Data assimilative modeling and OSSEs experiments for the monitoring of the Loop Current frontal dynamics—Application to the Deepwater Horizon oil spill

A methodology to develop integrated observational and modeling systems in the Gulf of Mexico (GoM) using Observing System Simulation Experiments (OSSEs) is presented. The example will focus on the predictability of Loop Current (LC) dynamics. The LC is the connectivity corridor of water masses (and associated fish larvae, and nutrients, but also pollutants) from the Yucatan to the Florida Straits, through the GoM interior, where it has a varying northern extension. When fully extended, it sheds a large and warm anticyclonic eddy, called a Loop Current Eddy (LCE), which then drifts westward inside the GoM, while the LC retracts. This LCE shedding involves interactions between the LC and surrounding mesoscale cyclonic eddies, developing at the edge of the LC. The complexity of these interactions have been highlighted during the Deepwater Horizon oil spill event, during which pollutants have been advected southward by a small frontal cyclonic eddy, before being trapped in Eddy Franklin, the LCE that was formed late May and fully detached in early July, 2010.

A long term, high resolution, atmospherically forced ocean model simulation has been set up to study the LC variability induced by frontal mesoscale cyclonic eddies, especially during LCE formation and shedding. The GoM-HYCOM is set at a resolution of 1/25o as free-running and of 1/50o as data assimilative. Two data assimilation (DA) schemes, one based on an optimal interpolation (OI), and a more advanced Singular Evolutive Extended Kalman (SEEK) filter, have been set up to integrate observations within the model. OSSEs are used to assess the impact of in situ or remotely sensed observations on the representation of the LC frontal dynamics, as well as the relative performances of the chosen DA schemes. In addition, data assimilative simulations before and after the Deepwater Horizon oil spill event, involving real and simulated observations, will facilitate the design of observation networks suitable for the monitoring of LC variability and its associated connectivity and transport.