Home

Context Navigation

source: palm/trunk/SOURCE/data_output_2d.f90 @ 1962

Last change on this file since 1962 was 1961, checked in by suehring, 8 years ago
last commit documented
Property svn:keywords set to `Id`
File size: 88.7 KB

Line
1	!> @file data_output_2d.f90
2	!--------------------------------------------------------------------------------!
3	! This file is part of PALM.
4	!
5	! PALM is free software: you can redistribute it and/or modify it under the terms
6	! of the GNU General Public License as published by the Free Software Foundation,
7	! either version 3 of the License, or (at your option) any later version.
8	!
9	! PALM is distributed in the hope that it will be useful, but WITHOUT ANY
10	! WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
11	! A PARTICULAR PURPOSE. See the GNU General Public License for more details.
12	!
13	! You should have received a copy of the GNU General Public License along with
14	! PALM. If not, see <http://www.gnu.org/licenses/>.
15	!
16	! Copyright 1997-2016 Leibniz Universitaet Hannover
17	!--------------------------------------------------------------------------------!
18	!
19	! Current revisions:
20	! -----------------
21	!
22	!
23	! Former revisions:
24	! -----------------
25	! $Id: data_output_2d.f90 1961 2016-07-12 16:37:58Z suehring $
26	!
27	! 1960 2016-07-12 16:34:24Z suehring
28	! Scalar surface flux added
29	! Rename INTEGER variable s into s_ind, as s is already assigned to scalar
30	!
31	! 1849 2016-04-08 11:33:18Z hoffmann
32	! precipitation_amount, precipitation_rate, prr moved to arrays_3d
33	!
34	! 1822 2016-04-07 07:49:42Z hoffmann
35	! Output of bulk cloud physics simplified.
36	!
37	! 1788 2016-03-10 11:01:04Z maronga
38	! Added output of z0q
39	!
40	! 1783 2016-03-06 18:36:17Z raasch
41	! name change of netcdf routines and module + related changes
42	!
43	! 1745 2016-02-05 13:06:51Z gronemeier
44	! Bugfix: test if time axis limit exceeds moved to point after call of check_open
45	!
46	! 1703 2015-11-02 12:38:44Z raasch
47	! bugfix for output of single (*) xy-sections in case of parallel netcdf I/O
48	!
49	! 1701 2015-11-02 07:43:04Z maronga
50	! Bugfix in output of RRTGM data
51	!
52	! 1691 2015-10-26 16:17:44Z maronga
53	! Added output of Obukhov length (ol) and radiative heating rates for RRTMG.
54	! Formatting corrections.
55	!
56	! 1682 2015-10-07 23:56:08Z knoop
57	! Code annotations made doxygen readable
58	!
59	! 1585 2015-04-30 07:05:52Z maronga
60	! Added support for RRTMG
61	!
62	! 1555 2015-03-04 17:44:27Z maronga
63	! Added output of r_a and r_s
64	!
65	! 1551 2015-03-03 14:18:16Z maronga
66	! Added suppport for land surface model and radiation model output. In the course
67	! of this action, the limits for vertical loops have been changed (from nzb and
68	! nzt+1 to nzb_do and nzt_do, respectively in order to allow soil model output).
69	! Moreover, a new vertical grid zs was introduced.
70	!
71	! 1359 2014-04-11 17:15:14Z hoffmann
72	! New particle structure integrated.
73	!
74	! 1353 2014-04-08 15:21:23Z heinze
75	! REAL constants provided with KIND-attribute
76	!
77	! 1327 2014-03-21 11:00:16Z raasch
78	! parts concerning iso2d output removed,
79	! -netcdf output queries
80	!
81	! 1320 2014-03-20 08:40:49Z raasch
82	! ONLY-attribute added to USE-statements,
83	! kind-parameters added to all INTEGER and REAL declaration statements,
84	! kinds are defined in new module kinds,
85	! revision history before 2012 removed,
86	! comment fields (!:) to be used for variable explanations added to
87	! all variable declaration statements
88	!
