3 |
|
|
4 |
\section{NetCDF I/O Integration: MNC} |
\section{NetCDF I/O Integration: MNC} |
5 |
\label{sec:pkg:mnc} |
\label{sec:pkg:mnc} |
6 |
|
\begin{rawhtml} |
7 |
|
<!-- CMIREDIR:package_mnc: --> |
8 |
|
\end{rawhtml} |
9 |
|
|
10 |
The \texttt{mnc} package is a set of convenience routines written to |
The \texttt{mnc} package is a set of convenience routines written to |
11 |
expedite the process of creating, appending, and reading NetCDF files. |
expedite the process of creating, appending, and reading NetCDF files. |
21 |
\begin{rawhtml} </A> \end{rawhtml} |
\begin{rawhtml} </A> \end{rawhtml} |
22 |
|
|
23 |
|
|
24 |
\subsection{Introduction} |
\subsection{Using MNC} |
25 |
|
|
26 |
|
\subsubsection{MNC Configuration and Inputs} |
27 |
|
|
28 |
|
As with all MITgcm packages, MNC can be turned on/off at compile time |
29 |
|
using the \texttt{packages.conf} file or the genmake2 |
30 |
|
\texttt{-enable=mnc} or \texttt{-disable=mnc} switches. |
31 |
|
|
32 |
|
For run-time configuration, most of the MNC--related model parameters |
33 |
|
are contained within a Fortran namelist file called \texttt{data.mnc}. |
34 |
|
If this file does not exist, then the MNC package will interpret that |
35 |
|
as an indication that it is not to be used. If the \texttt{data.mnc} |
36 |
|
file does exist, then it may contain the following parameters: |
37 |
|
|
38 |
|
\begin{center} |
39 |
|
{\footnotesize |
40 |
|
\begin{tabular}[htb]{|l|c|l|l|}\hline |
41 |
|
\textbf{Name} & \textbf{T} & |
42 |
|
\textbf{Default} & \textbf{Description} \\\hline |
43 |
|
& & & \\ |
44 |
|
\texttt{useMNC} & L & \texttt{.FALSE.} & |
45 |
|
\textbf{overall MNC ON/OFF switch} \\ |
46 |
|
\texttt{mnc\_echo\_gvtypes} & L & \texttt{.FALSE.} & |
47 |
|
echo pre-defined ``types'' (debugging) \\ |
48 |
|
\texttt{mnc\_use\_outdir} & L & \texttt{.FALSE.} & |
49 |
|
create a directory for output \\ |
50 |
|
\texttt{mnc\_outdir\_str} & S & \texttt{'mnc\_'} & |
51 |
|
output directory name \\ |
52 |
|
\texttt{mnc\_outdir\_date} & L & \texttt{.FALSE.} & |
53 |
|
embed date in the output dir name \\ |
54 |
|
\texttt{pickup\_write\_mnc} & L & \texttt{.FALSE.} & |
55 |
|
use MNC to write (create) pickup files \\ |
56 |
|
\texttt{pickup\_read\_mnc} & L & \texttt{.FALSE.} & |
57 |
|
use MNC to read pickup files \\ |
58 |
|
\texttt{mnc\_use\_indir} & L & \texttt{.FALSE.} & |
59 |
|
use a directory (path) for input \\ |
60 |
|
\texttt{mnc\_indir\_str} & S & \texttt{''} & |
61 |
|
input directory (or path) name \\ |
62 |
|
\texttt{snapshot\_mnc} & L & \texttt{.FALSE.} & |
63 |
|
write \texttt{snapshot} (instantaneous) w/MNC \\ |
64 |
|
\texttt{monitor\_mnc} & L & \texttt{.FALSE.} & |
65 |
|
write \texttt{monitor} w/MNC \\ |
66 |
|
\texttt{timeave\_mnc} & L & \texttt{.FALSE.} & |
67 |
|
write \texttt{timeave} w/MNC \\ |
68 |
|
\texttt{autodiff\_mnc} & L & \texttt{.FALSE.} & |
69 |
|
write \texttt{autodiff} w/MNC \\\hline |
70 |
|
\end{tabular} |
71 |
|
} |
72 |
|
\end{center} |
73 |
|
|
74 |
|
Additional MNC--related parameters are contained within the main |
75 |
|
\texttt{data} namelist file and in some of the namelist files for |
76 |
|
individual packages. These options are: |
77 |
|
\begin{center} |
78 |
|
{\footnotesize |
79 |
|
\begin{tabular}[htb]{|l|c|l|l|}\hline |
80 |
|
\textbf{Name} & \textbf{T} & |
81 |
|
\textbf{Default} & \textbf{Description} \\\hline |
82 |
|
\multicolumn{4}{|c|}{\ } \\ |
83 |
|
\multicolumn{4}{|c|}{Main namelist file: |
84 |
|
``\textbf{data}''} \\\hline |
85 |
|
\texttt{snapshot\_ioinc} & L & \texttt{.FALSE.} & |
86 |
|
write \texttt{snapshot} ``inclusively'' \\ |
87 |
|
\texttt{timeave\_ioinc} & L & \texttt{.FALSE.} & |
88 |
|
write \texttt{timeave} ``inclusively'' \\ |
