1 | ********************************** |
---|
2 | On line biogeochemistry coarsening |
---|
3 | ********************************** |
---|
4 | |
---|
5 | .. todo:: |
---|
6 | |
---|
7 | |
---|
8 | |
---|
9 | .. contents:: |
---|
10 | :local: |
---|
11 | |
---|
12 | Presentation |
---|
13 | ============ |
---|
14 | |
---|
15 | A capacity of coarsening physics to force a BGC model coupled to NEMO has been developed. |
---|
16 | This capacity allow to run 'online' a BGC model coupled to OCE-SI3 with a lower resolution, |
---|
17 | to reduce the CPU cost of the BGC model, while preserving the effective resolution of the dynamics. |
---|
18 | |
---|
19 | A presentation is available [attachment:crs_wiki_1.1.pdf here], where the methodology is presented. |
---|
20 | |
---|
21 | What is available and working for now in this version |
---|
22 | ===================================================== |
---|
23 | |
---|
24 | [To be completed] |
---|
25 | |
---|
26 | Description of the successful validation tests |
---|
27 | ============================================== |
---|
28 | |
---|
29 | [To be completed] |
---|
30 | |
---|
31 | What is not working yet with on line coarsening of biogeochemistry |
---|
32 | ================================================================== |
---|
33 | |
---|
34 | [To be completed] |
---|
35 | |
---|
36 | ''should include precise explanation on MPI decomposition problems too'' |
---|
37 | |
---|
38 | How to set up and use on line biogeochemistry |
---|
39 | ============================================= |
---|
40 | |
---|
41 | Extract the on line biogeochemistry branch |
---|
42 | ------------------------------------------ |
---|
43 | |
---|
44 | To get the appropriate source code with the on line coarsening of biogeochemistry feature: |
---|
45 | |
---|
46 | .. code-block:: console |
---|
47 | |
---|
48 | $ svn co https://forge.ipsl.jussieu.fr/nemo/browser/NEMO/branches/2018/dev_r5003_MERCATOR6_CRS |
---|
49 | |
---|
50 | |
---|
51 | How to activate coarsening? |
---|
52 | --------------------------- |
---|
53 | |
---|
54 | To activate the coarsening, ``key_crs`` should be added to list of CPP keys. |
---|
55 | This key will only activate the coarsening of dynamics. |
---|
56 | |
---|
57 | Some parameters are available in the namelist_cfg: |
---|
58 | |
---|
59 | .. code-block:: fortran |
---|
60 | |
---|
61 | ! passive tracer coarsened online simulations |
---|
62 | !----------------------------------------------------------------------- |
---|
63 | nn_factx = 3 ! Reduction factor of x-direction |
---|
64 | nn_facty = 3 ! Reduction factor of y-direction |
---|
65 | nn_msh_crs = 0 ! create (=1) a mesh file or not (=0) |
---|
66 | nn_crs_kz = 3 ! 0, volume-weighted MEAN of KZ |
---|
67 | ! 1, MAX of KZ |
---|
68 | ! 2, MIN of KZ |
---|
69 | ! 3, 10^(MEAN(LOG(KZ)) |
---|
70 | ! 4, MEDIANE of KZ |
---|
71 | ln_crs_wn = .false. ! wn coarsened (T) or computed using horizontal divergence ( F ) |
---|
72 | ! ! |
---|
73 | ln_crs_top = .true. !coarsening online for the bio |
---|
74 | / |
---|
75 | |
---|
76 | - Only ``nn_factx = 3`` is available and the coarsening only works for grids with a T-pivot point for |
---|
77 | the north-fold lateral boundary condition (ORCA025, ORCA12, ORCA36, ...). |
---|
78 | - ``nn_msh_crs = 1`` will activate the generation of the coarsened grid meshmask. |
---|
79 | - ``nn_crs_kz`` is the operator to coarsen the vertical mixing coefficient. |
---|
80 | - ``ln_crs_wn`` |
---|
81 | |
---|
82 | - when ``key_vvl`` is activated, this logical has no effect; |
---|
83 | the coarsened vertical velocities are computed using horizontal divergence. |
---|
84 | - when ``key_vvl`` is not activated, |
---|
85 | |
---|
86 | - coarsened vertical velocities are computed using horizontal divergence (``ln_crs_wn = .false.``) |
---|
87 | - or coarsened vertical velocities are computed with an average operator (``ln_crs_wn = .true.``) |