89	! 1318 2014-03-17 13:35:16Z raasch
90	! barrier argument removed from cpu_log.
91	! module interfaces removed
92	!
93	! 1311 2014-03-14 12:13:39Z heinze
94	! bugfix: close #if defined( __netcdf )
95	!
96	! 1308 2014-03-13 14:58:42Z fricke
97	! +local_2d_sections, local_2d_sections_l, ns
98	! Check, if the limit of the time dimension is exceeded for parallel output
99	! To increase the performance for parallel output, the following is done:
100	! - Update of time axis is only done by PE0
101	! - Cross sections are first stored on a local array and are written
102	! collectively to the output file by all PEs.
103	!
104	! 1115 2013-03-26 18:16:16Z hoffmann
105	! ql is calculated by calc_liquid_water_content
106	!
107	! 1076 2012-12-05 08:30:18Z hoffmann
108	! Bugfix in output of ql
109	!
110	! 1065 2012-11-22 17:42:36Z hoffmann
111	! Bugfix: Output of cross sections of ql
112	!
113	! 1053 2012-11-13 17:11:03Z hoffmann
114	! +qr, nr, qc and cross sections
115	!
116	! 1036 2012-10-22 13:43:42Z raasch
117	! code put under GPL (PALM 3.9)
118	!
119	! 1031 2012-10-19 14:35:30Z raasch
120	! netCDF4 without parallel file support implemented
121	!
122	! 1007 2012-09-19 14:30:36Z franke
123	! Bugfix: missing calculation of ql_vp added
124	!
125	! 978 2012-08-09 08:28:32Z fricke
126	! +z0h
127	!
128	! Revision 1.1 1997/08/11 06:24:09 raasch
129	! Initial revision
130	!
131	!
132	! Description:
133	! ------------
134	!> Data output of horizontal cross-sections in netCDF format or binary format
135	!> compatible to old graphic software iso2d.
136	!> Attention: The position of the sectional planes is still not always computed
137	!> --------- correctly. (zu is used always)!
138	!------------------------------------------------------------------------------!
139	SUBROUTINE data_output_2d( mode, av )
140
141
142	USE arrays_3d, &
143	ONLY: dzw, e, nr, ol, p, pt, precipitation_amount, precipitation_rate,&
144	prr,q, qc, ql, ql_c, ql_v, ql_vp, qr, qsws, rho, s, sa, shf, &
145	ssws, tend, ts, u, us, v, vpt, w, z0, z0h, z0q, zu, zw
146
147	USE averaging
148
149	USE cloud_parameters, &
150	ONLY: hyrho, l_d_cp, pt_d_t
151
152	USE control_parameters, &
153	ONLY: cloud_physics, data_output_2d_on_each_pe, data_output_xy, &
154	data_output_xz, data_output_yz, do2d, &
155	do2d_xy_last_time, do2d_xy_n, do2d_xy_time_count, &
156	do2d_xz_last_time, do2d_xz_n, do2d_xz_time_count, &
157	do2d_yz_last_time, do2d_yz_n, do2d_yz_time_count, &
158	ibc_uv_b, io_blocks, io_group, message_string, &
159	ntdim_2d_xy, ntdim_2d_xz, ntdim_2d_yz, &
160	psolver, section, simulated_time, simulated_time_chr, &
161	time_since_reference_point
162
163	USE cpulog, &
164	ONLY: cpu_log, log_point
165
166	USE grid_variables, &
167	ONLY: dx, dy
168
169	USE indices, &
170	ONLY: nbgp, nx, nxl, nxlg, nxr, nxrg, ny, nyn, nyng, nys, nysg, &
