Version 1 (modified by gm, 8 years ago) (diff)

Last edited Timestamp?

Author : Gurvan Madec & Fabien Roquet

Ticket : #927

Branch v3.5 : 2013/dev_r3858_CNRS3_Ediag


LOCEAN .3 (2013) - Energy diagnostics

Motivation: output 3D trends of tracers, momentum, kinetic energy and potential energy.
Status : the extraction of trends terms exists, but not the 3D output of the trends
Main tasks :

(1) implement the 3D output of tracers and momentum trends using iom_put
(2) compute and output the 3D trends of PE and KE. In particular introduce a consistent way of computing PE
(3) add Vallis equation of state in replacement of linear in T and in T-S eos (nn_eos=2)
(4) add the exact Jacket and Mc Dougall? eos (nn_eos=-1) keep for backward compatibility the nn_eos=0 old case
(5) add TEOS10 equation of state
(3) validation + documentation

Science Reviewer: NOCS guy?
System Reviewer: NOCS guy?
Deadline: automn 2013
Priority: high
Depends on: -
Principal Investigator : Gurvan Madec and Fabien Roquet (gurvan.madec@…)


Detail of the implementation

I. Port from v3.4 to v3.5

convert the v3.4 development branch 2012/dev_r3309_LOCEAN12_Ediag into the following v3.5 development branch 2013/dev_r3858_CNRS3_Ediag to v3.5

see Changeset [3874] ; [3876] and [3878]

II. First revision of the eos

  • introduce Vallis eos as the alternative to UNESCO type eos,
  • add systematic calculation of before and now partial derivative of density anomaly with respect to T and s (rab_b and rab_n 4D arrays) and compute rn2 and rrau using rab

To do so, the following changes have been made:

  • Added 4D variable rab_b(:,:,:,jp_ts) and rab_n(:,:,:,jp_ts) (oce.F90 ; istate.F90, ) and their calculation in eosbn2.F90
  • Add Vallis eos and suppress the 2 old eos.
  • introduce the computation of rab_pe, the primitive of rab in eosbn2.F90 module and use it in trdpen.F90

NB: rhd and rhop are kept for backward compatibility, but they are meant to become obsolete (especially rhop which could be easily removed, with minor changes in zdfkpp, zdfmxl and zdfric


Pending issues :

  • diagnostic of trends missing in the following modules :                             =⇒> To be done !!!'''

traldf_grif.F90 ; dynhdf_(iso,lap,bilap and bilapg).F90 ; dynnept.F90 ; dynzdf_(exp,imp).F90 ; dynspg_ts.F90

  • verify the change in sign in transport computation ticket #1043 (in traldf_iso and griff)
  • atmospheric pressure gradient trend not taken into account (see dynspg.F90) =⇒> To be done !!!'''
  • kpp non-local trend put in zdf trends !!! this will not work ! a additional trend term should be add To be done !!!'''
  • problems to be solved: vvl case for tracer sad trends ; flux form case for had (keg) and zad momentum trends
  • add separate modules for each option …
  • create the momentum diag over the ML
  • reshape trdtra so that T and S are treated separately in all trd routine (including mld diag…) create umask_i and vmask_i (2D) fields that mirror task_i field but for the velocity points

Fabien addition (copy from the v3.4 page):

In oce.F90:

  • rhd and rhop are kept for backward compatibility, but they are meant to become obsolete
    (especially rhop which could be easily removed, with minor changes in zdfkpp, zdfmxl and zdfric)
  • TODO rn2b might become obsolete (only one rnb variable might be enough).
  • TODO check restartibility

In step.F90:

  • Added "IF( .NOT.ln_bfrimp)" before the "CALL dyn_bfr( kstp )" to clarify that dyn_bfr is called only if bottom friction is explicit.
  • TODO Diagnostic of BFR should be improved!
  • added eos calls to set alpha and beta expansion coeffs before bn2 computation.
  • Consider removing calls to bn2, and compute it only where needed:
    • rn2b needed to call ldf_slp
    • also: asmtrj, step_c1d, ldfeiv, zdfddm, zdfevd, zdfgls, zdfkpp, zdfric, zdftke, zdftmx
  • Also, remove obsolete calls to eos when necessary.
  • TODO consider removing calls to eos to get rhop.

In trdtra.F90:

  • removed semi-colon on lines 118 and 267.
  • replaced ln_glo_trd by ln_PE_trd in line 239.
  • Moved wrk_alloc and wrk_dealloc in the IF (.NOT.lk_vvl ) structure.

In trddyn.F90:

  • Added wrk_alloc, wrk_dealloc and lbc_lnk in trd_dyn_iom.
    Remove unused variables ztswu, ztswv.
  • Following norm in the code, utrd_bfr and vtrd_bfr should be filled only when ln_bfrimp=.FALSE. (non implicit bottom friction).
  • Implicit bottom friction is calculated at the end of each timestep, in dynzdf_imp.
    The diagnostic of bottom friction in the implicit case is saved in separate variable names to avoid confusion: utrd_bfri and vtrd_bfri.
    Similarly, diagnostics of wind stress inputs are provided in 2D fields utrd_tau and vtrd_tau.

