1 |
C $Header: /u/gcmpack/MITgcm/pkg/ctrl/ctrl_obcsvol.F,v 1.11 2009/04/28 18:09:28 jmc Exp $ |
2 |
C $Name: $ |
3 |
|
4 |
#include "CTRL_CPPOPTIONS.h" |
5 |
#ifdef ALLOW_OBCS |
6 |
# include "OBCS_OPTIONS.h" |
7 |
#endif |
8 |
|
9 |
subroutine ctrl_obcsvol( |
10 |
I mytime, |
11 |
I myiter, |
12 |
I mythid |
13 |
& ) |
14 |
|
15 |
c ================================================================== |
16 |
c SUBROUTINE ctrl_obcsvol |
17 |
c ================================================================== |
18 |
c |
19 |
c o volumetrically balance the control vector contribution. |
20 |
c o Assume the calendar is identical |
21 |
c for all open boundaries. Need to save the barotropic adjustment |
22 |
c velocity so it can be used in all ctrl_getobcs files. |
23 |
c ================================================================== |
24 |
c SUBROUTINE ctrl_obcsvol |
25 |
c ================================================================== |
26 |
|
27 |
implicit none |
28 |
|
29 |
c == global variables == |
30 |
|
31 |
#include "EEPARAMS.h" |
32 |
#include "SIZE.h" |
33 |
#include "PARAMS.h" |
34 |
#include "GRID.h" |
35 |
#include "DYNVARS.h" |
36 |
#ifdef ALLOW_OBCS |
37 |
# include "OBCS.h" |
38 |
#endif |
39 |
|
40 |
#include "ctrl.h" |
41 |
#include "ctrl_dummy.h" |
42 |
#include "optim.h" |
43 |
|
44 |
c == routine arguments == |
45 |
|
46 |
integer myiter |
47 |
_RL mytime |
48 |
integer mythid |
49 |
|
50 |
#ifdef BALANCE_CONTROL_VOLFLUX_GLOBAL |
51 |
c == local variables == |
52 |
|
53 |
integer bi,bj |
54 |
integer i,j,k |
55 |
integer itlo,ithi |
56 |
integer jtlo,jthi |
57 |
integer jmin,jmax |
58 |
integer imin,imax |
59 |
integer irec |
60 |
integer il |
61 |
integer iobcs |
62 |
integer ip1 |
63 |
integer jp1 |
64 |
integer nrec |
65 |
integer ilfld |
66 |
integer igg |
67 |
|
68 |
_RL sumvol |
69 |
_RL sumarea |
70 |
_RL tmpflux |
71 |
_RL tmparea |
72 |
_RL dummy |
73 |
_RL gg |
74 |
_RL tmpx |
75 |
_RL tmpy |
76 |
character*(80) fnamefldn |
77 |
character*(80) fnameflds |
78 |
character*(80) fnamefldw |
79 |
character*(80) fnameflde |
80 |
|
81 |
logical doglobalread |
82 |
logical ladinit |
83 |
logical obcsnfirst, obcsnchanged |
84 |
integer obcsncount0, obcsncount1 |
85 |
_RL obcsnfac |
86 |
|
87 |
#ifdef ECCO_VERBOSE |
88 |
character*(MAX_LEN_MBUF) msgbuf |
89 |
#endif |
90 |
|
91 |
c == external functions == |
92 |
|
93 |
integer ilnblnk |
94 |
external ilnblnk |
95 |
|
96 |
c == end of interface == |
97 |
|
98 |
jtlo = mybylo(mythid) |
99 |
jthi = mybyhi(mythid) |
100 |
itlo = mybxlo(mythid) |
101 |
ithi = mybxhi(mythid) |
102 |
jmin = 1 |
103 |
jmax = sny |
104 |
imin = 1 |
105 |
imax = snx |
106 |
|
107 |
c-- Read tiled data. |
108 |
doglobalread = .false. |
109 |