171	nz, nzb, nzt
172
173	USE kinds
174
175	USE land_surface_model_mod, &
176	ONLY: c_liq, c_liq_av, c_soil_av, c_veg, c_veg_av, ghf_eb, &
177	ghf_eb_av, lai, lai_av, m_liq_eb, m_liq_eb_av, m_soil, &
178	m_soil_av, nzb_soil, nzt_soil, qsws_eb, qsws_eb_av, &
179	qsws_liq_eb, qsws_liq_eb_av, qsws_soil_eb, qsws_soil_eb_av, &
180	qsws_veg_eb, qsws_veg_eb_av, r_a, r_a_av, r_s, r_s_av, shf_eb, &
181	shf_eb_av, t_soil, t_soil_av, zs
182
183	#if defined( __netcdf )
184	USE NETCDF
185	#endif
186
187	USE netcdf_interface, &
188	ONLY: id_set_xy, id_set_xz, id_set_yz, id_var_do2d, id_var_time_xy, &
189	id_var_time_xz, id_var_time_yz, nc_stat, netcdf_data_format, &
190	netcdf_handle_error
191
192	USE particle_attributes, &
193	ONLY: grid_particles, number_of_particles, particle_advection_start, &
194	particles, prt_count
195
196	USE pegrid
197
198	USE radiation_model_mod, &
199	ONLY: rad_net, rad_net_av, rad_sw_in, rad_sw_in_av, rad_sw_out, &
200	rad_sw_out_av, rad_sw_cs_hr, rad_sw_cs_hr_av, rad_sw_hr, &
201	rad_sw_hr_av, rad_lw_in, rad_lw_in_av, rad_lw_out, &
202	rad_lw_out_av, rad_lw_cs_hr, rad_lw_cs_hr_av, rad_lw_hr, &
203	rad_lw_hr_av
204
205	IMPLICIT NONE
206
207	CHARACTER (LEN=2) :: do2d_mode !<
208	CHARACTER (LEN=2) :: mode !<
209	CHARACTER (LEN=4) :: grid !<
210	CHARACTER (LEN=25) :: section_chr !<
211	CHARACTER (LEN=50) :: rtext !<
212
213	INTEGER(iwp) :: av !<
214	INTEGER(iwp) :: ngp !<
215	INTEGER(iwp) :: file_id !<
216	INTEGER(iwp) :: i !<
217	INTEGER(iwp) :: if !<
218	INTEGER(iwp) :: is !<
219	INTEGER(iwp) :: iis !<
220	INTEGER(iwp) :: j !<
221	INTEGER(iwp) :: k !<
222	INTEGER(iwp) :: l !<
223	INTEGER(iwp) :: layer_xy !<
224	INTEGER(iwp) :: n !<
225	INTEGER(iwp) :: nis !<
226	INTEGER(iwp) :: ns !<
227	INTEGER(iwp) :: nzb_do !< lower limit of the data field (usually nzb)
228	INTEGER(iwp) :: nzt_do !< upper limit of the data field (usually nzt+1)
229	INTEGER(iwp) :: psi !<
230	INTEGER(iwp) :: s_ind !<
231	INTEGER(iwp) :: sender !<
232	INTEGER(iwp) :: ind(4) !<
233
234	LOGICAL :: found !<
235	LOGICAL :: resorted !<
236	LOGICAL :: two_d !<
237
238	REAL(wp) :: mean_r !<
239	REAL(wp) :: s_r2 !<
240	REAL(wp) :: s_r3 !<
241
242	REAL(wp), DIMENSION(:), ALLOCATABLE :: level_z !<
243	REAL(wp), DIMENSION(:,:), ALLOCATABLE :: local_2d !<
244	REAL(wp), DIMENSION(:,:), ALLOCATABLE :: local_2d_l !<
245	REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: local_pf !<
246	REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: local_2d_sections !<
247	REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: local_2d_sections_l !<
248
249	#if defined( __parallel )
250	REAL(wp), DIMENSION(:,:), ALLOCATABLE :: total_2d !<
251	#endif
252	REAL(wp), DIMENSION(:,:,:), POINTER :: to_be_resorted !<
253
254	NAMELIST /LOCAL/ rtext
255
256	!
