Version 26 (modified by lovato, 3 years ago) (diff)

TOP interface User Quick Guide

The present Quick Guide applies to NEMO 4.0+ releases.

TOP interface Structure

TOP synthetic Workflow

TOP namelists Walkthrough

MY_TRC interface for coupling external BGC models

Coupling an external BGC model using NEMO framework


TOP interface Structure

TOP (Tracers in the Ocean Paradigm) is the NEMO hardwired interface toward biogeochemical models and provide the physical constraints/boundaries for oceanic tracers. It consists of a modular framework to handle multiple ocean tracers, including also a variety of built-in modules. TOP interfaces has the following location in the code repository

<repository>/NEMOGCM/NEMO/TOP_SRC/

and the following modules are available:

TRP Interface to NEMO physical core for computing tracers transport
CFC Inert carbon tracers (CFC11,CFC12, SF6)
C14 Radiocarbon passive tracer
AGE Water age tracking
MY_TRC Template for creation of new modules and external BGC models coupling
PISCES Built in BGC model

The usage of TOP is activated by i) including in the configuration definition the component TOP_SRC and ii) adding the macro key_top in the configuration cpp file.

As an example, the user can refer to simplest configurations already available in the code GYRE_BFM or GYRE_PISCES.

Note that, since version 4.0, TOP interface core functionalities are activated by means of logical keys and all submodules preprocessing macros from previous versions were removed.


Here below the list of preprocessing keys that applies to the TOP interface (beside key_top):

key_iomput : use XIOS I/O

key_zdfddm & key_zdftke & key_zdfgls: vertical schemes (need to be updated after Merge2017 finalization)

key_trabbl : bottom boundary layer parameterization

key_agrif : enable AGRIF coupling

key_trdtrc & key_trdmxl_trc : trend computation for tracers


TOP synthetic Workflow

A synthetic description of the TOP interface workflow is given below to summarize the steps involved in the computation of biogeochemical and physical trends and their time integration and outputs, by reporting also the principal Fortran subroutine herein involved.


Model initialization (OPA_SRC/nemogcm.F90)

call to trc_init (trcini.F90)

↳ call trc_nam (trcnam.F90) to initialize TOP tracers and run setting

↳ call trc_ini_sms, to initialize each submodule

↳ call trc_ini_trp, to initialize transport for tracers

↳ call trc_ice_ini, to initialize tracers in seaice

↳ call trc_ini_state, read passive tracers from a restart or input data

↳ call trc_sub_ini, setup substepping if nn_dttrc /= 1


Time marching procedure (OPA_SRC/stp.F90)

call to trc_stp.F90 (trcstp.F90)

↳ call trc_sub_stp, averaging physical variables for sub-stepping

↳ call trc_wri, call XIOS for output of data

↳ call trc_sms, compute BGC trends for each submodule

↳ call trc_sms_my_trc, includes also surface and coastal BCs trends

↳ call trc_trp (TRP/trctrp.F90), compute physical trends

↳ call trc_sbc, get trend due to surface concentration/dilution

↳ call trc_adv, compute tracers advection

↳ call to trc_ldf, compute tracers lateral diffusion

↳ call to trc_zdf, vertical mixing and after tracer fields

↳ call to trc_nxt, tracer fields at next time step

↳ call to trc_rad, Correct artificial negative concentrations

↳ call trc_rst_wri, output tracers restart files


TOP namelists Walkthrough

namelist_top

Here below are listed the features/options of the TOP interface accessible through the namelist_top_ref and modifiable by means of namelist_top_cfg (as for NEMO physical ones).

Note that ## is used to refer to a number in an array field.