In tranxt.F90:

  • Important:
    If ln_dyn_hpg_imp=.TRUE., Brown and Campana effect is added. The pressure gradient term is computed using a linear combination version of T and S at times b, n and a:
    Tbc=Tb/4+Tn/2+Ta/4
    Sbc=Sb/4+Sn/2+Sa/4

In trdglo.F90:

  • In trd_glo_init, corrected definition for tvolt:
    tvolt = tvolt + SUM( e1e2t(:,:) * fse3t(:,:,jk) * tmask(:,:,jk) * tmask_i(:,:) )
  • Treatment of bfr has been clarified: if bfr is implicit (ln_bfrimp=.TRUE.), the bottom friction term remain 0, as bottom friction is implicitly included in the vertical diffusion term.
  • Correction of outputs of dynamic trends and hke.
  • What is the test: "pressure gradient u2 = - 1/rau0 u.dz(rhop)" ? Identity is not verified in my configuration!
  • Wind stress is always zero (although not in the 3d diag provided in trddyn.F90)
  • In glo_dyn_wri, missing zcof to compute density flux at w-point?!
  • Conversion is calculated as -g*div(rho*U)/rho0 !! (missing z)
  • TODO Once trddyn and trdtra are ready, use them for trdglo computations…

In trd_oce.F90

INTEGER, PUBLIC, PARAMETER ::   jptot_dyn  = 13     !: Total trend nb: change it when adding/removing one indice below    !                               ===============     !      
INTEGER, PUBLIC, PARAMETER ::   jpdyn_hpg  =  1     !: hydrostatic pressure gradient     
INTEGER, PUBLIC, PARAMETER ::   jpdyn_spg  =  2     !: surface     pressure gradient    
INTEGER, PUBLIC, PARAMETER ::   jpdyn_keg  =  3     !: kinetic energy gradient  or horizontal advection    
INTEGER, PUBLIC, PARAMETER ::   jpdyn_rvo  =  4     !: relative  vorticity      or metric term    
INTEGER, PUBLIC, PARAMETER ::   jpdyn_pvo  =  5     !: planetary vorticity    
INTEGER, PUBLIC, PARAMETER ::   jpdyn_zad  =  6     !: vertical advection    
INTEGER, PUBLIC, PARAMETER ::   jpdyn_ldf  =  7     !: horizontal diffusion       
INTEGER, PUBLIC, PARAMETER ::   jpdyn_zdf  =  8     !: vertical   diffusion    
INTEGER, PUBLIC, PARAMETER ::   jpdyn_bfr  =  9     !: bottom  stress     
INTEGER, PUBLIC, PARAMETER ::   jpdyn_atf  = 10     !: Asselin time filter    
INTEGER, PUBLIC, PARAMETER ::   jpdyn_tau  = 11     !: surface stress    
INTEGER, PUBLIC, PARAMETER ::   jpdyn_bfri = 12     !: implicit bottom friction (ln_bfrimp=.TRUE.)    
INTEGER, PUBLIC, PARAMETER ::   jpdyn_ken  = 13     !: use for calculation of KE
  • where jpdyn_tau and jpdyn_bfri are diagnostics of surface and bottom stress, respectively.
  • jpdyn_ken has been added for the calculation of KE in dynnxt.F90 (done just before the swap)

In trd_ken.F90:

  • l110: DO ji = 2, jpj should be DO ji = 2, jpi
  • Same in l133, l152 and l178.
  • In l183, replaced "ketrd_bfr" by "ketrd_bfri" to avoid confusion with "ketrd_bfr".
  • Important: contribution from atf is calculated with un and vn, after the swap.
    A small error is made: the contribution of ATF term at time t is recorded with trends taken at t-1.
  • In trd_ken_init, the 3 variables fse3x_n were replaced by their constant value fse3x for the non-vvl case.
  • Definition of KE (KE in xml) at time t+½: KE(t+½)=rau0*u(t)*u(t+1)/2.
    Add also subroutine ken_p2k to compute conversion rate.
    KE and ketrd_convP2K both calculated during dynnxt.F90, using ktrd=jpdyn_ken
  • l. 109: multiplication by rau0 of zke to obtain KE trends in W/m3, and KE in J/m3.
  • remove r1_2_rau0 variable, which is no more used.

In dynspg_flt:

  • Add a diagnostic of the explicit and implicit (due to filter) contributions to SPG.
  • ssh_flt: diagnostic of the ssh modification due to filter.

