Layer based ocean models like the MICOM (Miami Isopycnic Coordinate Ocean Model) family are intrinsically well adapted to the study of water mass development and spreading because of their capacity to conserve properties in baroclinic advective systems with minimal inter-layer property mixing related to model numerics. In the limit of small or vanishing diapycnic mixing, the property evolution in individual coordinate layers becomes a two-dimensional advective-diffusive boundary value problem, albeit one where the boundary conditions as well as the physical position of the boundary are generally time dependent. The thermodynamically quiescent interior pycnocline regime interacts along this mobile boundary with a domain dominated by the destruction and reestablishment of the seasonal thermocline, i.e., seasonal evolution and retreat of the surface mixed layer (ML).
From the perspective of a layer based model, the expanding and contracting high latitude ML regime is like a pumping heart, sucking in and reinjecting waters from and into the permanent pycnocline. The resulting net ventilation effect is then closely tied to the baroclinic transport patterns across its maximally extended boundary. The MICOM ML evolution is governed by entrainment/detrainment processes subject to property conservation constraints and potential energy budget considerations. These work in an intuitively simple way during periods of active ML deepening, but less so during periods of restratification.
Briefly put, as the mixed layer density decreases, new interior layers are created in discrete steps, dictated by the model coordinate discretization. The imposed consistency with the latter excludes accounting for "fossil" mixed layers of intermediate density. The mass potentially left behind in such layers is therefore in each model time step implicitly distributed between the uppermost interior layer and the newly formed mixed layer, until such time when the density of the later matches the next prescribed interior layer. Each newly formed layer is thus gradually inflated until the ML density reaches the next coordinate value, and the inflation of a new layer commences. This process of gradual layer inflation is potentially significant for the reinjection process of waters from the seasonally convective region into the permanent pycnocline domain, since it allows for intermittent layer thickness (potential vorticity) matching at the domain boundary.
Absent ambient advection and eddy diffusion effects, the one-dimensional ML evolution in response to cyclic forcing is well known to result in greater convective penetration in each consecutive cycle, until the bottom is reached. To balance this secular deepening trends in a poleward bounded basin, a convergent mass transport is required in the uppermost surviving interior layer, and a corresponding mean annual divergence must be provided in the time variable mixed layer environment. An effective diagnostic relevant to this process is in MICOM provided by the mean annual transfer rate across the lower ML boundary, since mere vertical displacements of the latter in response to the seasonal cooling and restratification processes does not contribute to it. We find in an ongoing long term integration of a global MICOM version that at one degree resolution entrainment into the ML is in the northwest Atlantic concentrated along the deep water outflow path at the western boundary, where the deep Isopycnic layers shoal for dynamic reasons, while at one eight degree resolution a distributed entrainment process is apparently associated with mesoscale eddy activity. It appears thus that the high resolution model representation of the latter during intensive surface buoyancy flux forcing needs careful attention.
ACKNOWLEDGMENTS:
Work supported under DOE contract GEFG05-94ER61943: Development and Evaluation of the Global Version of the Miami Isopycnic Coordinate Model.