&namtrc_run ! run information
nn_dttrc Time step frequency for TOP computation (if >1 substepping is used)
ln_top_euler Use Euler Forward integration instead of leap-frog
ln_rsttr Start from a restart file (T) or not (F)
nn_rsttr Restart control.
0 : initial time step is not compared to the restart file value.
1 : do not use the value in the restart file.
2 : calendar parameters read in the restart file.
cn_trcrst_in
cn_trcrst_indir
Suffix of tracers INPUT restart file
Root directory for the location of INPUT restart files
cn_trcrst_out
cn_trcrst_outdir
suffix of tracers OUTPUT restart file
Root directory for the location of OUTPUT restart files
&namtrc ! tracers definition
jp_bgc Number of passive tracers of the BGC model
ln_pisces / ln_my_trc Run PISCES or MY_TRC BGC modules. These modules cannot be run at the same time since they both rely on jp_bgc
ln_age, ln_cfc11, ln_cfc12, ln_sf6, ln_c14 Activate sub-modules: tracers are automatically defined within each module. These can run along with main BGC modules.
ln_trcdta Initialisation from data input file (T) or not (F)
ln_trcdmp
ln_trcdmp_clo
Add a damping term (T) or not (F)
Add a damping term (T) or not (F) for closed seas
jp_dia3d, jp_dia2d Number of 3D/2D diagnostic variables to populate dedicated array beside the direct usage of iom_put calls.
sn_tracer(##) Data structure to specify tracer name and metadata for the chosen BGC module. Here set initialization and Boundary conditions. See structure description after this table.
&namtrc_dta ! Initialisation from data input file
sn_trcdta(##) Data structure to read tracers Initial conditions according to sn_tracer%llinit option. See structure description after this table.
rf_trfac(##) Multiplicative factor for initialized tracer. If omitted default is 1.0.
cn_dir Root directory for the location of INPUT data files
&namtrc_adv ! advection scheme for passive tracer
ln_trcadv_cen
nn_cen_h
nn_cen_v
2nd order centered scheme
=2 or 4, horizontal 2nd / 4th order
=2 or 4, vertical 2nd / 4th order
ln_trcadv_fct
nn_fct_h
nn_fct_v
nn_fct_zts
FCT scheme
=2 or 4, horizontal 2nd / 4th order
=2 or 4, vertical 2nd / 4th order
≥1, number of sub-timestep for 2nd order vertical scheme
ln_trcadv_mus
ln_mus_ups
MUSCL scheme
use (T) or not (F) upstream scheme near river mouths
ln_trcadv_ubs
nn_ubs_v
UBS scheme
= 2 , vertical 2nd order FCT
ln_trcadv_qck QUICKEST scheme
&namtrc_ldf ! lateral diffusion scheme for passive tracer
ln_trcldf_lap
rn_ahtrc_0
Laplacian operator (use .T. or not .F.)
Lateral eddy diffusivity coefficient [m2/s]
ln_trcldf_bilap
rn_bhtrc_0
Bilaplacian operator (use .T. or not .F.)
Lateral eddy diffusivity coefficient [m2/s]
ln_trcldf_lev
ln_trcldf_hor
ln_trcldf_iso
ln_trcldf_triad
Iso-level direction (use .T. or not .F.)
Horizontal (geopotential) direction (use .T. or not .F.)
Iso-neutral (standard operator) direction (use .T. or not .F.)
Iso-neutral (triad operator) direction (use .T. or not .F.)
rn_fact_lap Multiplicative factor to enhance zonal eddy diffusivity.
&namtrc_zdf ! vertical physics
ln_trczdf_exp Split explicit (T) or implicit (F) time stepping
nn_trczdf_exp Number of sub-timestep for ln_trczdfexp=T
&namtrc_rad ! treatment of negative concentrations
ln_trcrad Artificially correct negative concentrations (T) or not (F)
&namtrc_dmp ! passive tracer newtonian damping
nn_zdmp_tr Vertical shape of damping term
0 : damping throughout the water column
1 : no damping in the mixing layer (kz criteria)
2 : no damping in the mixed layer (rho crieria)
cn_resto_tr NetCDF filename with input damping coefficients
&namtrc_ice ! Representation of sea ice growth & melt effects
nn_ice_tr Tracer concentration in sea ice. Available options