In dynnxt.F90

  • computation of z1_2dt must be put before the IF( ln_dyn_trd ) block (l. 194 and 195)

In eosbn2.F90

  • Vallis equation of state added (Vallis 2006, p34-35): simple EOS which accounts for thermobaricity, cabelling and compressibility.
    Generalization of linear equation of state, so T-linear and TS-linear cases have been removed.
    Vallis EOS: nn_eos=1.
  • Add original Jackett and McDougall? (1995) EOS (nn_eos=-1)
    nn_eos=0 case is a modified version of the Jackett and Mc Dougall? (1995) EOS!! Numerically close though…
  • eos_alpbet modified, which now provides alpha and beta (as a 4d variable) instead of alpha/beta and beta0=0 or 1.
    ldf_slp_grif in ldfslp.F90 modified accordingly (only place eos_alpbet was used).
    Now, there is no problem if beta=0, because its inverse is never used.
  • bn2 modified:
    • use an exact formulation based on each eos
    • bn2(ts,pn2): call first eos_alpbet to compute alpha and beta coefficients, then interpolating alpha/beta on w-points and finally calculating pn2.
    • bn2(ts,alpbet,pn2): alpha and beta coefficients are directly given in arguments, then interpolated to compute pn2.
    • Consider removing bn2 from eosbn2.F90
    • TODO Modify trabbl.F90 to use alpbet
    • TODO Modify tranpc.F90 to use alpbet
    • TODO Modify zdfkpp.F90 to use alpbet
  • eos has now a unified interface:
    • eos(ts,rhd) gives density anomaly (3D field)
    • eos(ts,rhd,rhop) should become obsolete (density anomaly + sigma0)
    • eos(ts,dep,rhop) density anomaly at one zt(pdep) (2D)
    • eos(ts,alpbet) providing alpha and beta (4D field jpi,jpj,jpk,jpts)
    • eos(ts,rhd,alpbet) density+alpha/beta (3D + 4D fields)
  • eos_pen added to compute PE anomaly, and equivalents of alpha and beta for the PE state variable (called alpha_pe and beta_pe, given in alpbet_pe).
    function is called at each time step trdpen.F90 (PE diagnostic).
    PE anomaly is defined as: (PE - rau0*gz)/(rau0*gz). For a linear case, PE anomaly is equal to density anomaly (nice isn't it?)
    When z=0, PE anomaly is defined asymptotically, converging toward the density anomaly value (in the code, if z<0 —> z=1).
    TODO Waiting for a new EOS from Trevor Mc Dougall? that would allow a more efficient computation of each terms.

In trdtra.F90

  • Add 'trdt(:,:,:) = ptrd(:,:,:) * tmask(:,:,:)' in CASE jptra_bbc in trd_tra (l. 103)

Documentation:

  • In Annex A,
    • change label name Apdx_A_grad_p in Apdx_A_grad_p2 l. 418 and 430.
    • sign error for vertical advection in Eq. A.18 (tracer equation)
  • In Annex C, sign errors corrected in discretization of HPG (p. 293)
  • In Chap_TRA,
    • update part 5.8 on equation of state to include reference to Vallis2006+JM95 true formulation
  • In Chap_DYN,
    • correction of title 6.1.2, and references
    • Changes on Asselin description
  • In Chap_STP, modification on Asselin description

General comments:

  • implement e3t = dk(gdept), and e3w=dk(gdpew).
  • N2 calculation could be speed up by precalculating ratios: (zw[k]-zt[k])/e3w[k] and (zt[k-1]-zw[k])/e3z[k]
  • put KE and PE directly in diawri?
  • Energetics of Brown and Campana effect?
  • Modify HPG to use alpha/beta + partial step formulation on fixed z-levels

Testing

Testing could consider (where appropriate) other configurations in addition to NVTK].

NVTK Tested '''YES/NO'''
Other model configurations '''YES/NO'''
Processor configurations tested [ Enter processor configs tested here ]
If adding new functionality please confirm that the

New code doesn't change results when it is switched off

and ''works'' when switched on
'''YES/NO/NA'''

(Answering UNSURE is likely to generate further questions from reviewers.)

'Please add further summary details here'

  • Processor configurations tested
  • etc——

Bit Comparability

Does this change preserve answers in your tested standard configurations (to the last bit) ? '''YES/NO '''
Does this change bit compare across various processor configurations. (1xM, Nx1 and MxN are recommended) '''YES/NO'''
Is this change expected to preserve answers in all possible model configurations? '''YES/NO'''
Is this change expected to preserve all diagnostics?

,,''Preserving answers in model runs does not necessarily imply preserved diagnostics. ''
'''YES/NO'''

If you answered '''NO''' to any of the above, please provide further details:

  • Which routine(s) are causing the difference?
  • Why the changes are not protected by a logical switch or new section-version
  • What is needed to achieve regression with the previous model release (e.g. a regression branch, hand-edits etc). If this is not possible, explain why not.
  • What do you expect to see occur in the test harness jobs?
  • Which diagnostics have you altered and why have they changed?Please add details here……..

System Changes

Does your change alter namelists? '''YES/NO '''
Does your change require a change in compiler options? '''YES/NO '''

If any of these apply, please document the changes required here…….


Resources

''Please ''summarize'' any changes in runtime or memory use caused by this change……''


IPR issues

Has the code been wholly (100%) produced by NEMO developers staff working exclusively on NEMO? '''YES/ NO '''

If No:

  • Identify the collaboration agreement details
  • Ensure the code routine header is in accordance with the agreement, (Copyright/Redistribution? etc).Add further details here if required……….