89 |
|
\texttt{monitor\_ioinc} & L & \texttt{.FALSE.} & |
90 |
|
write \texttt{monitor} ``inclusively'' \\ |
91 |
|
\texttt{the\_run\_name} & C & ``name...'' & |
92 |
|
name is included in all MNC output \\\hline |
93 |
|
\multicolumn{4}{|c|}{\ } \\ |
94 |
|
\multicolumn{4}{|c|}{Diagnostics namelist file: |
95 |
|
``\textbf{data.diagnostics}''} \\\hline |
96 |
|
\texttt{diag\_mnc} & L & \texttt{.FALSE.} & |
97 |
|
write \texttt{diagnostics} w/MNC \\ |
98 |
|
\texttt{diag\_ioinc} & L & \texttt{.FALSE.} & |
99 |
|
write \texttt{diagnostics} ``inclusively'' \\\hline |
100 |
|
\end{tabular} |
101 |
|
} |
102 |
|
\end{center} |
103 |
|
|
104 |
|
By default, turning on MNC for a particular output type will result in |
105 |
|
turning off all the corresponding (usually, default) MDSIO or STDOUT |
106 |
|
output mechanisms. In other words, output defaults to being an |
107 |
|
exclusive selection. To enable multiple kinds of simultaneous output, |
108 |
|
flags of the form \texttt{NAME\_ioinc} have been created where |
109 |
|
\texttt{NAME} corresponds to the various MNC output flags. When a |
110 |
|
\texttt{NAME\_ioinc} flag is set to \texttt{.TRUE.}, then multiple |
111 |
|
simultaneous forms of output are allowed for the \texttt{NAME} output |
112 |
|
mechanism. The intent of this design is that typical users will only |
113 |
|
want one kind of output while people debugging the code (particularly |
114 |
|
the I/O routines) may want simultaneous types of output. |
115 |
|
|
116 |
|
This ``inclusive'' versus ``exclusive'' design is easily applied in |
117 |
|
cases where three or more kinds of output may be generated. Thus, it |
118 |
|
can be readily extended to additional new output types (eg. HDF5). |
119 |
|
|
120 |
|
Input types are always exclusive. |
121 |
|
|
122 |
|
\subsubsection{MNC Output} |
123 |
|
|
124 |
|
While NetCDF files are supposed to be ``self-describing'', it is |
125 |
|
helpful to note the following: |
126 |
|
|
127 |
|
\begin{itemize} |
128 |
|
\item The constraints placed upon the ``unlimited'' (or ``record'') |
129 |
|
dimension inherent with NetCDF v3.x make it very inefficient to put |
130 |
|
variables written at potentially different intervals within the same |
131 |
|
file. For this reason, MNC output is split into a few file ``base |
132 |
|
names'' which try to reflect the nature of their content. |
133 |
|
|
134 |
|
\item All MNC output is currently done in a ``tile-per-file'' fashion |
135 |
|
since most NetCDF v3.x implementions cannot write safely within MPI |
136 |
|
or multi-threaded environments. This tiling is done in a global |
137 |
|
fashion and the tile numbers are appended to the base names |
138 |
|
described above. Some scripts to ``assemble'' output are available |
139 |
|
(\texttt{MITgcm/utils/matlab}). More general manipulations can be |
140 |
|
accomplished with the |
141 |
|
\begin{rawhtml} |
142 |
|
<A href="http://nco.sourceforge.net"> |
143 |
|
\end{rawhtml} |
144 |
|
\begin{verbatim} |
145 |
|
NetCDF Operators (or ``NCO'') at http://nco.sourceforge.net |
146 |
|
\end{verbatim} |
147 |
|
\begin{rawhtml} </A> \end{rawhtml} |
148 |
|
which is a very powerful and convenient set of tools for working |
149 |
|
with all NetCDF files. |
150 |
|
|
151 |
|
\item MNC does not (yet) provide a mechanism for reading information |
152 |
|
from a single ``global'' file as can be done with the MDSIO |
153 |
|
package. |
154 |
|
|
155 |
|
\end{itemize} |
156 |
|
|
157 |
|
|
158 |
|
\subsection{MNC Internals} |
159 |
|
|
160 |
The \texttt{mnc} package is a two-level convenience library (or |
The \texttt{mnc} package is a two-level convenience library (or |
161 |
``wrapper'') for most of the NetCDF Fortran API. Its purpose is to |
