cvmix_coeffs_kpp doesn't exactly match POP
Closed this issue · 4 comments
The way CVMix handles the nonlocal term doesn't account for enhanced diffusivity... need to multiply by Kenh/Kold at bottom of boundary layer
When you address this, can we make the scaling optional to retain the current formulation. As mentioned in the telecon, @StephenGriffies will need to revise the documentation too.
I 'fixed' this issue without re-reading the ticket, so I didn't account for @adcroft-gfdl comment above. A simple fix will be to make Tnonlocal and Snonlocal optional inputs, and only update them if they are present... I expect to be able to do that later tonight or tomorrow, but I'm leaving this comment in case it takes longer to come back to.
In issue #35 @adcroft-gfdl mentions that we don't want to compute enhanced diffusivity unless we are matching the gradient term or both the gradient and nonlocal terms; in commit eb7c85c I added logic to only compute enhanced diffusivity if we are matching both terms (as done in POP). This needs to be updated to compute the enhanced diffusivity if we are just matching the gradient term (without updating the nonlocal term, which, per my last comment, needs to be made an optional argument to compute_enhanced_diff() anyway)
I still need to add logic to call enhanced diffusivity if we're only matching the gradient terms, which will require making the nonlocal terms optional inputs, but the enhanced diffusivity routine is returning what I expect when matching both the gradient and nonlocal terms.