This repository contains the MITgcm model setup and analysis scripts that were used for the paper Submesoscale Vertical Velocities Enhance Tracer Subduction in an Idealized Antarctic Circumpolar Current, Balwada, Smith and Abernathey 2018, GRL (https://doi.org/10.1029/2018GL079244).
The model setup is similar to the one in Abernathey and Cessi 2014 (https://github.com/rabernat/channel_topography), and details are present in our paper and supplementary material. The basic setup is a zonally periodic channel, with walls to the north and south. The zonal symmetry is broken by the presence of a meridional ridge. The forcing is only at the surface with winds and buoyancy restoring.
The major updates from the previous studies were to the subgrid numerics, with the new setup using Leith style parameterizations same as the ones used in the LLC4320 setup. These numerics promise to be grid aware, and changing/tuning parameters with resolution is supposedly not required. We did find that at the highest resolutions (1km) the model solution generated a lot of internal waves, and it is unclear if these waves are "realistic" (generated by loss of balance at boundaries) or an artifact of the subgrid scheme. We also changed the wind profile from sin(y) to sin^2(y) profile, as the new profile has a curl that goes to zero at the boundaries. However the influence of this change was modest.
In addition to the physical model setup, we forced a passive tracer with a surface source. This tracer was forced to be representative of atmospheric gasses that dissolve in the surface waters, and are then propogate into the interior via ventilation/subduction. The time characteristics of the forcing are similar to the step change in atmospheric concentrations, with an infinite amount of tracer - so subduction in the ocean does not influence the forcing.
The main analysis correspond to the follows:
- Basic descriptive plots; mean flows, EKE etc.
- Tracer budget analysis - how much tracer changes in a control, and what processes (advection or diffusion) at the boundaries influence this.
- Spectral analysis - we calculate energy spectra, and flux cross spectra using the https://github.com/xgcm/xrft package.
The 3D figures were made using paraview, and scripts are not provided here.