-1 (no vvl: identical cc in ice and ocean / vvl: cc_ice = 0)

0 (no vvl: cc_ice = zero / vvl: cc_ice = )

1 prescribed to a namelist value (implemented in pisces or through user definition in trcice_my_trc)
&namtrc_trd ! diagnostics on tracer trends ('key_trdtrc')
nn_trd_trc Time step frequency and tracers trends
nn_ctls_trc Control surface type in mixed-layer trends (0,1 or n<jpk)
rn_ucf_trc Unit conversion factor (=1 → /seconds ; =86400. → /day)
ln_trdmld_trc_restart Use restart (T) or not (F) for ML diagnostics
ln_trdmld_trc_instant Diagnose trends of instantaneous (T) or mean (F) ML T/S
ln_trdtrc(##) Compute trend for nth tracer. Array of logical values.
&namtrc_bc ! Forcing data for boundary conditions
sn_trcsbc(##), sn_trccbc(##), sn_trcobc(##) Data structure to read tracers boundary conditions according to sn_tracer%llsbc,sn_tracer%llcbc, sn_tracer%llobc options. See structure description after this table.
rn_trsfac(##), rn_trcfac(##), rn_trofac(##) Multiplicative factor for SURFACE (sbc), COASTAL (cbc), and LATERAL (obc) forcing data files
cn_dir_sbc, cn_dir_cbc, cn_dir_obc Root directory for the location of SURFACE (sbc), COASTAL (cbc), and LATERAL (obc) forcing data files
ln_rnf_ctl Remove runoff dilution on tracers with absent river load (only without VVL)
rn_bc_time Time scaling factor for SBC and CBC data (seconds in a day)
&namtrc_bdy ! Setup of tracer boundary conditions
cn_trc_dflt Open Boundary Condition applied by default to all tracers
cn_trc Boundary conditions used for tracers with data files (selected in namtrc using sn_tracer%llobc)
nn_trcdmp_bdy Use damping timescales defined in nambdy of namelist
0 : NO damping of tracers at open boundaries
1 : Only for tracers forced with external data
2 : Damping applied to all tracers
&namage ! Water Age module
rn_age_depth Depth over which age tracer reset to zero
rn_age_kill_rate Relaxation timescale (s) for age tracer shallower than age depth

Two main types of data structure are used within TOP interface to initialize tracer properties (1) and to provide related initial and boundary conditions (2).

1. TOP tracers initialization : sn_tracer (namtrc)

Beside providing name and metadata for tracers, here are also defined the use of initial (sn_tracer%llinit) and boundary (sn_tracer%llsbc, sn_tracer%llcbc, sn_tracer%llobc) conditions.

In the following, an example of the full structure definition is given for two idealized tracer both with initial conditions given and the first with surface boundary forcing prescribes, while the second has surface and coastal forcing:

!             !    name   !           title of the field            !   units    ! initial data ! sbc   !   cbc  !   obc  !
sn_tracer(1)  = 'TRC1'    , 'Tracer 1 Concentration                ',   ' - '    ,  .true.      , .true., .false., .true.
sn_tracer(2)  = 'TRC2 '   , 'Tracer 2 Concentration                ',   ' - '    ,  .true.      , .true., .true. , .false.

As tracers in BGC models are increasingly growing, the same structure can be written also in a more compact and readable way:

!             !    name   !           title of the field            !   units    ! initial data !
sn_tracer(1)  = 'TRC1'    , 'Tracer 1 Concentration                ',   ' - '    ,   .true.
sn_tracer(2)  = 'TRC2 '   , 'Tracer 2 Concentration                ',   ' - '    ,   .true.
! sbc
sn_tracer(1)%llsbc = .true.
sn_tracer(2)%llsbc = .true.
! cbc
sn_tracer(2)%llcbc = .true.

