New URL for NEMO forge!   http://forge.nemo-ocean.eu

Since March 2022 along with NEMO 4.2 release, the code development moved to a self-hosted GitLab.
This present forge is now archived and remained online for history.
Chap_MISC.tex in branches/2016/dev_r6522_SIMPLIF_4/DOC/TexFiles/Chapters – NEMO

source: branches/2016/dev_r6522_SIMPLIF_4/DOC/TexFiles/Chapters/Chap_MISC.tex @ 6713

Last change on this file since 6713 was 6713, checked in by lovato, 8 years ago

#1730 - trunk: Update manual and history in header of lib_fortran

File size: 17.8 KB
Line 
1% ================================================================
2% Chapter ——— Miscellaneous Topics
3% ================================================================
4\chapter{Miscellaneous Topics}
5\label{MISC}
6\minitoc
7
8\newpage
9$\ $\newline    % force a new ligne
10
11% ================================================================
12% Representation of Unresolved Straits
13% ================================================================
14\section{Representation of Unresolved Straits}
15\label{MISC_strait}
16
17In climate modeling, it often occurs that a crucial connections between water masses
18is broken as the grid mesh is too coarse to resolve narrow straits. For example, coarse
19grid spacing typically closes off the Mediterranean from the Atlantic at the Strait of
20Gibraltar. In this case, it is important for climate models to include the effects of salty
21water entering the Atlantic from the Mediterranean. Likewise, it is important for the
22Mediterranean to replenish its supply of water from the Atlantic to balance the net
23evaporation occurring over the Mediterranean region. This problem occurs even in
24eddy permitting simulations. For example, in ORCA 1/4\deg several straits of the Indonesian
25archipelago (Ombai, Lombok...) are much narrow than even a single ocean grid-point.
26
27We describe briefly here the three methods that can be used in \NEMO to handle
28such improperly resolved straits. The first two consist of opening the strait by hand
29while ensuring that the mass exchanges through the strait are not too large by
30either artificially reducing the surface of the strait grid-cells or, locally increasing
31the lateral friction. In the third one, the strait is closed but exchanges of mass,
32heat and salt across the land are allowed.
33Note that such modifications are so specific to a given configuration that no attempt
34has been made to set them in a generic way. However, examples of how
35they can be set up is given in the ORCA 2\deg and 0.5\deg configurations. For example,
36for details of implementation in ORCA2, search:
37\texttt{ IF( cp\_cfg == "orca" .AND. jp\_cfg == 2 ) }
38
39% -------------------------------------------------------------------------------------------------------------
40%       Hand made geometry changes
41% -------------------------------------------------------------------------------------------------------------
42\subsection{Hand made geometry changes}
43\label{MISC_strait_hand}
44
45$\bullet$ reduced scale factor in the cross-strait direction to a value in better agreement
46with the true mean width of the strait. (Fig.~\ref{Fig_MISC_strait_hand}).
47This technique is sometime called "partially open face" or "partially closed cells".
48The key issue here is only to reduce the faces of $T$-cell ($i.e.$ change the value
49of the horizontal scale factors at $u$- or $v$-point) but not the volume of the $T$-cell.
50Indeed, reducing the volume of strait $T$-cell can easily produce a numerical
51instability at that grid point that would require a reduction of the model time step.
52The changes associated with strait management are done in \mdl{domhgr},
53just after the definition or reading of the horizontal scale factors.
54
55$\bullet$ increase of the viscous boundary layer thickness by local increase of the
56fmask value at the coast (Fig.~\ref{Fig_MISC_strait_hand}). This is done in
57\mdl{dommsk} together with the setting of the coastal value of fmask
58(see Section \ref{LBC_coast})
59
60%>>>>>>>>>>>>>>>>>>>>>>>>>>>>
61\begin{figure}[!tbp]     \begin{center}
62\includegraphics[width=0.80\textwidth]{./TexFiles/Figures/Fig_Gibraltar.pdf}
63\includegraphics[width=0.80\textwidth]{./TexFiles/Figures/Fig_Gibraltar2.pdf}
64\caption{   \label{Fig_MISC_strait_hand} 
65Example of the Gibraltar strait defined in a $1\deg \times 1\deg$ mesh.
66\textit{Top}: using partially open cells. The meridional scale factor at $v$-point
67is reduced on both sides of the strait to account for the real width of the strait
68(about 20 km). Note that the scale factors of the strait $T$-point remains unchanged.
