add_MARBL (compute): are river fluxes in MOM correct?
mnlevy1981 opened this issue · comments
In MOM runs, we are seeing an order-of-magnitude change in some tracers in regions near river mouths, where companion POP runs stay relatively constant. It's possible that something weird is happening with river fluxes. Some things to investigate:
- In POP, does the global integral of each
*_RIV_FLUX
diagnostic match the global integral in the forcing dataset? - Same question, but for MOM
- In MOM, are we applying the river flux forcing correctly?
It is possible that there is a unit conversion issue somewhere, but it may just be the case that the different models are reacting to river flux in different ways
Note that we are applying river fluxes as an in_flux_optional
argument to applyTracerBoundaryFluxesInOut()
(unlike POP, where it goes through the EBM)
I noticed this issue while looking at open tickets, and did some quick analysis.
-
POP global integrals vs forcing:
Relative difference between global integrals from MOM and those from MOM forcing ---- NO3 - din: 7.307e-16 PO4 - dip: -4.248e-16 DON + DONr - don: 5.322e-16 DOP + DOPr - dop: -1.850e-16 SiO3 - dsi: 0.000e+00 Fe - dfe: 0.000e+00 DIC - dic: -3.425e-16 ALK + NO3 - alk: 5.137e-16 DOC + DOCr - doc: 2.610e-16
-
MOM global integrals vs forcing
Relative difference between global integrals from POP and those from POP forcing ---- NO3 - din: 0.000e+00 PO4 - dip: 0.000e+00 DON + DONr - don: -3.569e-16 DOP + DOPr - dop: 1.900e-16 SiO3 - dsi: -1.843e-16 Fe - dfe: 0.000e+00 DIC - dic: 0.000e+00 ALK + NO3 - alk: 0.000e+00 DOC + DOCr - doc: -1.270e-16
-
MOM global integrals vs POP global integrals
Relative difference between global integrals from POP and those from MOM ---- PO4_RIV_FLUX: 1.01% NO3_RIV_FLUX: 1.43% SiO3_RIV_FLUX: 2.25% Fe_RIV_FLUX: 2.68% DIC_RIV_FLUX: 3.75% ALK_RIV_FLUX: 3.90% DOC_RIV_FLUX: 2.74% DON_RIV_FLUX: 0.60% DOP_RIV_FLUX: 2.68% DOPr_RIV_FLUX: 2.68% DONr_RIV_FLUX: 0.60% DOCr_RIV_FLUX: 2.74%
So there are small differences between the two models that are likely due to differences in the grids / land masks resulting in differences from the horizontal interpolation of the forcing data, but in general the forcing dataset differences look small and both models appear to be applying the forcing in similar manners.