ladinit = .false. |
110 |
|
111 |
cgg Assume the number of records is the same for |
112 |
cgg all boundaries. |
113 |
|
114 |
tmpflux= 0. d 0 |
115 |
tmparea= 0. d 0 |
116 |
sumarea= 0. d 0 |
117 |
sumvol = 0. d 0 |
118 |
|
119 |
#ifdef ECCO_VERBOSE |
120 |
_BEGIN_MASTER( mythid ) |
121 |
write(msgbuf,'(a)') ' ' |
122 |
call print_message( msgbuf, standardmessageunit, |
123 |
& SQUEEZE_RIGHT , mythid) |
124 |
write(msgbuf,'(a)') ' ' |
125 |
call print_message( msgbuf, standardmessageunit, |
126 |
& SQUEEZE_RIGHT , mythid) |
127 |
write(msgbuf,'(a,i9.8)') |
128 |
& ' ctrl_obcsvol: number of records to process: ',nrec |
129 |
call print_message( msgbuf, standardmessageunit, |
130 |
& SQUEEZE_RIGHT , mythid) |
131 |
write(msgbuf,'(a)') ' ' |
132 |
call print_message( msgbuf, standardmessageunit, |
133 |
& SQUEEZE_RIGHT , mythid) |
134 |
_END_MASTER( mythid ) |
135 |
#endif |
136 |
|
137 |
c-- Get the counters, flags, and the interpolation factor. |
138 |
call ctrl_GetRec( 'xx_obcsn', |
139 |
O obcsnfac, obcsnfirst, obcsnchanged, |
140 |
O obcsncount0,obcsncount1, |
141 |
I mytime, myiter, mythid ) |
142 |
|
143 |
if (optimcycle .ge. 0) then |
144 |
c |
145 |
ilfld=ilnblnk( xx_obcsn_file ) |
146 |
write(fnamefldn(1:80),'(2a,i10.10)') |
147 |
& xx_obcsn_file(1:ilfld),'.', optimcycle |
148 |
ilfld=ilnblnk( xx_obcss_file ) |
149 |
write(fnameflds(1:80),'(2a,i10.10)') |
150 |
& xx_obcss_file(1:ilfld),'.',optimcycle |
151 |
ilfld=ilnblnk( xx_obcsw_file ) |
152 |
write(fnamefldw(1:80),'(2a,i10.10)') |
153 |
& xx_obcsw_file(1:ilfld),'.',optimcycle |
154 |
ilfld=ilnblnk( xx_obcse_file ) |
155 |
write(fnameflde(1:80),'(2a,i10.10)') |
156 |
& xx_obcse_file(1:ilfld),'.',optimcycle |
157 |
c |
158 |
endif |
159 |
|
160 |
c-- Loop over records. For north boundary, we only need V velocity. |
161 |
|
162 |
if ( obcsnfirst ) then |
163 |
|
164 |
shiftvel(1) = 0. d0 |
165 |
shiftvel(2) = 0. d0 |
166 |
|
167 |
#ifdef ALLOW_OBCSN_CONTROL |
168 |
jp1 = 0 |
169 |
|
170 |
call active_read_xz(fnamefldn,tmpfldxz, |
171 |
& (obcsncount0-1)*nobcs+4, doglobalread, |
172 |
& ladinit, optimcycle, mythid |
173 |
& , xx_obcsn_dummy ) |
174 |
|
175 |
c-- Loop over this thread tiles. |
176 |
do bj = jtlo,jthi |
177 |
do bi = itlo,ithi |
178 |
|
179 |
tmpflux = 0. d0 |
180 |
tmparea = 0. d0 |
181 |
|
182 |
do k = 1, Nr |
183 |
do i = imin,imax |
184 |
j = Ob_Jn(I,bi,bj) |
185 |
if (maskS(i,j+jp1,k,bi,bj) .ne. 0.) then |
186 |
cgg -- Do not let the corners contribute to the volume flux. |
187 |
if(ob_iw(j,bi,bj).ne.i .and. ob_ie(j,bi,bj).ne.i) then |
188 |
tmpx = tmpfldxz(i,k,bi,bj) |
189 |
cgg -- Positive is flux in. |
190 |
tmpflux = tmpflux - tmpx* delR(k)*dxg(i,j+jp1,bi,bj) |
191 |
tmparea = tmparea + delR(k) * dxg(i,j+jp1,bi,bj) |
