecasting”. DP Baumhefner (personal communication) has suggested that the term “Numerical Extended Range Forecasting” would be preferable, but for this review we will use the expression “Dynamic Extended Range Forecasting” (DERF), since it has already been used in several publications. The current use of the adjective “dynamical” rather than “deterministic”, or “numerical”, puts more emphasis on the dynamical mechanisms by which the initial conditions can influence the time averaged properties of the flow (Shukla, 1981), and is an implicit recognition that, while the model used is deterministic, the forecasts contain a stochastic element due to the growth of the initial uncertainties. Initial conditions and boundary conditions need to be specified in order to be able to integrate a numerical model. The DERF can be considered as a transition between the two extreme kinds of predictions: deterministic short range prediction where the influence of the initial conditions is dominant, and climate prediction which is by its nature a statistical concept and where the key role is played by the boundary conditions.