``wrapper'') for most of the NetCDF Fortran API. Its purpose is to |
203 |
\end{description} |
\end{description} |
204 |
|
|
205 |
|
|
206 |
\subsection{Using MNC} |
\subsubsection{MNC Grid--Types and Variable--Types} |
|
|
|
|
\subsubsection{Grid--Types and Variable--Types} |
|
207 |
|
|
208 |
As a convenience for users, the MNC package includes numerous routines |
As a convenience for users, the MNC package includes numerous routines |
209 |
to aid in the writing of data to NetCDF format. Probably the biggest |
to aid in the writing of data to NetCDF format. Probably the biggest |
234 |
\begin{center} |
\begin{center} |
235 |
\begin{tabular}[h]{|ccc|c|c|}\hline |
\begin{tabular}[h]{|ccc|c|c|}\hline |
236 |
\multicolumn{3}{|c|}{Horizontal} & Vertical & Time \\ |
\multicolumn{3}{|c|}{Horizontal} & Vertical & Time \\ |
237 |
\textit{H0}: location & \textit{H1}: dimensions & \textit{H2}: halo |
\textbf{H0}: location & \textbf{H1}: dimensions & \textbf{H2}: halo |
238 |
& \textit{V}: location & \textit{T}: level \\\hline |
& \textbf{V}: location & \textbf{T}: level \\\hline |
239 |
\texttt{-} & xy & Hn & \texttt{-} & \texttt{-} \\ |
\texttt{-} & xy & Hn & \texttt{-} & \texttt{-} \\ |
240 |
U & x & Hy & i & t \\ |
U & x & Hy & i & t \\ |
241 |
V & y & & c & \\ |
V & y & & c & \\ |
249 |
\texttt{pkg/mnc/pre-defined\_grids.txt}. |
\texttt{pkg/mnc/pre-defined\_grids.txt}. |
250 |
\end{center} |
\end{center} |
251 |
|
|
252 |
|
The variable type is an association between a variable type name and the |
253 |
|
following items: |
254 |
|
\begin{center} |
255 |
|
\begin{tabular}[h]{|l|l|}\hline |
256 |
|
\textbf{Item} & \textbf{Purpose} \\\hline |
257 |
|
grid type & defines the in-memory arrangement \\ |
258 |
|
\texttt{bi,bj} dimensions & tiling indices, if present \\\hline |
259 |
|
\end{tabular} |
260 |
|
\end{center} |
261 |
|
and is used by the \texttt{mnc\_cw\_*\_[R|W]} subroutines for reading |
262 |
|
and writing variables. |
263 |
|
|
264 |
|
|
265 |
\subsubsection{An Example} |
\subsubsection{Using MNC: Examples} |
266 |
|
|
267 |
Writing variables to NetCDF files can be accomplished in as few as two |
Writing variables to NetCDF files can be accomplished in as few as two |
268 |
function calls. The first function call defines a variable type, |
function calls. The first function call defines a variable type, |
272 |
necessary, the current time level within the model. |
necessary, the current time level within the model. |
273 |
|
|
274 |
Examples of the initialization calls can be found in the file |
Examples of the initialization calls can be found in the file |
275 |
\filelink{model/src/initialise\_fixed.F}{model-src-initialise_fixed.F} |
\filelink{model/src/ini\_mnc\_io.F}{model-src-ini_mnc_io.F} |
276 |
where these four function calls: |
where these function calls: |
277 |
{\footnotesize |
{\footnotesize |
278 |
\begin{verbatim} |
\begin{verbatim} |
279 |
C Create MNC definitions for DYNVARS.h variables |
C Create MNC definitions for DYNVARS.h variables |
280 |
CALL MNC_CW_ADD_VNAME(myThid, 'iter', '-_-_--__-__t', 0,0) |
CALL MNC_CW_ADD_VNAME('iter', '-_-_--__-__t', 0,0, myThid) |
281 |
CALL MNC_CW_ADD_VATTR_TEXT(myThid,'iter',1, |
CALL MNC_CW_ADD_VATTR_TEXT('iter',1, |
282 |
& 'long_name','iteration_count') |
& 'long_name','iteration_count', myThid) |
283 |
CALL MNC_CW_ADD_VNAME(myThid, 'U', 'U_xy_Hn__C__t', 4,5) |
|
284 |
CALL MNC_CW_ADD_VATTR_TEXT(myThid,'U',1,'units','m/s') |
CALL MNC_CW_ADD_VNAME('model_time', '-_-_--__-__t', 0,0, myThid) |
285 |
|
CALL MNC_CW_ADD_VATTR_TEXT('model_time',1, |
286 |
|
& 'long_name','Model Time', myThid) |
287 |
|
CALL MNC_CW_ADD_VATTR_TEXT('model_time',1,'units','s', myThid) |
288 |
|
|
289 |