192 |
endif |
193 |
endif |
194 |
enddo |
195 |
enddo |
196 |
|
197 |
sumarea = sumarea+ tmparea |
198 |
sumvol = sumvol + tmpflux |
199 |
enddo |
200 |
enddo |
201 |
#endif |
202 |
|
203 |
#ifdef ALLOW_OBCSS_CONTROL |
204 |
jp1 = 1 |
205 |
|
206 |
call active_read_xz(fnameflds,tmpfldxz, |
207 |
& (obcsncount0-1)*nobcs+4, doglobalread, |
208 |
& ladinit, optimcycle, mythid |
209 |
& , xx_obcss_dummy ) |
210 |
|
211 |
c-- Loop over this thread tiles. |
212 |
do bj = jtlo,jthi |
213 |
do bi = itlo,ithi |
214 |
|
215 |
tmpflux = 0. d 0 |
216 |
tmparea = 0. d 0 |
217 |
|
218 |
do k = 1, Nr |
219 |
do i = imin,imax |
220 |
j = Ob_Js(I,bi,bj) |
221 |
if (maskS(i,j+jp1,k,bi,bj) .ne. 0.) then |
222 |
cgg -- Do not let the corners contribute to the volume flux. |
223 |
if (ob_iw(j,bi,bj).ne.i .and.ob_ie(j,bi,bj).ne.i) then |
224 |
tmpx = tmpfldxz(i,k,bi,bj) |
225 |
cgg -- Positive is flux in. |
226 |
tmpflux = tmpflux + tmpx* delR(k)*dxg(i,j+jp1,bi,bj) |
227 |
tmparea = tmparea + delR(k) * dxg(i,j+jp1,bi,bj) |
228 |
endif |
229 |
endif |
230 |
enddo |
231 |
enddo |
232 |
sumarea = sumarea+ tmparea |
233 |
sumvol = sumvol + tmpflux |
234 |
enddo |
235 |
enddo |
236 |
#endif |
237 |
|
238 |
#ifdef ALLOW_OBCSW_CONTROL |
239 |
ip1 = 1 |
240 |
|
241 |
call active_read_yz( fnamefldw, tmpfldyz, |
242 |
& (obcsncount0-1)*nobcs+3, doglobalread, |
243 |
& ladinit, optimcycle, mythid |
244 |
& , xx_obcsw_dummy ) |
245 |
|
246 |
c-- Loop over this thread tiles. |
247 |
do bj = jtlo,jthi |
248 |
do bi = itlo,ithi |
249 |
|
250 |
c-- Determine the weights to be used. |
251 |
tmpflux = 0. d 0 |
252 |
tmparea = 0. d 0 |
253 |
|
254 |
do k = 1, Nr |
255 |
do j = jmin,jmax |
256 |
i = ob_iw(j,bi,bj) |
257 |
if (maskW(i+ip1,j,k,bi,bj) .ne. 0.) then |
258 |
cgg -- Do not let the corners contribute to the volume flux. |
259 |
if (ob_jn(i,bi,bj).ne.j .and. ob_js(i,bi,bj).ne.j)then |
260 |
tmpy = tmpfldyz(j,k,bi,bj) |
261 |
cgg -- Positive is flux in. |
262 |
tmpflux = tmpflux + tmpy* delR(k)*dyg(i+ip1,j,bi,bj) |
263 |
tmparea = tmparea + delR(k)*dyg(i+ip1,j,bi,bj) |
264 |
endif |
265 |
endif |
266 |
enddo |
267 |
enddo |
268 |
sumarea =sumarea + tmparea |
269 |
sumvol = sumvol + tmpflux |
270 |
enddo |
271 |
enddo |
272 |
#endif |
273 |
|
274 |
#ifdef ALLOW_OBCSE_CONTROL |
275 |
ip1 = 0 |
276 |
|
277 |
call active_read_yz( fnameflde, tmpfldyz, |
278 |
(obcsncount0-1)*nobcs+3, doglobalread, |
279 |
ladinit, optimcycle, mythid |
280 |
& , xx_obcse_dummy ) |
281 |
|
282 |
c-- Loop over this thread tiles. |
283 |
do bj = jtlo,jthi |
284 |
do bi = itlo,ithi |
285 |
|
286 |
c-- Determine the weights to be used. |
287 |
tmpflux = 0. d 0 |
288 |
tmparea = 0. d 0 |
289 |
|
290 |
do k = 1, Nr |
291 |
do j = jmin,jmax |
