1 |
\subsection {RBCS Package} |
2 |
\label{sec:pkg:rbcs} |
3 |
\begin{rawhtml} |
4 |
<!-- CMIREDIR:package_rbcs: --> |
5 |
\end{rawhtml} |
6 |
|
7 |
\subsubsection {Introduction} |
8 |
|
9 |
A package which provides the flexibility |
10 |
to relax fields (temperature, salinity, ptracers) |
11 |
in any 3-D location: |
12 |
so could be used as a sponge layer, or as a |
13 |
"source" anywhere in the domain. |
14 |
|
15 |
\noindent |
16 |
For a tracer ($T$) at every grid point the tendency is modified so that: |
17 |
\[ |
18 |
\frac{dT}{dt}=\frac{dT}{dt} - \frac{M_{rbc}}{\tau_T} (T-T_{rbc}) |
19 |
\] |
20 |
|
21 |
\noindent |
22 |
where $M_{rbc}$ is a 3-D mask (no time dependence) with |
23 |
values between 0 and 1. Where $M_{rbc}$ is 1, relaxing timescale |
24 |
is $1/\tau_T$. Where it is 0 there is no relaxing. |
25 |
The value relaxed to is a 3-D (potentially varying in |
26 |
time) field given by $T_{rbc}$. |
27 |
|
28 |
A seperate mask can be used for T,S and ptracers and |
29 |
each of these |
30 |
can be relaxed or not and can have its own timescale |
31 |
$\tau_T$. These are set in data.rbcs (see below). |
32 |
|
33 |
\subsubsection {Key subroutines and parameters} |
34 |
|
35 |
The only change need in the code might be in {RBCS.h}, for |
36 |
PARAMETER(maskLEN = 3 ), if you need more than 3 |
37 |
masks (see below). |
38 |
|
39 |
\vspace{.5cm} |
40 |
|
41 |
\noindent |
42 |
There are runtime parameters |
43 |
set in {\it data.rbcs}:\\ |
44 |
These runtime options include\\ |
45 |
Set in {RBCS\_PARM01}:\\ |
46 |
%$\bullet$ Parameters to set the timing for periodic fields to |
47 |
%relax to are to |
48 |
%be loaded are: |
49 |
$\bullet$ {\bf rbcsForcingPeriod}, {\bf rbcsForcingCycle}: timing of |
50 |
fields to relax to. |
51 |
The former is how often to load, the latter is how often to cycle |
52 |
through those fields (eg. period could be monthly and cycle one year). |
53 |
rbcs\_ForcingCycle=0 means non-cyclic forcing, and |
54 |
rbcs\_ForcingPeriod=0 non-time-varying forcing, where the relax field |
55 |
is only read in at the beginning of the run and kept constant |
56 |
the rest of the run. Default is 0. |
57 |
\\ |
58 |
$\bullet$ {\bf rbcsForcingOffset}: if you want to offset rbcs forcing |
59 |
timing. This is the time at the beginning of the first forcing period |
60 |
(in seconds). Default is 0. (This used to be rbcsInIter and was in |
61 |
units of iterations.)\\ |
62 |
$\bullet$ {\bf rbcsSingleTimeFiles}: if true, relax fields are given |
63 |
1 file per time labeled by iteration number.\\ |
64 |
$\bullet$ {\bf deltaTrbcs}: time step used to compute iteration numbers |
65 |
for rbcsSingleTimeFiles.\\ |
66 |
$\bullet$ {\bf rbcsIter0}: rbcsSingleTimeFiles iteration number |
67 |
corresponding to rbcsForcingOffset.\\ |
68 |
$\bullet$ {\bf useRBCtemp}: true or false (default false)\\ |
69 |
$\bullet$ {\bf useRBCsalt}: true or false (default false)\\ |
70 |
$\bullet$ {\bf useRBCptracers}: true or false (default false), must be using |
71 |
ptracers to set true\\ |
72 |
$\bullet$ {\bf tauRelaxT}: timescale in seconds of relaxing |
73 |
in temperature ($\tau_T$ in equation above). |
74 |
Where mask is 1, relax rate will be |
75 |
1/tauRelaxT. Default is 1. |
76 |
$\bullet$ {\bf tauRelaxS}: same for salinity. |
77 |
$\bullet$ {\bf relaxMaskFile(irbc)}: filename of 3-D file |
78 |
with mask ($M_{rbc}$ in equation above. |
79 |
Need a file for each irbc. 1=temperature, |
80 |
2=salinity, 3=ptracer01, 4=ptracer02 etc. If the mask numbers |
81 |
end (see maskLEN) are less than the number tracers, then |
82 |