257	!-- Immediate return, if no output is requested (no respective sections
258	!-- found in parameter data_output)
259	IF ( mode == 'xy' .AND. .NOT. data_output_xy(av) ) RETURN
260	IF ( mode == 'xz' .AND. .NOT. data_output_xz(av) ) RETURN
261	IF ( mode == 'yz' .AND. .NOT. data_output_yz(av) ) RETURN
262
263	CALL cpu_log (log_point(3),'data_output_2d','start')
264
265	two_d = .FALSE. ! local variable to distinguish between output of pure 2D
266	! arrays and cross-sections of 3D arrays.
267
268	!
269	!-- Depending on the orientation of the cross-section, the respective output
270	!-- files have to be opened.
271	SELECT CASE ( mode )
272
273	CASE ( 'xy' )
274	s_ind = 1
275	ALLOCATE( level_z(nzb:nzt+1), local_2d(nxlg:nxrg,nysg:nyng) )
276
277	IF ( netcdf_data_format > 4 ) THEN
278	ns = 1
279	DO WHILE ( section(ns,s_ind) /= -9999 .AND. ns <= 100 )
280	ns = ns + 1
281	ENDDO
282	ns = ns - 1
283	ALLOCATE( local_2d_sections(nxlg:nxrg,nysg:nyng,1:ns) )
284	local_2d_sections = 0.0_wp
285	ENDIF
286
287	!
288	!-- Parallel netCDF4/HDF5 output is done on all PEs, all other on PE0 only
289	IF ( myid == 0 .OR. netcdf_data_format > 4 ) THEN
290	CALL check_open( 101+av*10 )
291	ENDIF
292
293	IF ( data_output_2d_on_each_pe ) THEN
294	CALL check_open( 21 )
295	ELSE
296	IF ( myid == 0 ) THEN
297	#if defined( __parallel )
298	ALLOCATE( total_2d(-nbgp:nx+nbgp,-nbgp:ny+nbgp) )
299	#endif
300	ENDIF
301	ENDIF
302
303	CASE ( 'xz' )
304	s_ind = 2
305	ALLOCATE( local_2d(nxlg:nxrg,nzb:nzt+1) )
306
307	IF ( netcdf_data_format > 4 ) THEN
308	ns = 1
309	DO WHILE ( section(ns,s_ind) /= -9999 .AND. ns <= 100 )
310	ns = ns + 1
311	ENDDO
312	ns = ns - 1
313	ALLOCATE( local_2d_sections(nxlg:nxrg,1:ns,nzb:nzt+1) )
314	ALLOCATE( local_2d_sections_l(nxlg:nxrg,1:ns,nzb:nzt+1) )
315	local_2d_sections = 0.0_wp; local_2d_sections_l = 0.0_wp
316	ENDIF
317
318	!
319	!-- Parallel netCDF4/HDF5 output is done on all PEs, all other on PE0 only
320	IF ( myid == 0 .OR. netcdf_data_format > 4 ) THEN
321	CALL check_open( 102+av*10 )
322	ENDIF
323
324	IF ( data_output_2d_on_each_pe ) THEN
325	CALL check_open( 22 )
326	ELSE
327	IF ( myid == 0 ) THEN
328	#if defined( __parallel )
329	ALLOCATE( total_2d(-nbgp:nx+nbgp,nzb:nzt+1) )
330	#endif
331	ENDIF
332	ENDIF
333
334	CASE ( 'yz' )
335	s_ind = 3
336	ALLOCATE( local_2d(nysg:nyng,nzb:nzt+1) )
337
338	IF ( netcdf_data_format > 4 ) THEN
339	ns = 1
340	DO WHILE ( section(ns,s_ind) /= -9999 .AND. ns <= 100 )
341	ns = ns + 1
342	ENDDO
343	ns = ns - 1
344	ALLOCATE( local_2d_sections(1:ns,nysg:nyng,nzb:nzt+1) )
345	ALLOCATE( local_2d_sections_l(1:ns,nysg:nyng,nzb:nzt+1) )
346	local_2d_sections = 0.0_wp; local_2d_sections_l = 0.0_wp
347	ENDIF
348
349	!