292 |
i = ob_ie(j,bi,bj) |
293 |
if (maskW(i+ip1,j,k,bi,bj) .ne. 0.) then |
294 |
cgg -- Do not let the corners contribute to the volume flux. |
295 |
if (ob_jn(i,bi,bj).ne.j .and. ob_js(i,bi,bj).ne.j)then |
296 |
tmpy = tmpfldyz(j,k,bi,bj) |
297 |
cgg -- Positive is flux in. |
298 |
tmpflux = tmpflux - tmpy* delR(k)*dyg(i+ip1,j,bi,bj) |
299 |
tmparea = tmparea + delR(k) *dyg(i+ip1,j,bi,bj) |
300 |
endif |
301 |
endif |
302 |
enddo |
303 |
enddo |
304 |
sumarea = sumarea+ tmparea |
305 |
sumvol = sumvol + tmpflux |
306 |
enddo |
307 |
enddo |
308 |
#endif |
309 |
|
310 |
c-- Do the global summation. |
311 |
_GLOBAL_SUM_RL( sumvol, mythid ) |
312 |
_GLOBAL_SUM_RL( sumarea,mythid ) |
313 |
|
314 |
shiftvel(2) = sumvol /sumarea |
315 |
endif |
316 |
cgg End of the obcsnfirst loop. |
317 |
|
318 |
if ( ( obcsnfirst) .or. (obcsnchanged)) then |
319 |
|
320 |
cgg Swap the value. |
321 |
shiftvel(1) = shiftvel(2) |
322 |
|
323 |
sumvol = 0. d0 |
324 |
sumarea= 0. d0 |
325 |
|
326 |
#ifdef ALLOW_OBCSN_CONTROL |
327 |
jp1 = 0 |
328 |
|
329 |
call active_read_xz(fnamefldn,tmpfldxz, |
330 |
& (obcsncount1-1)*nobcs+4, doglobalread, |
331 |
& ladinit, optimcycle, mythid |
332 |
& , xx_obcsn_dummy ) |
333 |
c-- Loop over this thread tiles. |
334 |
do bj = jtlo,jthi |
335 |
do bi = itlo,ithi |
336 |
|
337 |
tmpflux = 0. d0 |
338 |
tmparea = 0. d0 |
339 |
|
340 |
do k = 1, Nr |
341 |
do i = imin,imax |
342 |
j = Ob_Jn(I,bi,bj) |
343 |
if (maskS(i,j+jp1,k,bi,bj) .ne. 0.) then |
344 |
cgg -- Do not let the corners contribute to the volume flux. |
345 |
if (ob_iw(j,bi,bj).ne.i .and. ob_ie(j,bi,bj).ne.i)then |
346 |
tmpx = tmpfldxz(i,k,bi,bj) |
347 |
cgg -- Positive is flux in. |
348 |
tmpflux = tmpflux - tmpx* delR(k)*dxg(i,j+jp1,bi,bj) |
349 |
tmparea = tmparea + delR(k) * dxg(i,j+jp1,bi,bj) |
350 |
endif |
351 |
endif |
352 |
enddo |
353 |
enddo |
354 |
|
355 |
sumarea = sumarea+ tmparea |
356 |
sumvol = sumvol + tmpflux |
357 |
enddo |
358 |
enddo |
359 |
#endif |
360 |
|
361 |
#ifdef ALLOW_OBCSS_CONTROL |
362 |
jp1 = 1 |
363 |
|
364 |
call active_read_xz(fnameflds,tmpfldxz, |
365 |
& (obcsncount1-1)*nobcs+4, doglobalread, |
366 |
& ladinit, optimcycle, mythid |
367 |
& , xx_obcss_dummy ) |
368 |
|
369 |
c-- Loop over this thread tiles. |
370 |
do bj = jtlo,jthi |
371 |
do bi = itlo,ithi |
372 |
|
373 |
tmpflux = 0. d 0 |
374 |
tmparea = 0. d 0 |
375 |
|
376 |
do k = 1, Nr |
377 |
do i = imin,imax |
378 |
j = Ob_Js(I,bi,bj) |
379 |
if (maskS(i,j+jp1,k,bi,bj) .ne. 0.) then |
380 |
cgg -- Do not let the corners contribute to the volume flux. |
381 |
if (ob_iw(j,bi,bj).ne.i .and. ob_ie(j,bi,bj).ne.i)then |
382 |
tmpx = tmpfldxz(i,k,bi,bj) |
383 |
cgg -- Positive is flux in. |
384 |
