High resolution ensemble forecasting for the Gulf of Mexico eddies and fronts

Stochastic forecasting is widely used in meteorology, in order to provide confidence indices of the prediction (Molteni et al., 1996). Stochastic forecasting is introduced by Yin and Oey (2007) for application to the GOM.

In a paper published in Ocean Dynamics Drs. Francois Counillon and Laurent Bertino present a sensitivity study to different perturbation systems is carried out by perturbing the initial state, lateral forcing, and atmospheric forcing. The perturbation of the initial state seems to control the positioning of the large-scale features, whereas the perturbation of the boundary conditions (lateral and atmospheric) controls the growth of instabilities.

The assimilation results indicate that a strong assimilation allows for higher initial accuracy, but diverges faster with the model integration. This implies that an optimal value can be chosen in order to obtain the maximum accuracy at a prescribed forecasting horizon. Finally, the ensemble spread of a 10-member dynamic ensemble appears to be correlated in space and time with the forecast error. This result indicates that even a small ensemble can provide a confidence index of the forecast.

The figure resolves; Overlay of model ensemble fronts (pink line) with the non-assimilated OC map (contour); for the nowcast on the 12th July. Blue color (resp. green) indicates low (resp. high) concentration of chlorophyll, and cloud covered areas are in white. The thick black line represents the front derived from SSH altimeter maps, and the red thick line is the ensemble mean.


Add comment

Login to post comments