The data structure is internally initialized by code with dummy names and all initialization/forcing logical fields set to .false. .

2. SBC structure to read input fields : sn_trcdta (namtrc_dta), sn_trcsbc/sn_trccbc/sn_trcobc (namtrc_bc)

This data structure is based on the general one defined for NEMO core in the SBC component (see details in User Manual SBC Chapter on Input Data specification).

The following example illustrates the data structure in the case of a single tracer with initial conditions contained in the file named tracer_1_data.nc (.nc is implicitly assumed in namelist filename), with a doubled initial value, and located in the usr/work/model/inputdata/ folder:

!               !  file name             ! frequency (hours) ! variable  ! time interp. !  clim  ! 'yearly'/ ! weights  ! rotation ! land/sea mask !
!               !                        !  (if <0  months)  !   name    !   (logical)  !  (T/F) ! 'monthly' ! filename ! pairing  ! filename      !
  sn_trcdta(1)  = 'tracer_1_data'        ,        -12        ,  'TRC1'   ,    .false.   , .true. , 'yearly'  , ''       , ''       , ''
  rf_trfac(1) = 2.0
  cn_dir = “usr/work/model/inputdata/”

Note that, the Lateral Open Boundaries conditions are applied on the segments defined for the physical core of NEMO (see BDY description in the User Manual).


namelist_trc

Here below the description of namelist_trc_ref used to handle Carbon tracers modules, namely CFC and C14.

&namcfc ! CFC
ndate_beg datedeb1
nyear_res iannee1
clname Name of formatted file with annual hemisperic CFCs concentration in the atmosphere (ppt)
&namc14_typ ! C14 - type of C14 tracer, default values of C14/C and pco2
kc14typ Type of C14 tracer (0=equilibrium; 1=bomb transient; 2=past transient)
rc14at Default value for atmospheric C14/C (used for equil run)
pco2at Default value for atmospheric pcO2 [atm] (used for equil run)
rc14init Default value for initialization of ocean C14/C (when no restart)
&namc14_sbc ! C14 - surface BC
ln_chemh Use (T) or not (F) Chemical enhancement in piston velocity
xkwind Coefficient for gas exchange velocity
xdicsur Reference DIC surface concentration (mol/m3)
&namc14_fcg ! files & dates
cfileco2 Name of formatted file with atmospheric co2 - Bomb
cfilec14 Name of formatted file with atmospheric c14 - Bomb
tyrc14_beg starting year of experiment - Bomb


MY_TRC interface for coupling external BGC models

The generalized interface is pivoted on MY_TRC module that contains template files to build the coupling between NEMO and any external BGC model.

The call to MY_TRC is activated by setting ln_my_trc = .true. (in namtrc)

The following 6 fortran files are available in MY_TRC with the specific purposes here described.


par_my_trc.F90

This module allows to define additional arrays and public variables to be used within the MY_TRC interface

trcini_my_trc.F90

Here are initialized user defined namelists and the call to the external BGC model initialization procedures to populate general tracer array (trn and trb). Here are also likely to be defined suport arrays related to system metrics that could be needed by the BGC model.

trcnam_my_trc.F90

This routine is called at the beginning of trcini_my_trc and should contain the initialization of additional namelists for the BGC model or user-defined code.

trcsms_my_trc.F90

The routine performs the call to Boundary Conditions and its main purpose is to contain the Source-Minus-Sinks terms due to the biogeochemical processes of the external model. Be aware that lateral boundary conditions are applied in trcnxt routine. IMPORTANT: the routines to compute the light penetration along the water column and the tracer vertical sinking should be defined/called in here, as generalized modules are still missing in the code.

trcice_my_trc.F90

Here it is possible to prescribe the tracers concentrations in the seaice that will be used as boundary conditions when ice melting occurs (nn_ice_tr =1 in namtrc_ice). See e.g. the correspondent PISCES subroutine.

trcwri_my_trc.F90

This routine performs the output of the model tracers (only those defined in namtrc) using IOM module (see Manual Chapter “Output and Diagnostics”). It is possible to place here the output of additional variables produced by the model, if not done elsewhere in the code, using the call to iom_put.


