Application of a hybrid EnKF-OI to ocean forecasting

[摘要] Data assimilation methods often use an ensemble to represent the backgrounderror covariance. Two approaches are commonly used; a simple one with astatic ensemble, or a more advanced one with a dynamic ensemble. The latteris often non-practical due to its high computational requirements. Somerecent studies suggested using a hybrid covariance, which is a linearcombination of the covariances represented by a static and a dynamicensemble. Here, the use of the hybrid covariance is first extensively testedwith a quasi-geostrophic model and with different analysis schemes, namelythe Ensemble Kalman Filter (EnKF) and the Ensemble Square Root Filter (ESRF).The hybrid covariance ESRF (ESRF-OI) is more accurate and more stable thanthe hybrid covariance EnKF (EnKF-OI), but the overall conclusions are similarregardless of the analysis scheme used. The benefits of using the hybridcovariance are large compared to both the static and the dynamic methods witha small dynamic ensemble. The benefits over the dynamic methods becomenegligible, but remain, for large dynamic ensembles. The optimal value of thehybrid blending coefficient appears to decrease exponentially with the sizeof the dynamic ensemble. Finally, we consider a realistic application withthe assimilation of altimetry data in a hybrid coordinate ocean model (HYCOM)for the Gulf of Mexico, during the shedding of Eddy Yankee (2006). A10-member EnKF-OI is compared to a 10-member EnKF and a static method calledthe Ensemble Optimal Interpolation (EnOI). While 10 members seem insufficientfor running the EnKF, the 10-member EnKF-OI reduces the forecast errorcompared to the EnOI, and improves the positions of the fronts.

[学科分类] 海洋学与技术