search

NOAA-CEFI-Regional-Ocean-Modeling / ocean_BGC

3 stars 8 forks source link

Update interfaces to pass a photoacclimation MLD from MOM6 #88

Closed theresa-morrison closed 3 months ago

theresa-morrison commented 3 months ago

This PR changes the intefaces between MOM6 and COBALT to allow a mixed layer depth to be passed from MOM6. The new argument is optional, and if present used instead of calculating the photoacclimation depth in COBALT.

yichengt900 commented 3 months ago

Hi @theresa-morrison , thank you for opening this draft PR. I understand that it has a few dependencies and may require further refinement. I will begin by reviewing it and running several tests, including a 1D test case and a 5-year simulation using the NWA12 domain. Will keep you updated on the progress.

theresa-morrison commented 3 months ago

Hi @yichengt900 thanks! I was planning to talk through the code with @charliestock tomorrow to make sure it looks like it will do what the COBALT group wants before starting the 5 year test.

yichengt900 commented 3 months ago

@theresa-morrison @andrew-c-ross, and @charliestock:

The preliminary results (5yr NWA run) with the new interface (MLD pre-calculated in MOM6 and passed to COBALT) are quite promising. First, the new MLD_003 diagnostics (set HREF_FOR_MLD=10.0) gives a slight better bias when compared to the dBM 2022 (see figures below):

new MLD003 when HREF_FOR_MLD=10.0: mld003_eval

Org MLD003: mld003_eval

The domain and regional surface chl also look ok to me: domain_chl regional_chl

yichengt900 commented 3 months ago

As a reference, below are the surface chl values from our most recent bi-weekly regression testing. As you can see, the results are quite close to those using the MOM6-calculated photoacclimation MLD (figures above):

domain_chl

regional_chl

charliestock commented 3 months ago

Thanks @yichengt900 and @theresa-morrison, this looks great. I am tempted to make the leap to this improved approach sooner than later. Let's discuss whether there should be any final checks on this when we get together tomorrow. If not, let's move ahead with the merge. If we think a few more checks are necessary, we can set them in motion before the weekend and confirm things next week. Thank you both!

yichengt900 commented 3 months ago

Ok here are some new updates from the 1993-2019 long term NWA12 runs:

Regional Chl: regional_chl

Domain Chl: domain_chl

GoM Hypoxia : hypoxic_area_from_daily

MLD003: mld003_eval_new

Although it’s not an exact apple-to-apple comparison to the results presented in Andrew's paper, the overall consistency is within the expected range. One observation from the domain chlorophyll plots is that the discrepancy along the OBC appears to be larger in the new run. This could potentially be related to the shorter OBC update frequency I used in this test. Hopefully, the new BGC nudging PR proposed by @theresa-morrison could mitigate this issue.

Overall, the new interface changes look promising. I would suggest retaining the COBALT-MLD calculation for now until we’re all comfortable with the new interface and MOM6-calculated MLD for various regions. We can still remove the eq_of_state argument from the MOM6 side, which should not impact this PR.

theresa-morrison commented 3 months ago

Thanks, Yi-Cheng. I decided pha_all_same=true is a better default so there is one more commit. I think this PR should be ready to go and can be merged before the changes to MOM6. This PR alone should have no impact on the model since photo_acc_dpth will not be present in the call to update COBALT.