69\textit{Bottom}: using viscous boundary layers. The four fmask parameters
70along the strait coastlines are set to a value larger than 4, $i.e.$ "strong" no-slip
71case (see Fig.\ref{Fig_LBC_shlat}) creating a large viscous boundary layer
72that allows a reduced transport through the strait.}
73\end{center}   \end{figure}
74%>>>>>>>>>>>>>>>>>>>>>>>>>>>>
75
76
77% ================================================================
78% Closed seas
79% ================================================================
80\section{Closed seas (\mdl{closea})}
81\label{MISC_closea}
82
83\colorbox{yellow}{Add here a short description of the way closed seas are managed}
84
85
86% ================================================================
87% Sub-Domain Functionality (\textit{nizoom, njzoom}, namelist parameters)
88% ================================================================
89\section{Sub-Domain Functionality (\np{jpizoom}, \np{jpjzoom})}
90\label{MISC_zoom}
91
92The sub-domain functionality, also improperly called the zoom option
93(improperly because it is not associated with a change in model resolution)
94is a quite simple function that allows a simulation over a sub-domain of an
95already defined configuration ($i.e.$ without defining a new mesh, initial
96state and forcings). This option can be useful for testing the user settings
97of surface boundary conditions, or the initial ocean state of a huge ocean
98model configuration while having a small computer memory requirement.
99It can also be used to easily test specific physics in a sub-domain (for example,
100see \citep{Madec_al_JPO96} for a test of the coupling used in the global ocean
101version of OPA between sea-ice and ocean model over the Arctic or Antarctic
102ocean, using a sub-domain). In the standard model, this option does not
103include any specific treatment for the ocean boundaries of the sub-domain:
104they are considered as artificial vertical walls. Nevertheless, it is quite easy
105to add a restoring term toward a climatology in the vicinity of such boundaries
106(see \S\ref{TRA_dmp}).
107
108In order to easily define a sub-domain over which the computation can be
109performed, the dimension of all input arrays (ocean mesh, bathymetry,
110forcing, initial state, ...) are defined as \np{jpidta}, \np{jpjdta} and \np{jpkdta} 
111( in \ngn{namcfg} namelist), while the computational domain is defined through
112\np{jpiglo}, \np{jpjglo} and \jp{jpk} (\ngn{namcfg} namelist). When running the
113model over the whole domain, the user sets \np{jpiglo}=\np{jpidta} \np{jpjglo}=\np{jpjdta} 
114and \jp{jpk}=\jp{jpkdta}. When running the model over a sub-domain, the user
115has to provide the size of the sub-domain, (\np{jpiglo}, \np{jpjglo}, \np{jpkglo}),
116and the indices of the south western corner as \np{jpizoom} and \np{jpjzoom} in
117the  \ngn{namcfg} namelist (Fig.~\ref{Fig_LBC_zoom}).
118
119Note that a third set of dimensions exist, \jp{jpi}, \jp{jpj} and \jp{jpk} which is
120actually used to perform the computation. It is set by default to \jp{jpi}=\np{jpjglo} 
121and \jp{jpj}=\np{jpjglo}, except for massively parallel computing where the
122computational domain is laid out on local processor memories following a 2D
123horizontal splitting. % (see {\S}IV.2-c) ref to the section to be updated
124
125\subsection{Simple subsetting of input files via netCDF attributes}
126
127The extended grids for use with the under-shelf ice cavities will result in redundant rows
128around Antarctica if the ice cavities are not active. A simple mechanism for subsetting
129input files associated with the extended domains has been implemented to avoid the need to
130maintain different sets of input fields for use with or without active ice cavities. The
131existing 'zoom' options are overly complex for this task and marked for deletion anyway.