---|
88 | - ``ln_crs_top = .true.``: should be activated to run BCG model in coarsened space; |
---|
89 | so only works when ``key_top`` is in the cpp list and eventually ``key_pisces`` or ``key_my_trc``. |
---|
90 | |
---|
91 | Choice of operator to coarsene KZ |
---|
92 | --------------------------------- |
---|
93 | |
---|
94 | A sensiblity test has been done with an Age tracer to compare the different operators. |
---|
95 | The 3 and 4 options seems to provide the best results. |
---|
96 | |
---|
97 | Some results can be found [xxx here] |
---|
98 | |
---|
99 | Example of xml files to output coarsened variables with XIOS |
---|
100 | ------------------------------------------------------------ |
---|
101 | |
---|
102 | In the [attachment:iodef.xml iodef.xml] file, a "nemo" context is defined and |
---|
103 | some variable defined in [attachment:file_def.xml file_def.xml] are writted on the ocean-dynamic grid. |
---|
104 | To write variables on the coarsened grid, and in particular the passive tracers, |
---|
105 | a "nemo_crs" context should be defined in [attachment:iodef.xml iodef.xml] and |
---|
106 | the associated variable are listed in [attachment:file_crs_def.xml file_crs_def.xml ]. |
---|
107 | |
---|
108 | Passive tracers tracers initial conditions |
---|
109 | ------------------------------------------ |
---|
110 | |
---|
111 | When initial conditions are provided in NetCDF files, the field might be: |
---|
112 | |
---|
113 | - on the coarsened grid |
---|
114 | - or they can be on another grid and |
---|
115 | interpolated `on-the-fly <http://forge.ipsl.jussieu.fr/nemo/wiki/Users/SetupNewConfiguration/Weight-creator>`_. |
---|
116 | Example of namelist for PISCES : |
---|
117 | |
---|
118 | .. code-block:: fortran |
---|
119 | |
---|
120 | !----------------------------------------------------------------------- |
---|
121 | &namtrc_dta ! Initialisation from data input file |
---|
122 | !----------------------------------------------------------------------- |
---|
123 | ! |
---|
124 | sn_trcdta(1) = 'DIC_REG1' , -12 , 'DIC' , .false. , .true. , 'yearly' , 'reshape_REG1toeORCA075_bilin.nc' , '' , '' |
---|
125 | sn_trcdta(2) = 'ALK_REG1' , -12 , 'ALK' , .false. , .true. , 'yearly' , 'reshape_REG1toeORCA075_bilin.nc' , '' , '' |
---|
126 | sn_trcdta(3) = 'O2_REG1' , -1 , 'O2' , .true. , .true. , 'yearly' , 'reshape_REG1toeORCA075_bilin.nc' , '' , '' |
---|
127 | sn_trcdta(5) = 'PO4_REG1' , -1 , 'PO4' , .true. , .true. , 'yearly' , 'reshape_REG1toeORCA075_bilin.nc' , '' , '' |
---|
128 | sn_trcdta(7) = 'Si_REG1' , -1 , 'Si' , .true. , .true. , 'yearly' , 'reshape_REG1toeORCA075_bilin.nc' , '' , '' |
---|
129 | sn_trcdta(10) = 'DOC_REG1' , -12 , 'DOC' , .false. , .true. , 'yearly' , 'reshape_REG1toeORCA075_bilin.nc' , '' , '' |
---|
130 | sn_trcdta(14) = 'Fe_REG1' , -12 , 'Fe' , .false. , .true. , 'yearly' , 'reshape_REG1toeORCA075_bilin.nc' , '' , '' |
---|
131 | sn_trcdta(23) = 'NO3_REG1' , -1 , 'NO3' , .true. , .true. , 'yearly' , 'reshape_REG1toeORCA075_bilin.nc' , '' , '' |
---|
132 | rn_trfac(1) = 1.0e-06 ! multiplicative factor |
---|
133 | rn_trfac(2) = 1.0e-06 ! - - - - |
---|
134 | rn_trfac(3) = 44.6e-06 ! - - - - |
---|
135 | rn_trfac(5) = 122.0e-06 ! - - - - |
---|
136 | rn_trfac(7) = 1.0e-06 ! - - - - |
---|
137 | rn_trfac(10) = 1.0e-06 ! - - - - |
---|
138 | rn_trfac(14) = 1.0e-06 ! - - - - |
---|
139 | rn_trfac(23) = 7.6e-06 ! - - - - |
---|
140 | |
---|
141 | cn_dir = './' ! root directory for the location of the data files |
---|
142 | |
---|
143 | PISCES forcing files |
---|
144 | -------------------- |
---|
145 | |
---|
146 | They might be on the coarsened grid. |
---|
147 | |
---|
148 | Perspectives |
---|
149 | ============ |
---|
150 | |
---|
151 | For the future, a few options are on the table to implement coarsening for biogeochemistry in 4.0 and |
---|
152 | future releases. |
---|
153 | Those will be discussed in Autumn 2018 |
---|