Repository navigation
Conversation
|
Thanks @dougiesquire, this looks cool too. Maybe there could be something shared with |
Possibly?... I'm actually thinking I'll generalize the |
The boundary layer used for the melt parameterisation must match the depth over which the forcing is applied. The only controls for these are global: HMIX_SFC_PROP/HMIX_UV_SFC_PROP control the boundary layer for computing the parameterisation, and MINIMUM_FORCING_DEPTH controls the depth over which the melt is applied. We introduce new optional parameters to applyBoundaryFluxesInOut() to modify IforcingDepthScale to use a different `shelf_forcing_depth` according to the fraction of the cell covered by ice shelf.
This is the analogue to MINIMUM_FORCING_DEPTH, applied under ice shelves. It is only read when an ice shelf coverage field is provided to the diabatic driver initialisation routine.
29c655c to
a36db5d
Compare
This only affects the path when using the non-legacy ALE diabatic driver. In this case, the frac_shelf_h array is passed through, and used to apply a different forcing depth as required.
This adds HMIX_SHELF_PROP and HMIX_UV_SHELF_PROP, which are analogues to the `SFC` parameters. These determine the boundary layer depth for near-shelf temperature, salinity and velocities, which are used for calculation of melt parameterisation.
This now takes frac_shelf_h into account when computing depth_ml used for calculating the surface state.
a36db5d to
8a93ac8
Compare
|
Hi @angus-g and @claireyung, I've had a think about whether/how this relates to #72. My thinking now is that they're just serving different purposes and we should keep them separate (at least for now). My understanding is that this PR is about controlling the region over which the 3-equation stuff is handled with an active ice shelf. The melt occurs at depth over some small configurable region at the shelf/ocean interface. #72 is about better handling of coupled ice melt fluxes that are provided as 2D fields despite actually having a depth dependence in the real world (e.g Antarctic basal melt or iceberg calving fields from JRA55-do). It provides the ability to distribute the mass/heat over a depth range. I'm not sure there's much parameter unification to be had between the two PRs since the parameters represent quite different things. I could well be missing something here so keen to hear your thoughts. Also, out of curiosity, was this change motivated by physical or numerical reasons? |
|
Hi @dougiesquire As for the motivation, it is kind of a numerical trick but related to a physical process - sampling the meltwater from a bigger depth range means the effect of cold meltwater stratifying the upper ocean is reduced, so your thermal driving is bigger. Distributing it over a bigger range is needed for consistency with the sampling depth, but also implicitly increases mixing and reduces stratification. Gwyther 2020 has more details. In reality, we expect strong stratification in some places, but in the model we have a very high resolution compared to most models and maybe we don't have enough mixing to compensate (tides or related to sub-grid stuff or otherwise). We were trying this in the pan-Antarctic in the hope it would increase melting and fix our bias claireyung/mom6-panAn-iceshelf-tools#82 To demonstrate it works, this is a zoom in of a temperature profile under Thwaites ice shelf; left is 1.08m sampling/distribution distance, right is 3m distance; and stratification seems to follow those distances :) |
|
Sorry @angus-g, I've just noticed this is coming from your fork. Could you please redo from a branch on this repo? |
|
Oops, one day I'll remember to open it from the right place 😁 |
This is a draft for @claireyung to test whether we can specify different depths for a) the calculation of the surface boundary layer that feeds into the melt parameterisation; b) the application of the melt fluxes. There are three new parameters:
SHELF_FORCING_DEPTH: the equivalent toMINIMUM_FORCING_DEPTHapplied in shelf regions;HMIX_SHELF_PROP; the equivalent toHMIX_SFC_PROP(for T/S) applied in shelf regions;HMIX_UV_SHELF_PROP; the equivalent toHMIX_UV_SFC_PROP(for velocity) applied in shelf regions.