Predictability of upper ocean dynamics at SWOT scales

Abstract:

Ocean processes that evolve and decorrelate on timescales much faster than the observation sampling time are effectively random and unpredictable. For the SWOT mission, this includes submesoscale processes, which evolve on timescales of days, and unbalanced internal tides and other internal gravity waves, with periods of minutes to hours. Sea-surface height variations due to internal waves are a particular challenge for SWOT observations because their spatial scales overlap with those of balanced motions at submesoscales and so can dominate the SSH signal and derived quantities such as geostrophic velocities and surface vorticity.

In this work, we examine novel stochastic data assimilation techniques that seek to address two key obstacles: filtering geostrophic currents from unbalanced internal wave motions, and interpolation of high-spatial-resolution but low-temporal-resolution satellite measurements of the surface signal. The approach taken combines simulated satellite observations with an efficient stochastic parameterization for the fast timescales. The surface signal is modelled as a superposition of slow geostrophic motions, fast-evolving internal wave motions, and incoherent turbulent noise with known spectra and dispersion relations. These contributions are filtered from the observations using standard Kalman filtering techniques, providing an estimate of the slowly evolving geostrophic component. The technique is demonstrated in numerical simulations driven by climatological shear and stratification profiles and realistic satellite sampling.