search

ai2cm / fv3gfs-fortran

FV3GFS Fortran for internal development at AI2
Other
6 stars 9 forks source link

Add surface type fraction diagnostics #377

Closed spencerkclark closed 1 year ago

spencerkclark commented 1 year ago

This PR adds diagnostics-manager-controlled surface-type-fraction diagnostics for the ocean_fraction, land_fraction, and sea_ice_fraction. These are all added as part of the gfs_sfc diagnostics module, similar to the slmsksfc diagnostic containing the integer-encoded surface type. As stated in #371, the advantage of these diagnostics is that they can be sensible time-and-area-averaged, so they retain their value upon standard time-averaging, coarse-graining, and conservative regridding approaches.

This notebook shows that these diagnostics are working as expected.

Note that it is theoretically possible to run FV3GFS with fractional land (this PR assumes one is always running in the mode of one grid cell, one surface type). Since we have not explored running in that mode yet, we have elected to omit trying to handle that case in this PR (it would make things more complicated).

Resolves #371.

spencerkclark commented 1 year ago

A question I have related to this is: should we modify things such that in sea-ice grid cells the sea-ice fraction is equal to sfcprop%fice and the ocean fraction is equal to 1 - sfcprop%fice? This is different than how we currently do things offline, but I think may be truer to the internals of the model.

spencerkclark commented 1 year ago

Surface skin temperature in sea ice grid cells is determined by the following weighted sum, weighing the temperature over the sea ice with the temperature over open water:

https://github.com/ai2cm/fv3gfs-fortran/blob/b3793303df9800de341a3f8e8b234c6f4dc2b7a1/FV3/gfsphysics/physics/sfc_sice.f#L332

fice is the ice fraction; ffw is 1.0 - fice. tgice is a constant, set to 271.2 K. This means that over all sea-ice grid cells, open water has a constant assumed skin temperature. This is different than over ocean grid cells, where the temperature is set either by the climatology or how we prescribe it. Therefore, it seems to make sense to keep these surface type fractions defined as they are, and output the existing fice diagnostic if we have an additional interest in the fraction of sea-ice-type grid cells covered with sea ice.