Coupling an external BGC model using NEMO framework

The coupling with an external BGC model through the NEMO compilation framework can be achieved in different ways according to the degree of coding complexity (e.g., code is only in one file or it has multiple modules and interfaces) of the BGC model itself.


Beside the 6 core files of MY_TRC module, let’s assume an external BGC model named “MYBGC” and constituted by a rather essential coding structure, likely few Fortran files. The new coupled configuration name is NEMO_MYBGC.

The best solution is to have all files (the modified MY_TRC routines and the BGC model ones) place in a single folder and use the makenemo external addressing of MY_SRC folder.

The coupled configuration listed in cfg.txt will look like

NEMO_MYBGC OPA_SRC TOP_SRC

and the related cpp_MYBGC.fcm content will be

bld::tool::fppkeys  key_zdftke key_dynspg_ts key_iomput key_mpp_mpi key_top

the compilation qith makenemo will be executed the following syntax

makenemo -n NEMO_MYBGC -m <arch_my_machine> -j 8 -e <MYBGCPATH>

The makenemo feature “-e” was introduced to readdress at compilation time the standard MY_SRC folder (usually found in NEMO configurations) with a user defined external one.


The compilation of more articulated BGC model code & infrastructure, like in the case of BFM (BFM-NEMO coupling manual) requires some additional features.

As before, let’s assume a coupled configuration name NEMO_MYBGC, but in this case MYBGC model repository has 3 different subfolders for biogeochemistry, named initialization, pelagic, and benthic, and a separate one named nemo_coupling that contains the modified MY_SRC routines. The latter folder containing the modified NEMO coupling interface will be still linked using the makenemo “-e” option.

To include the BGC model subfolders in the compilation of NEMO code it will be necessary to extend the configuration cpp_NEMO_MYBGC.fcm file to include the specific paths of MYBGC folders, as in the following example

bld::tool::fppkeys  key_zdftke key_dynspg_ts key_iomput key_mpp_mpi key_top
src::MYBGC::initialization         <MYBGCPATH>/initialization
src::MYBGC::pelagic                <MYBGCPATH>/pelagic
src::MYBGC::benthic                <MYBGCPATH>/benthic

bld::pp::MYBGC      1
bld::tool::fppflags::MYBGC   %FPPFLAGS
bld::tool::fppkeys           %bld::tool::fppkeys MYBGC_MACROS

where MYBGC_MACROS is the space delimited list of macros used in MYBGC model for selecting/excluding specific parts of the code. The BGC model code will be preprocessed in the configuration BLD folder as for NEMO, but with an independent path, like NEMO_MYBGC/BLD/MYBGC/<subforlders>.

The compilation will be performed as in the previous case with the following

makenemo -n NEMO_MYBGC -m <arch_my_machine> -j 8 -e <MYBGCPATH>

Note that, the additional lines specific for the BGC model source and build paths, can be written into a separate file, e.g. named MYBGC.fcm, and then simply included in the cpp_NEMO_MYBGC.fcm as follow

bld::tool::fppkeys  key_zdftke key_dynspg_ts key_iomput key_mpp_mpi key_top
inc <MYBGCPATH>/MYBGC.fcm

This will enable a more portable compilation structure for all MYBGC related configurations.

Important: the coupling interface contained in nemo_coupling cannot be added using the FCM syntax, as the same files already exists in NEMO and they are overridden only with the readdressing of MY_SRC contents to avoid compilation conflicts due to duplicate routines.

All modifications illustrated above, can be easily implemented using shell or python scripting to edit the NEMO configuration cpp.fcm file and to create the BGC model specific FCM compilation file with code paths.