relaxMaskFile(maskLEN) is used for all remaining ptracers.\\ |
83 |
$\bullet$ {\bf relaxTFile}: name of file where temperatures |
84 |
that need to be realxed to ($T_{rbc}$ in equation above) |
85 |
are stored. Need 3-D fields to |
86 |
match model domain, and as many entries as given by |
87 |
rbcsForcingPeriod and rbcsForcingCycle.\\ |
88 |
$\bullet$ {\bf relaxSFile}: same for salinity.\\ |
89 |
|
90 |
\vspace{.5cm} |
91 |
\noindent |
92 |
Set in {RBCS\_PARM02} for each of the ptracers (iTrc):\\ |
93 |
$\bullet$ {\bf useRBCptrnum(iTrc)}: true or false (default |
94 |
is false).\\ |
95 |
$\bullet$ {\bf tauRelaxPTR(iTrc)}: relax timescale.\\ |
96 |
$\bullet$ {\bf relaxPtracerFile(iTrc)}: file with relax |
97 |
fields.\\ |
98 |
|
99 |
\noindent |
100 |
Typical ways of specifying timing of relaxation fields: |
101 |
\begin{enumerate} |
102 |
\item Constant-in-time forcing: |
103 |
\begin{quote} |
104 |
rbcsForcingPeriod = 0 |
105 |
\end{quote} |
106 |
One field is read and used for all times. Use this to emulate the result of |
107 |
rbcsForcingCycle=0 before 2010-11-10. |
108 |
|
109 |
\item Non-cyclic time-varying forcing: |
110 |
\begin{quote} |
111 |
rbcsForcingPeriod = period in seconds\\ |
112 |
rbcsForcingCycle = 0\\ |
113 |
rbcsForcingOffset = beginning of first forcing period in seconds (default 0) |
114 |
\end{quote} |
115 |
When starting the run at time 0 (as usually the case), a period with center before |
116 |
or at time 0 is needed for time interpolation, hence Offset needs to be negative. |
117 |
For aligned periods (one period starting at time 0) and one extra record before |
118 |
time 0 (and ending at time 0), set Offset=-Period. |
119 |
|
120 |
This option makes most sense with rbcsSingleTimeFiles=T (see below), but may be |
121 |
used with either setting. |
122 |
|
123 |
\item Cyclic Forcing: |
124 |
\begin{quote} |
125 |
rbcsForcingPeriod = period in seconds\\ |
126 |
rbcsForcingCycle = cycle in seconds\\ |
127 |
rbcsForcingOffset = beginning of first forcing period in seconds (default 0) |
128 |
\end{quote} |
129 |
Here Offset may be at or after the start time (actually starttime-period/2), |
130 |
in which case records from the end of the cycle are used for missing records at |
131 |
the beginning (via cyclicity). Not very nice, but works. |
132 |
|
133 |
This is the traditional way used with rbcsSingleTimeFiles=F, but again may be |
134 |
used with either setting. |
135 |
\end{enumerate} |
136 |
|
137 |
\noindent |
138 |
Ways to organize the files: |
139 |
\begin{enumerate} |
140 |
\item One big file with many time records: |
141 |
\begin{quote} |
142 |
rbcsSingleTimeFiles = .False. |
143 |
\end{quote} |
144 |
All time records are in one big file. |
145 |
|
146 |
\item A separate file for each time: |
147 |
\begin{quote} |
148 |
rbcsSingleTimeFiles = .True.\\ |
149 |
deltaTrbcs = time step used to generate forcing files\\ |
150 |
rbcsIter0 = forcing file iteration number corresponding to rbcsForcingOffset |
151 |
\end{quote} |
152 |
The rbcs field for each time needed is in a separate file, labeled by the |
153 |
iteration number at the end of the forcing period. If a different timestep |
154 |
was used for generating the files (and the file names), set deltaTrbcs. |
155 |
If there is a shift in time, set rbcsIter0. |
156 |
\end{enumerate} |
157 |
|
158 |
|
159 |
|
160 |
\subsubsection{Do's and Don'ts} |
161 |
|
162 |
\subsubsection{Reference Material} |
163 |
|
164 |
\subsubsection{Experiments and tutorials that use rbcs} |
165 |
\label{sec:pkg:rbcs:experiments} |
166 |
|
167 |
|
168 |
|
169 |
%%% \end{itemize} |
170 |
|