Search

The highlights for the manuscript, "Ocean Drifter Velocity Data Assimilation, Part 2: Forecast Validation", include: • 3DVAR velocity data assimilation within an operational prediction system. • Eularian velocities inferred from surface drifter position observations. • Velocity assimilation reduces temperature and salinity forecast error. • Including velocity observations improves the position of fronts and eddies. A large deployment of drifters conducted during August-December, 2020 in the Gulf of Mexico offers a test bed for a data assimilation system developed specifically to include velocity observations. This updated Navy Coupled Ocean Data Assimilation system employs the three-dimensional variational approach and is described in part one of this two-part paper (Helber et al, 2023). In this paper, we examine the impact of velocity data assimilation on the ensuing forecasts of the ocean state including not only velocity but also temperature and salinity fields below the surface. Two high-resolution (1 km) experiments were performed in the Gulf of Mexico; one with velocity data assimilation and the other without. The resulting 48 h forecasts of temperature, salinity, and velocity are examined and compared relative to the observations being assimilated (including the inferred velocities from the drifters) and unassimilated observations of temperature, salinity, and velocity from two gliders near the drifters. In addition, we assess eddy positioning and Lagrangian trajectory separation. Comparisons of these two experiments, with and without velocity data assimilation, suggest that adding velocity observations to the assimilation increases skill in predicting velocity and the subsurface temperature and salinity. [ABSTRACT FROM AUTHOR]

Ocean prediction systems rely on an array of assumptions to optimize their data assimilation schemes. Many of these remain untested, especially at smaller scales, because sufficiently dense observations are very rare. A set of 295 drifters deployed in July 2012 in the north-eastern Gulf of Mexico provides a unique opportunity to test these systems down to scales previously unobtainable. In this study, background error covariance assumptions in the 3DVar assimilation process are perturbed to understand the effect on the solution relative to the withheld dense drifter data. Results show that the amplitude of the background error covariance is an important factor as expected, and a proposed new formulation provides added skill. In addition, the background error covariance time correlation is important to allow satellite observations to affect the results over a period longer than one daily assimilation cycle. The results show the new background error covariance formulations provide more accurate placement of frontal positions, directions of currents and velocity magnitudes. These conclusions have implications for the implementation of 3DVar systems as well as the analysis interval of 4DVar systems. [ABSTRACT FROM AUTHOR]