Bug fixes for GEOMETRIC parameterisation
Bug fixes for GEOMETRIC
Various bugs in the formulation of the parameterised eddy-energy budget:
- Missing an e[12] factor in mean-flow advection, so that advection by the mean flow was doing essentially nothing even if option was switched on
- Minus signs in the computation of the advective trend
- Inconsistencies with definition by [uv]n_adv, which has changed since NEMO 3.6
- Inconsistencies with depth-averaging on different points in advection and diffusion
- Rossby advection is currently along -i direction, when it wants to be Westward
- keS (KE -> EKE) trend has one e3 factor too many (though option not used normally)
When correctly activated the main benefit comes from the mean flow advection, resulting in smoother energy fields, fewer extreme values, and more sensible values of aeiw.
-
Branches impacted: main (LDF)
Analysis
Relatively easy fixes, as identified by Gurvan Madec and Romain Torres (tested by the latter in CNRM NEMO 3.6).
Fix
- add/remove scaling factors as appropriate
- remove minus signs
- add in various multiplications/divisions by h[tuv]
- make a call to rot_rep in SBC/geo2ocean for rotating the Rossby advection when the ORCA grid is used