Preliminary Results for Mesocale Ocean Forecast Model Ensembling in the Gulf of Mexico

The ocean circulation of the Gulf of Mexico is dominated by the state of the Loop Current; e.g., its location, strength, and eddy-shedding condition. {The Loop Current is the section of the Gulf Stream System that enters the Gulf through the Yucatan Channel and exits through the Straits of Florida.} More or less annually, the Loop Current sheds an enormous ca. 400 km diameter anticyclone that extends through the water column and translates westward to the Mexican coast over the course of a year or so. It also sheds order-of-magnitude smaller, and more frequent, cyclones in the upper ocean. The above mesoscale features, and their associated ocean fronts lead to near-surface current jets strong enough (ca. 1 m/s) to impact the efficacy and safety of offshore oil & gas recovery operations and facilities. Conversely, the advective and turbulent transports of chemical spills and ambient biota are strongly dependent on the space-time structures of the above mesoscale variability. Hence, there is considerable motivation to establish an operational (several-month) mesoscale ocean forecast circulation model for the Gulf.

From prior observational and modeling studies, the Gulf of Mexico circulation is understood to be influenced by large scale and synoptic scale atmospheric forcing, variability in the Loop Current throughflow, complex bottom topography, and, to a lesser extent, river runoff and tides. The skill of a forecast model depends, i n part, on the quality of initial and boundary conditions provided to the model. However, the observed initial and boundary conditions are generally incomplete and otherwise uncertain. There are several possible probabilistic approaches; e.g., multi-model ensembling, single-model ensembling, or a combination of those two ensembling approaches, for estimating the ensemble mean and variance. {The multi-model approach has been tried previously with mixed results.} Here, the exclusive focus is on a single-model approach using the Naval Research Laboratory's Naval Coastal Ocean Model (NCOM), which has been used in global and regional forecasts for a decade, in an implementation for the Gulf of Mexico with a horizontal resolution of ca. 3 km and a vertical representation of ca. 40 vertical levels . Atmospheric forcing is provided by the Navy's operational atmospheric regional mesoscale model (COAMPS) and open boundary conditions are taken from the Navy's operational, global implementation of NCOM. The Gulf of Mexico model also assimilates satellite sea surface heights & sea surface temperatures, and vertical temperature & salinity profiles when and where available in real-time.

Preliminary results for an ongoing experiment running for at least a year of validation indicate useful skill in forecasting an anticyclonic eddy separation event with a forecast horizon of up to 2 mos. However, this preliminary result must be recognized as being based on only one eddy-shedding event, and the validation period needs to be extended for ca. 10 yrs or longer for statistical significance.