Opened 4 years ago

Closed 3 years ago

#1686 closed Bug (wontfix)

Computed bbl diffusive coefficients & tracer conservation

Reported by: mouchet Owned by: nemo
Priority: normal Milestone:
Component: TOP Version: release-3.6
Severity: Keywords: BBL
Cc: anne.mouchet@…

Description

The present problem may be related to that described in ticket #1529

Here i use NEMO off-line and fixed volume.
In that case conservation of a fully passive tracer may only be obtained by imposing w=0 at the surface as well as emps=0 (it is not physically realistic but useful for testing tracer modules).
Under these conditions:

  • when bbl coefficients are read in dynamic files conservation is perfect
  • when bbl coefficients are computed conservation degrades by up to 7 order of magnitude for a 1 year long experiment (see below).

The 2 tracers considered here differ simply by their initial distribution; they have no source sink terms. The only difference between the 2 exps is the value of ln_dynbbl.
Only diffusive bbl was tested.

Thanks,
Anne

bbl coeff computed:
==================

ln_dynbbl = .false. ! bbl coef are in files, so read them - requires ("key_trabbl")

curie70:EXP00>grep drift *ocean*
tracer nb : 1 name : MYT1 mean : 0.9999999982E+00 min : 0.8026565369E+00 max : 0.1158012835E+01 drift : -0.1763840643E-06 %
tracer nb : 2 name : MYT2 mean : 0.2131708811E-04 min : 0.0000000000E+00 max : 0.4190379461E-04 drift : 0.9901576322E-03 %

bbl coeff read:
==================

ln_dynbbl = .true. ! bbl coef are in files, so read them - requires ("key_trabbl")

curie70:EXP00>grep drift *ocean*
tracer nb : 1 name : MYT1 mean : 0.1000000000E+01 min : 0.7985668668E+00 max : 0.1158035137E+01 drift : 0.6255409052E-10 %
tracer nb : 2 name : MYT2 mean : 0.2131687704E-04 min : 0.0000000000E+00 max : 0.4237807293E-04 drift : 0.7412127798E-10 %

Commit History (0)

(No commits)

Change History (3)

comment:1 Changed 4 years ago by cetlod

It's probably because bbl coefficient must be computed in dtadyn.F90 before interpolating tn/sn/un/vn in the same way than the isopycnal slopes.

This requires significant modification in the code - trabbl.F90 & dtadyn.F90 routines - 

Since the conservation is ok when reading the coef, the advice is either to have the bbl coef. stored in the dynamical files, either to run without the bbl diffusive parameterization or either to live with the "non conservation"

For the next releases - offline with vvl - the option to compute bbl coefficient won't exist anymore

comment:2 Changed 4 years ago by nicolasmartin

  • Keywords BBL added; bbl removed

comment:3 Changed 3 years ago by clevy

  • Resolution set to wontfix
  • Status changed from new to closed
Note: See TracTickets for help on using tickets.