State estimates and forecasts of the loop current in the Gulf of Mexico using the MITgcm and its adjoint

TitleState estimates and forecasts of the loop current in the Gulf of Mexico using the MITgcm and its adjoint
Publication TypeJournal Article
Year of Publication2013
AuthorsGopalakrishnan G, Cornuelle BD, Hoteit I, Rudnick DL, W. Owens B
JournalJournal of Geophysical Research: Oceans
Volume118
Pagination3292-3314
Date Published2013/07
ISBN Number2169-9291
Keywords4219 Continental shelf and slope processes; 4220 Coral reef systems; 4260 Ocean data assimilation and reanalysis; 4262 Ocean observing systems; 4263 Ocean predictability and prediction; Gulf of Mexico; hindcast and forecast; loop current; MITgcm; ocean modeling; state estimation
Abstract

An ocean state estimate has been developed for the Gulf of Mexico (GoM) using the MIT general circulation model and its adjoint. The estimate has been tested by forecasting loop current (LC) evolution and eddy shedding in the GoM. The adjoint (or four-dimensional variational) method was used to match the model evolution to observations by adjusting model temperature and salinity initial conditions, open boundary conditions, and atmospheric forcing fields. The model was fit to satellite-derived along-track sea surface height, separated into temporal mean and anomalies, and gridded sea surface temperature for 2 month periods. The optimized state at the end of the assimilation period was used to initialize the forecast for 2 months. Forecasts explore practical LC predictability and provide a cross-validation test of the state estimate by comparing it to independent future observations. The model forecast was tested for several LC eddy separation events, including Eddy Franklin in May 2010 during the deepwater horizon oil spill disaster in the GoM. The forecast used monthly climatological open boundary conditions, atmospheric forcing, and run-off fluxes. The model performance was evaluated by computing model-observation root-mean-square difference (rmsd) during both the hindcast and forecast periods. The rmsd metrics for the forecast generally outperformed persistence (keeping the initial state fixed) and reference (forecast initialized using assimilated Hybrid Coordinate Ocean Model 1/12° global analysis) model simulations during LC eddy separation events for a period of 1̃2 months.

DOI10.1002/jgrc.20239
Integrated Research Themes: 
Student Publication: 
No
sharknado