Impact of GODAE products on nested HYCOM simulations of the West Florida Shelf

Abstract

Nested non-assimilative simulations of the West Florida Shelf for 2004–2005 are used to quantify the impact of initial and boundary conditions provided by Global Ocean Data Assimilation Experiment ocean products. Simulations are nested within an optimum interpolation hindcast of the Atlantic Ocean, the initial test of the US Navy Coupled Ocean Data Assimilation system for the Gulf of Mexico, and a global ocean hindcast that used the latter assimilation system. These simulations are compared to one that is nested in a non-assimilative Gulf of Mexico model to document the importance of assimilation in the outer model. Simulations are evaluated by comparing model results to moored Acoustic Doppler Current Profiler measurements and moored sea surface temperature time series. The choice of outer model has little influence on simulated velocity fluctuations over the inner and middle shelf where fluctuations are dominated by the deterministic wind-driven response. Improvement is documented in the representation of alongshore flow variability over the outer shelf, driven in part by the intrusion of the Loop Current and associated cyclones at the shelf edge near the Dry Tortugas. This improvement was realized in the simulation nested in the global ocean hindcast, the only outer model choice that contained a realistic representation of Loop Current transport associated with basin-scale wind-driven gyre circulation and the Atlantic Meridional Overturning Circulation. For temperature, the non-assimilative outer model had a cold bias in the upper ocean that was substantially corrected in the data-assimilative outer models, leading to improved temperature representation in the simulations nested in the assimilative outer models.

Appendix: Vertical coordinate selection

The first application of HYCOM as a coastal model (Winther and Evensen 2006) demonstrated that the model could produce realistic circulation and water mass structure in shallow water regions. They used the standard approach for coastal regions of allowing the offshore isopycnic and nearsurface level p coordinates to transition to σ coordinates over shallow water. The classic seamount problem (e.g., Beckmann and Haidvogel 1993) is used in this study to demonstrate the superiority of using p instead of σ coordinates in coastal regions with sloping topography.

The seamount domain was set up in a 360 × 360-km f-plane box, with f set to its value at 30° N, and uses the seamount structure of Shchepetkin and McWilliams (2003):

$$h\left( {x,y} \right) = H_0 - H\exp \left( {\frac{{ - \left( {x^2 + y^2 } \right)}}{{L^2 }}} \right)$$

where L = 5,000 m, H = 4,500 m, and L = 40 km. The continuous initial density profile is exponential and roughly representative of summer density profiles observed in the subtropical Atlantic. The Burger number is ∼3, an intermediate value in the range of cases considered in earlier seamount tests (e.g., Beckmann and Haidvogel 1993; Mellor et al. 1998; Shchepetkin and McWilliams 2003). Twenty-two vertical layers were used for both the p and σ coordinate cases. To initialize model fields, the density value of each layer was first assigned as a function of central layer depth based on the initial continuous density profile. To assign initial T and S values to each layer, an exponential temperature profile roughly representative of the summer subtropical Atlantic was used to assign T values at central layer depths; then, S values were calculated using the model equation of state. Initial cross-sections of density for both vertical coordinate choices are presented in Fig. 12. In addition to the two cases run with the existing Montgomery potential formulation of the pressure gradient force (MP), two additional cases were run implementing the pressure gradient formulation used in the ROMS ocean model (Shchepetkin and McWilliams 2003).

Fig. 12

Initial density field and vertical coordinates for the p and σ coordinate seamount tests (left). Temporal evolution of kinetic energy over the first day of integration for the four seamount tests

For each of the four cases, the model was run unforced for 24 h. The resulting currents resulted from errors in the pressure gradient force, and the effects of these errors were monitored by graphing the total kinetic energy as a function of time (Fig. 12). The most rapid increase in KE occurs for the σ coordinate, MP case. The rate of increase was much smaller for the p coordinate, MP case because pressure gradient errors at any grid point are confined to the deepest layer with non-zero thickness that intersects a sloping bottom while errors exist in shallower σ coordinate layers. A similar scenario is observed for the two ROMS cases, with the smallest rate of KE increase occurring in the p coordinate case. These results demonstrate the superiority of using p coordinates over sloping topography in HYCOM. Ideally, the ROMS pressure gradient formulation should also be used, but there is a significant problem in the interior isopycnic coordinate domain. If a level isopycnic layer intersects a sloping bottom, there should be zero pressure gradient force, and this is achieved with high accuracy by the MP formulation. However, this situation produces adjacent grid points where the sloping bottom is shallower than the level interface at the bottom of the layer, resulting in a change of central layer depth between the grid points. Given the constant density in this layer, the ROMS formulation detects a sloping density interface and produces a non-zero pressure gradient force where none should exist. Tests demonstrated that this problem more than nullified the improvements produced by the ROMS formulation in the non-isopycnic coordinate domain (not shown). Therefore, the MP formulation was retained and p coordinates used for the nested coastal simulations in this study. This is a reasonable choice because the rate of KE increase in the seamount tests was comparable to the rate of increase for the ROMS formulation used with σ coordinates (Fig. 12).