132This alternative subsetting operates for the j-direction only and works by optionally
133looking for and using a global file attribute (named: \np{open\_ocean\_jstart}) to
134determine the starting j-row for input. The use of this option is best explained with an
135example: Consider an ORCA1 configuration using the extended grid bathymetry and coordinate
136files:
137\vspace{-10pt}
138\begin{alltt}
139\tiny
140\begin{verbatim}
141eORCA1_bathymetry_v2.nc
142eORCA1_coordinates.nc
143\end{verbatim}
144\end{alltt}
145\noindent These files define a horizontal domain of 362x332. Assuming the first row with
146open ocean wet points in the non-isf bathymetry for this set is row 42 (Fortran indexing)
147then the formally correct setting for \np{open\_ocean\_jstart} is 41. Using this value as the
148first row to be read will result in a 362x292 domain which is the same size as the original
149ORCA1 domain. Thus the extended coordinates and bathymetry files can be used with all the
150original input files for ORCA1 if the ice cavities are not active (\np{ln\_isfcav =
151.false.}). Full instructions for achieving this are:
152
153\noindent Add the new attribute to any input files requiring a j-row offset, i.e:
154\vspace{-10pt}
155\begin{alltt}
156\tiny
157\begin{verbatim}
158ncatted  -a open_ocean_jstart,global,a,d,41 eORCA1_coordinates.nc
159ncatted  -a open_ocean_jstart,global,a,d,41 eORCA1_bathymetry_v2.nc
160\end{verbatim}
161\end{alltt}
162 
163\noindent Add the logical switch to \ngn{namcfg} in the configuration namelist and set true:
164%--------------------------------------------namcfg--------------------------------------------------------
165\namdisplay{namcfg_orca1}
166%--------------------------------------------------------------------------------------------------------------
167
168\noindent Note the j-size of the global domain is the (extended j-size minus
169\np{open\_ocean\_jstart} + 1 ) and this must match the size of all datasets other than
170bathymetry and coordinates currently. However the option can be extended to any global, 2D
171and 3D, netcdf, input field by adding the:
172\vspace{-10pt}
173\begin{alltt}
174\tiny
175\begin{verbatim}
176lrowattr=ln_use_jattr
177\end{verbatim}
178\end{alltt}
179optional argument to the appropriate \np{iom\_get} call and the \np{open\_ocean\_jstart} attribute to the corresponding input files. It remains the users responsibility to set \np{jpjdta} and \np{jpjglo} values in the \np{namelist\_cfg} file according to their needs.
180
181%>>>>>>>>>>>>>>>>>>>>>>>>>>>>
182\begin{figure}[!ht]    \begin{center}
183\includegraphics[width=0.90\textwidth]{./TexFiles/Figures/Fig_LBC_zoom.pdf}
184\caption{   \label{Fig_LBC_zoom}
185Position of a model domain compared to the data input domain when the zoom functionality is used.}
186\end{center}   \end{figure}
187%>>>>>>>>>>>>>>>>>>>>>>>>>>>>
188
189
190% ================================================================
191% Accuracy and Reproducibility
192% ================================================================
193\section{Accuracy and Reproducibility (\mdl{lib\_fortran})}
194\label{MISC_fortran}
195
196\subsection{Issues with intrinsinc SIGN function (\key{nosignedzero})}
197\label{MISC_sign}
198
199The SIGN(A, B) is the \textsc {Fortran} intrinsic function delivers the magnitude
200of A with the sign of B. For example, SIGN(-3.0,2.0) has the value 3.0.
201The problematic case is when the second argument is zero, because, on platforms
202that support IEEE arithmetic, zero is actually a signed number.
203There is a positive zero and a negative zero.
204
205In \textsc{Fortran}~90, the processor was required always to deliver a positive result for SIGN(A, B)
206if B was zero. Nevertheless, in \textsc{Fortran}~95, the processor is allowed to do the correct thing
207and deliver ABS(A) when B is a positive zero and -ABS(A) when B is a negative zero.
208This change in the specification becomes apparent only when B is of type real, and is zero,
209and the processor is capable of distinguishing between positive and negative zero,
210and B is negative real zero. Then SIGN delivers a negative result where, under \textsc{Fortran}~90
211rules,  it used to return a positive result.
212This change may be especially sensitive for the ice model, so we overwrite the intrinsinc
213function with our own function simply performing :   \\
214\verb?   IF( B >= 0.e0 ) THEN   ;   SIGN(A,B) = ABS(A)  ?    \\
215\verb?   ELSE                   ;   SIGN(A,B) =-ABS(A)     ?  \\
216\verb?   ENDIF    ? \\
217This feature can be found in \mdl{lib\_fortran} module and is effective when \key{nosignedzero}
218is defined. We use a CPP key as the overwritting of a intrinsic function can present
219performance issues with some computers/compilers.
220
221
222\subsection{MPP reproducibility}
223\label{MISC_glosum}
224
225The numerical reproducibility of simulations on distributed memory parallel computers
226is a critical issue. In particular, within NEMO global summation of distributed arrays
227is most susceptible to rounding errors, and their propagation and accumulation cause
228uncertainty in final simulation reproducibility on different numbers of processors.
229To avoid so, based on \citet{He_Ding_JSC01} review of different technics,
230we use a so called self-compensated summation method. The idea is to estimate
231the roundoff error, store it in a buffer, and then add it back in the next addition.
232
233Suppose we need to calculate $b = a_1 + a_2 + a_3$. The following algorithm
234will allow to split the sum in two ($sum_1 = a_{1} + a_{2}$ and $b = sum_2 = sum_1 + a_3$)
235with exactly the same rounding errors as the sum performed all at once.