350	!-- Parallel netCDF4/HDF5 output is done on all PEs, all other on PE0 only
351	IF ( myid == 0 .OR. netcdf_data_format > 4 ) THEN
352	CALL check_open( 103+av*10 )
353	ENDIF
354
355	IF ( data_output_2d_on_each_pe ) THEN
356	CALL check_open( 23 )
357	ELSE
358	IF ( myid == 0 ) THEN
359	#if defined( __parallel )
360	ALLOCATE( total_2d(-nbgp:ny+nbgp,nzb:nzt+1) )
361	#endif
362	ENDIF
363	ENDIF
364
365	CASE DEFAULT
366	message_string = 'unknown cross-section: ' // TRIM( mode )
367	CALL message( 'data_output_2d', 'PA0180', 1, 2, 0, 6, 0 )
368
369	END SELECT
370
371	!
372	!-- For parallel netcdf output the time axis must be limited. Return, if this
373	!-- limit is exceeded. This could be the case, if the simulated time exceeds
374	!-- the given end time by the length of the given output interval.
375	IF ( netcdf_data_format > 4 ) THEN
376	IF ( mode == 'xy' .AND. do2d_xy_time_count(av) + 1 > &
377	ntdim_2d_xy(av) ) THEN
378	WRITE ( message_string, * ) 'Output of xy cross-sections is not ', &
379	'given at t=', simulated_time, '&because the', &
380	' maximum number of output time levels is exceeded.'
381	CALL message( 'data_output_2d', 'PA0384', 0, 1, 0, 6, 0 )
382	CALL cpu_log( log_point(3), 'data_output_2d', 'stop' )
383	RETURN
384	ENDIF
385	IF ( mode == 'xz' .AND. do2d_xz_time_count(av) + 1 > &
386	ntdim_2d_xz(av) ) THEN
387	WRITE ( message_string, * ) 'Output of xz cross-sections is not ', &
388	'given at t=', simulated_time, '&because the', &
389	' maximum number of output time levels is exceeded.'
390	CALL message( 'data_output_2d', 'PA0385', 0, 1, 0, 6, 0 )
391	CALL cpu_log( log_point(3), 'data_output_2d', 'stop' )
392	RETURN
393	ENDIF
394	IF ( mode == 'yz' .AND. do2d_yz_time_count(av) + 1 > &
395	ntdim_2d_yz(av) ) THEN
396	WRITE ( message_string, * ) 'Output of yz cross-sections is not ', &
397	'given at t=', simulated_time, '&because the', &
398	' maximum number of output time levels is exceeded.'
399	CALL message( 'data_output_2d', 'PA0386', 0, 1, 0, 6, 0 )
400	CALL cpu_log( log_point(3), 'data_output_2d', 'stop' )
401	RETURN
402	ENDIF
403	ENDIF
404
405	!
406	!-- Allocate a temporary array for resorting (kji -> ijk).
407	ALLOCATE( local_pf(nxlg:nxrg,nysg:nyng,nzb:nzt+1) )
408
409	!
410	!-- Loop of all variables to be written.
411	!-- Output dimensions chosen
412	if = 1
413	l = MAX( 2, LEN_TRIM( do2d(av,if) ) )
414	do2d_mode = do2d(av,if)(l-1:l)
415
416	DO WHILE ( do2d(av,if)(1:1) /= ' ' )
417
418	IF ( do2d_mode == mode ) THEN
419
420	nzb_do = nzb
421	nzt_do = nzt+1
422	!
423	!-- Store the array chosen on the temporary array.
424	resorted = .FALSE.
