|dc.description.abstract||The circulation of the Gulf of Mexico is controlled by presence of large mesoscale structures (10-500 km). We investigate its variability and predictability from interannual to intraseasonal time scales, and the dynamical interactions between physical circulation and biological productivity. We do so by analyzing an ensemble of numerical integrations using the Regional Ocean Modeling System and hydrographic and biogeochemistry observations collected during summer field campaigns in 2010, 2011, and 2012.
First, we explore the potential relationships and linkages between Mississippi-Atchafalaya River runoff, nutrient loads, and ocean dynamics from our field data. A negative correlation between nutrient concentration and salinity was confirmed at the surface and in the upper 60m of the water column for nitrite, nitrate, phosphate and silicate. No major changes in the nutrient concentrations were found between our data and previous measurements from twenty years ago. The biological activity in the stations sampled (northern Gulf) is nitrogen limited in 79% of them and phosphorus limited in 8%. Besides the direct input of nutrients from river discharges, the distribution of nutrients in intermediate and high salinity waters in the euphotic layer is influenced by dynamical processes at the ocean mesoscales such as eddies, coastal upwelling events and Loop Current (LC) intrusions. Then, using an ensemble of four model integrations we investigate how mesoscale motions dominate the variability of the Gulf of Mexico circulation both at the surface and in deep waters on intraseasonal time scales. We focus on its predictability by exploring the impact of small variations in the initial conditions and the role of the boundary conditions in the circulation evolution. In all runs, the model provides a good representation of the mean circulation features. However, the shedding of the Loop Current Eddies (LCE) differs in each run considered, and our analysis shows that the detachment of the LCE is a stochastic process. We show that the interannual variability at the model boundaries affects the representation of the LC strength and of the Yucatan Channel transport. On the other hand, the circulation in the LATEX Shelf, TAVE Shelf, and Bay of Campeche is insensitive to the details of the model boundaries, is not affected by the LC, but depends only on the wind variability, and it is therefore predictable if the atmospheric conditions are known. On the contrary, the circulation in the central basin is affected by the LC extension and by the Rings, and dominated by mesoscale features. In most of the basin, mesoscale features are coherent in the top ~ 1000 m of the water column, and below it, but not correlated between the surface and the deep layer. Coherency throughout the whole water column is attributed to particular topographic features such as the south-west corner of the Sigsbee Deep. The chaotic behavior associated with the propagation of the LCE and the elevated mesoscale activity restricts the predictability of the system at intra-seasonal scales to the coastal areas. In consequence, assimilation of continuous in-situ measurements is necessary to insure good hindcasts and forecasts at surface and below 1000 m depth.
Finally, since mesoscale activity is key to understand the horizontal and vertical dynamics in the Gulf, we further analyze the model representation of mesoscale circulation under low (monthly) and high (6 hourly) frequency atmospheric forcing. The temporal scale variation from monthly to 6-hourly in the wind forcing impacts the timing of horizontal dynamics, but not the strength. However, high frequency winds impact the model representation of vertical transport that increases as the temporal resolution of the forcing increases. Vertical velocities in the simulation forced by 6-hourly winds are ten times greater than the one obtained using monthly averaged winds. The energy injected by the winds into the ocean is transported in the water column by mesoscale eddies and near-inertial oscillations. If the forcing used by the model does not resolve the inertial frequency (1.4 days in the Gulf), then vertical transport processes are underestimated. Those processes are particularly important for the model representation of biological activity in the ocean upper layers, since they contribute to the input of nutrients into the euphotic zone.||