236\begin{align*}
237   sum_1 \ \  &= a_1 + a_2 \\
238   error_1     &= a_2 + ( a_1 - sum_1 ) \\
239   sum_2 \ \  &= sum_1 + a_3 + error_1 \\
240   error_2     &= a_3 + error_1 + ( sum_1 - sum_2 ) \\
241   b \qquad \ &= sum_2 \\
242\end{align*}
243An example of this feature can be found in \mdl{lib\_fortran} module.
244It is systematicallt used in glob\_sum function (summation over the entire basin excluding
245duplicated rows and columns due to cyclic or north fold boundary condition as well as
246overlap MPP areas). The self-compensated summation method should be used in all summation
247in i- and/or j-direction. See closea.F90 module for an example.
248Note also that this implementation may be sensitive to the optimization level.
249
250\subsection{MPP scalability}
251\label{MISC_mppsca}
252
253The default method of communicating values across the north-fold in distributed memory applications
254(\key{mpp\_mpi}) uses a \textsc{MPI\_ALLGATHER} function to exchange values from each processing
255region in the northern row with every other processing region in the northern row. This enables a
256global width array containing the top 4 rows to be collated on every northern row processor and then
257folded with a simple algorithm. Although conceptually simple, this "All to All" communication will
258hamper performance scalability for large numbers of northern row processors. From version 3.4
259onwards an alternative method is available which only performs direct "Peer to Peer" communications
260between each processor and its immediate "neighbours" across the fold line. This is achieved by
261using the default \textsc{MPI\_ALLGATHER} method during initialisation to help identify the "active"
262neighbours. Stored lists of these neighbours are then used in all subsequent north-fold exchanges to
263restrict exchanges to those between associated regions. The collated global width array for each
264region is thus only partially filled but is guaranteed to be set at all the locations actually
265required by each individual for the fold operation. This alternative method should give identical
266results to the default \textsc{ALLGATHER} method and is recommended for large values of \np{jpni}.
267The new method is activated by setting \np{ln\_nnogather} to be true ({\bf nammpp}). The
268reproducibility of results using the two methods should be confirmed for each new, non-reference
269configuration.
270
271% ================================================================
272% Model optimisation, Control Print and Benchmark
273% ================================================================
274\section{Model Optimisation, Control Print and Benchmark}
275\label{MISC_opt}
276%--------------------------------------------namctl-------------------------------------------------------
277\namdisplay{namctl} 
278%--------------------------------------------------------------------------------------------------------------
279
280 \gmcomment{why not make these bullets into subsections?}
281Options are defined through the  \ngn{namctl} namelist variables.
282
283$\bullet$ Vector optimisation:
284
285\key{vectopt\_loop} enables the internal loops to collapse. This is very
286a very efficient way to increase the length of vector calculations and thus
287to speed up the model on vector computers.
288 
289% Add here also one word on NPROMA technique that has been found useless, since compiler have made significant progress during the last decade.
290 
291% Add also one word on NEC specific optimisation (Novercheck option for example)
292 
293$\bullet$ Control print %: describe here 4 things:
294
2951- \np{ln\_ctl} : compute and print the trends averaged over the interior domain
296in all TRA, DYN, LDF and ZDF modules. This option is very helpful when
297diagnosing the origin of an undesired change in model results.
298
2992- also \np{ln\_ctl} but using the nictl and njctl namelist parameters to check
300the source of differences between mono and multi processor runs.
301
302%%gm   to be removed both here and in the code
3033- last digit comparison (\np{nn\_bit\_cmp}). In an MPP simulation, the computation of
304a sum over the whole domain is performed as the summation over all processors of
305each of their sums over their interior domains. This double sum never gives exactly
306the same result as a single sum over the whole domain, due to truncation differences.
307The "bit comparison" option has been introduced in order to be able to check that
308mono-processor and multi-processor runs give exactly the same results.
309%THIS is to be updated with the mpp_sum_glo  introduced in v3.3
310% nn_bit_cmp  today only check that the nn_cla = 0 (no cross land advection)
311%%gm end
312
313$\bullet$  Benchmark (\np{nn\_bench}). This option defines a benchmark run based on
314a GYRE configuration (see \S\ref{CFG_gyre}) in which the resolution remains the same
315whatever the domain size. This allows a very large model domain to be used, just by
316changing the domain size (\jp{jpiglo}, \jp{jpjglo}) and without adjusting either the time-step
317or the physical parameterisations.
318
319% ================================================================
320
321
322
323
324
Note: See TracBrowser for help on using the repository browser.