As a starting point for discussion on setting up the first idealized numerical model intercomparison experiment, here (as PDF) is a document written by John Hunter describing the proposed model configuration and experiment. Comments are welcome.
Additional ODU experiments using different "horizontal" mixing schemes found here.
Based on the concern for the treatment of the pressure gradient at the ice-ocean interface (see document written by Adrian Jenkins), two sub-experiments have been proposed. The first is to use a homogeneous ocean by repeating Experiment 1 (i.e., same initial hydrographic properties) and no buoyancy exchanges at the ice-ocean interface. The second is to uniformly stratify the ocean (again no buoyancy exchanges). The temperature is everywhere -1.9 ºC while the salinity varies linearly from 34.4 psu at the 200 m depth horizon to 34.8 at 900 m. It is expected that the ocean should not move (much!) in either of these cases.
As a follow-on experiment, we may consider a simpler cavity than in Experiment 1 -- namely, flat bottom, flat ice shelf base and buoyancy input/output PRESCRIBED (i.e. rather that being derived from the temperature/salinity structure at the ice/ocean interface). A particularly simple model is a rectangular cavity, with constant depth and ice draft, with buoyancy injected as a line source at one end and removed as a line sink at the other. This would be a useful diagnosis of problems if we find that the models significantly disagree when applied to the Experiment 1 configuration, described above. It may even have an analytic solution.
Currently, we are also seeking to develop an experimental configuration that is amenable to an analytical solution technique. This would provide a known solution against which to intercompare the numerical model solutions. Suggestions on domains suitable for analytical solutions (and the actual analytical solutions!) are welcome.
|
. © David Holland. All Rights Reserved. |
|
If you would like further information concerning any of the above topics please send email . |
