Fafmip access

[russfiedler]

These commits address issues that arose when fixing the frazil computation for the ACCESS suite of models.

1. Compute frazil formed using redistributed heat "temperature"
2. Pass equivalent "redistributed potential temperature" to the coupler when conservative temperature is used as the prognostic temperature.
3. The diagnostic temperature was being passed to the diagnostic package at time taup1 rather than time tau as verified by snapshots with contemp_equal_pottemp .eqv. .TRUE.

#280

#284

#285

[russfiedler]

@Fabiodias Could you describe the test runs you did with this commit so we can get it accepted?

[fabiobdias]

Hi @russfiedler,

I have performed the FAFMIP sanity test (using "do_fafmip_heat = true." but using "temp_correction_scale = 0.0") and can confirm that the variables "temp" and "redist_heat" are equal through the simulation (6-month long).

I also have performed a control run ("do_fafmip_heat = false.") and can confirm that the variable "temp" is equal in both control and sanity test experiments (when using "pottemp_equal_contemp = true"). (This is not the case for control x sanity test using "pottemp_equal_contemp = false", with small differences appearing after a few timesteps).

These two metrics were previously related with the fixes (1) and (2) Russ wrote in the first comment.

I also can certify that the fix (3) is working and the variables "pot_temp" and "temp" (conservative temperature) are equal for every time outputted. This wasn't the case for runs previous of this fix.

[russfiedler]

The recent fafmip_stress pull request #290 includes all the changes here. I'll close this request.