tmpflux = tmpflux + tmpx* delR(k)*dxg(i,j+jp1,bi,bj) |
385 |
tmparea = tmparea + delR(k) * dxg(i,j+jp1,bi,bj) |
386 |
endif |
387 |
endif |
388 |
enddo |
389 |
enddo |
390 |
sumarea = sumarea+ tmparea |
391 |
sumvol = sumvol + tmpflux |
392 |
enddo |
393 |
enddo |
394 |
#endif |
395 |
|
396 |
#ifdef ALLOW_OBCSW_CONTROL |
397 |
ip1 = 1 |
398 |
|
399 |
call active_read_yz( fnamefldw, tmpfldyz, |
400 |
& (obcsncount1-1)*nobcs+3, doglobalread, |
401 |
& ladinit, optimcycle, mythid |
402 |
& , xx_obcsw_dummy ) |
403 |
|
404 |
c-- Loop over this thread tiles. |
405 |
do bj = jtlo,jthi |
406 |
do bi = itlo,ithi |
407 |
|
408 |
c-- Determine the weights to be used. |
409 |
tmpflux = 0. d 0 |
410 |
tmparea = 0. d 0 |
411 |
|
412 |
do k = 1, Nr |
413 |
do j = jmin,jmax |
414 |
i = ob_iw(j,bi,bj) |
415 |
if (maskW(i+ip1,j,k,bi,bj) .ne. 0.) then |
416 |
cgg -- Do not let corners contribute. |
417 |
if (ob_jn(i,bi,bj).ne.j .and. ob_js(i,bi,bj).ne.j)then |
418 |
tmpy = tmpfldyz(j,k,bi,bj) |
419 |
cgg -- Positive is flux in. |
420 |
tmpflux = tmpflux + tmpy* delR(k)*dyg(i+ip1,j,bi,bj) |
421 |
tmparea = tmparea + delR(k)*dyg(i+ip1,j,bi,bj) |
422 |
endif |
423 |
endif |
424 |
enddo |
425 |
enddo |
426 |
sumarea =sumarea + tmparea |
427 |
sumvol = sumvol + tmpflux |
428 |
enddo |
429 |
enddo |
430 |
#endif |
431 |
|
432 |
#ifdef ALLOW_OBCSE_CONTROL |
433 |
ip1 = 0 |
434 |
|
435 |
call active_read_yz( fnameflde, tmpfldyz, |
436 |
(obcsncount1-1)*nobcs+3, doglobalread, |
437 |
ladinit, optimcycle, mythid |
438 |
& , xx_obcse_dummy ) |
439 |
|
440 |
c-- Loop over this thread tiles. |
441 |
do bj = jtlo,jthi |
442 |
do bi = itlo,ithi |
443 |
|
444 |
c-- Determine the weights to be used. |
445 |
tmpflux = 0. d 0 |
446 |
tmparea = 0. d 0 |
447 |
|
448 |
do k = 1, Nr |
449 |
do j = jmin,jmax |
450 |
i = ob_ie(j,bi,bj) |
451 |
if (maskW(i+ip1,j,k,bi,bj) .ne. 0.) then |
452 |
cgg -- Do not let the corners contribute to the volume flux. |
453 |
if (ob_jn(i,bi,bj).ne.j .and. ob_js(i,bi,bj).ne.j)then |
454 |
tmpy = tmpfldyz(j,k,bi,bj) |
455 |
cgg -- Positive is flux in. |
456 |
tmpflux = tmpflux - tmpy* delR(k)*dyg(i+ip1,j,bi,bj) |
457 |
tmparea = tmparea + delR(k) *dyg(i+ip1,j,bi,bj) |
458 |
endif |
459 |
endif |
460 |
enddo |
461 |
enddo |
462 |
sumarea = sumarea+ tmparea |
463 |
sumvol = sumvol + tmpflux |
464 |
enddo |
465 |
enddo |
466 |
#endif |
467 |
|
468 |
c-- Do the global summation. |
469 |
_GLOBAL_SUM_RL( sumvol, mythid ) |
470 |
_GLOBAL_SUM_RL( sumarea,mythid ) |
471 |
|
472 |
shiftvel(2) = sumvol /sumarea |
473 |
endif |
474 |
cgg End of the obcsnfirst, obcsnchanged loop. |
475 |
|
476 |
#endif /* BALANCE_CONTROL_VOLFLUX_GLOBAL */ |
477 |
|
478 |
return |
479 |
end |
480 |
|
481 |
|
482 |
|
483 |
|
484 |
|
485 |
|
486 |
|