425	SELECT CASE ( TRIM( do2d(av,if) ) )
426
427	CASE ( 'e_xy', 'e_xz', 'e_yz' )
428	IF ( av == 0 ) THEN
429	to_be_resorted => e
430	ELSE
431	to_be_resorted => e_av
432	ENDIF
433	IF ( mode == 'xy' ) level_z = zu
434
435	CASE ( 'c_liq*_xy' ) ! 2d-array
436	IF ( av == 0 ) THEN
437	DO i = nxlg, nxrg
438	DO j = nysg, nyng
439	local_pf(i,j,nzb+1) = c_liq(j,i) * c_veg(j,i)
440	ENDDO
441	ENDDO
442	ELSE
443	DO i = nxlg, nxrg
444	DO j = nysg, nyng
445	local_pf(i,j,nzb+1) = c_liq_av(j,i)
446	ENDDO
447	ENDDO
448	ENDIF
449	resorted = .TRUE.
450	two_d = .TRUE.
451	level_z(nzb+1) = zu(nzb+1)
452
453	CASE ( 'c_soil*_xy' ) ! 2d-array
454	IF ( av == 0 ) THEN
455	DO i = nxlg, nxrg
456	DO j = nysg, nyng
457	local_pf(i,j,nzb+1) = 1.0_wp - c_veg(j,i)
458	ENDDO
459	ENDDO
460	ELSE
461	DO i = nxlg, nxrg
462	DO j = nysg, nyng
463	local_pf(i,j,nzb+1) = c_soil_av(j,i)
464	ENDDO
465	ENDDO
466	ENDIF
467	resorted = .TRUE.
468	two_d = .TRUE.
469	level_z(nzb+1) = zu(nzb+1)
470
471	CASE ( 'c_veg*_xy' ) ! 2d-array
472	IF ( av == 0 ) THEN
473	DO i = nxlg, nxrg
474	DO j = nysg, nyng
475	local_pf(i,j,nzb+1) = c_veg(j,i)
476	ENDDO
477	ENDDO
478	ELSE
479	DO i = nxlg, nxrg
480	DO j = nysg, nyng
481	local_pf(i,j,nzb+1) = c_veg_av(j,i)
482	ENDDO
483	ENDDO
484	ENDIF
485	resorted = .TRUE.
486	two_d = .TRUE.
487	level_z(nzb+1) = zu(nzb+1)
488
489	CASE ( 'ghf_eb*_xy' ) ! 2d-array
490	IF ( av == 0 ) THEN
491	DO i = nxlg, nxrg
492	DO j = nysg, nyng
493	local_pf(i,j,nzb+1) = ghf_eb(j,i)
494	ENDDO
495	ENDDO
496	ELSE
497	DO i = nxlg, nxrg
498	DO j = nysg, nyng
499	local_pf(i,j,nzb+1) = ghf_eb_av(j,i)
500	ENDDO
501	ENDDO
502	ENDIF
503	resorted = .TRUE.
504	two_d = .TRUE.
505	level_z(nzb+1) = zu(nzb+1)
506
507	CASE ( 'lai*_xy' ) ! 2d-array
508	IF ( av == 0 ) THEN
509	DO i = nxlg, nxrg
510	DO j = nysg, nyng
511	local_pf(i,j,nzb+1) = lai(j,i)
512	ENDDO
513	ENDDO
514	ELSE
515	DO i = nxlg, nxrg
516	DO j = nysg, nyng
517	local_pf(i,j,nzb+1) = lai_av(j,i)
518	ENDDO
519	ENDDO
520	ENDIF
521	resorted = .TRUE.
522	two_d = .TRUE.
523	level_z(nzb+1) = zu(nzb+1)
524
525	CASE ( 'lpt_xy', 'lpt_xz', 'lpt_yz' )
526	IF ( av == 0 ) THEN
527	to_be_resorted => pt
528	ELSE
529	to_be_resorted => lpt_av
530	ENDIF
531	IF ( mode == 'xy' ) level_z = zu
532
533	CASE ( 'lwp*_xy' ) ! 2d-array
534	IF ( av == 0 ) THEN
535	DO i = nxlg, nxrg
536	DO j = nysg, nyng
537	local_pf(i,j,nzb+1) = SUM( ql(nzb:nzt,j,i) * &
538	dzw(1:nzt+1) )
539	ENDDO
540	ENDDO
541	ELSE
542	DO i = nxlg, nxrg
543	DO j = nysg, nyng
544	local_pf(i,j,nzb+1) = lwp_av(j,i)
545	ENDDO
546	ENDDO
547	ENDIF
548	resorted = .TRUE.