|
CALL MNC_CW_ADD_VNAME('U', 'U_xy_Hn__C__t', 4,5, myThid) |
290 |
|
CALL MNC_CW_ADD_VATTR_TEXT('U',1,'units','m/s', myThid) |
291 |
|
CALL MNC_CW_ADD_VATTR_TEXT('U',1, |
292 |
|
& 'coordinates','XU YU RC iter', myThid) |
293 |
|
|
294 |
|
CALL MNC_CW_ADD_VNAME('T', 'Cen_xy_Hn__C__t', 4,5, myThid) |
295 |
|
CALL MNC_CW_ADD_VATTR_TEXT('T',1,'units','degC', myThid) |
296 |
|
CALL MNC_CW_ADD_VATTR_TEXT('T',1,'long_name', |
297 |
|
& 'potential_temperature', myThid) |
298 |
|
CALL MNC_CW_ADD_VATTR_TEXT('T',1, |
299 |
|
& 'coordinates','XC YC RC iter', myThid) |
300 |
\end{verbatim} |
\end{verbatim} |
301 |
} |
} |
302 |
{\noindent initialize two \texttt{VNAME}s and add one NetCDF |
{\noindent initialize four \texttt{VNAME}s and add one or more NetCDF |
303 |
attribute to each.} |
attributes to each.} |
304 |
|
|
305 |
The two variables defined above are subsequently written at specific |
The four variables defined above are subsequently written at specific |
306 |
time steps within |
time steps within |
307 |
\filelink{model/src/write\_state.F}{model-src-write_state.F} |
\filelink{model/src/write\_state.F}{model-src-write_state.F} |
308 |
using the function calls: |
using the function calls: |
309 |
{\footnotesize |
{\footnotesize |
310 |
\begin{verbatim} |
\begin{verbatim} |
311 |
C Write the DYNVARS.h variables using the MNC package |
C Write dynvars using the MNC package |
312 |
mnc_iter = myIter |
CALL MNC_CW_SET_UDIM('state', -1, myThid) |
313 |
CALL MNC_CW_RL_W_R(myThid,'state',0,0,'iter',-1,mnc_iter) |
CALL MNC_CW_I_W('I','state',0,0,'iter', myIter, myThid) |
314 |
CALL MNC_CW_RL_W_D(myThid,'state',0,0,'U', 0, uVel) |
CALL MNC_CW_SET_UDIM('state', 0, myThid) |
315 |
|
CALL MNC_CW_RL_W('D','state',0,0,'model_time',myTime, myThid) |
316 |
|
CALL MNC_CW_RL_W('D','state',0,0,'U', uVel, myThid) |
317 |
|
CALL MNC_CW_RL_W('D','state',0,0,'T', theta, myThid) |
318 |
\end{verbatim} |
\end{verbatim} |
319 |
} |
} |
320 |
|
|
321 |
|
While it is easiest to write variables within typical 2D and 3D fields |
322 |
|
where all data is known at a given time, it is also possible to write |
323 |
|
fields where only a portion (\textit{eg.} a ``slab'' or ``slice'') is |
324 |
|
known at a given instant. An example is provided within |
325 |
|
\filelink{pkg/mom\_vecinv/mom\_vecinv.F}{pkg-mom_vecinv-mom_vecinv.F} |
326 |
|
where an offset vector is used: {\footnotesize |
327 |
|
\begin{verbatim} |
328 |
|
IF (useMNC .AND. snapshot_mnc) THEN |
329 |
|
CALL MNC_CW_RL_W_OFFSET('D','mom_vi',bi,bj, 'fV', uCf, |
330 |
|
& offsets, myThid) |
331 |
|
CALL MNC_CW_RL_W_OFFSET('D','mom_vi',bi,bj, 'fU', vCf, |
332 |
|
& offsets, myThid) |
333 |
|
ENDIF |
334 |
|
\end{verbatim} |
335 |
|
} |
336 |
|
to write a 3D field one depth slice at a time. |
337 |
|
|
338 |
\subsection{Key subroutines, parameters and files} |
Each element in the offset vector corresponds (in order) to the |
339 |
|
dimensions of the ``full'' (or virtual) array and specifies which are |
340 |
All of the variables used to implement the lookup tables are described |
known at the time of the call. A zero within the offset array means |
341 |
in \filelink{model/src/write\_state.F}{model-src-write_state.F} |
that all values along that dimension are available while a positive |
342 |
|
integer means that only values along that index of the dimension are |
343 |
|
available. In all cases, the matrix passed is assumed to start (that |
344 |
|
is, have an in-memory structure) coinciding with the start of the |
345 |
\subsection{Package Reference} |
specified slice. Thus, using this offset array mechanism, a slice |
346 |
|
can be written along any single dimension or combinations of |
347 |
|
dimensions. |
348 |
|
|