549	two_d = .TRUE.
550	level_z(nzb+1) = zu(nzb+1)
551
552	CASE ( 'm_liq_eb*_xy' ) ! 2d-array
553	IF ( av == 0 ) THEN
554	DO i = nxlg, nxrg
555	DO j = nysg, nyng
556	local_pf(i,j,nzb+1) = m_liq_eb(j,i)
557	ENDDO
558	ENDDO
559	ELSE
560	DO i = nxlg, nxrg
561	DO j = nysg, nyng
562	local_pf(i,j,nzb+1) = m_liq_eb_av(j,i)
563	ENDDO
564	ENDDO
565	ENDIF
566	resorted = .TRUE.
567	two_d = .TRUE.
568	level_z(nzb+1) = zu(nzb+1)
569
570	CASE ( 'm_soil_xy', 'm_soil_xz', 'm_soil_yz' )
571	nzb_do = nzb_soil
572	nzt_do = nzt_soil
573	IF ( av == 0 ) THEN
574	to_be_resorted => m_soil
575	ELSE
576	to_be_resorted => m_soil_av
577	ENDIF
578	IF ( mode == 'xy' ) level_z = zs
579
580	CASE ( 'nr_xy', 'nr_xz', 'nr_yz' )
581	IF ( av == 0 ) THEN
582	to_be_resorted => nr
583	ELSE
584	to_be_resorted => nr_av
585	ENDIF
586	IF ( mode == 'xy' ) level_z = zu
587
588	CASE ( 'ol*_xy' ) ! 2d-array
589	IF ( av == 0 ) THEN
590	DO i = nxlg, nxrg
591	DO j = nysg, nyng
592	local_pf(i,j,nzb+1) = ol(j,i)
593	ENDDO
594	ENDDO
595	ELSE
596	DO i = nxlg, nxrg
597	DO j = nysg, nyng
598	local_pf(i,j,nzb+1) = ol_av(j,i)
599	ENDDO
600	ENDDO
601	ENDIF
602	resorted = .TRUE.
603	two_d = .TRUE.
604	level_z(nzb+1) = zu(nzb+1)
605
606	CASE ( 'p_xy', 'p_xz', 'p_yz' )
607	IF ( av == 0 ) THEN
608	IF ( psolver /= 'sor' ) CALL exchange_horiz( p, nbgp )
609	to_be_resorted => p
610	ELSE
611	IF ( psolver /= 'sor' ) CALL exchange_horiz( p_av, nbgp )
612	to_be_resorted => p_av
613	ENDIF
614	IF ( mode == 'xy' ) level_z = zu
615
616	CASE ( 'pc_xy', 'pc_xz', 'pc_yz' ) ! particle concentration
617	IF ( av == 0 ) THEN
618	IF ( simulated_time >= particle_advection_start ) THEN
619	tend = prt_count
620	CALL exchange_horiz( tend, nbgp )
621	ELSE
622	tend = 0.0_wp
623	ENDIF
624	DO i = nxlg, nxrg
625	DO j = nysg, nyng
626	DO k = nzb, nzt+1
627	local_pf(i,j,k) = tend(k,j,i)
628	ENDDO
629	ENDDO
630	ENDDO
631	resorted = .TRUE.
632	ELSE
633	CALL exchange_horiz( pc_av, nbgp )
634	to_be_resorted => pc_av
635	ENDIF
636
637	CASE ( 'pr_xy', 'pr_xz', 'pr_yz' ) ! mean particle radius (effective radius)
638	IF ( av == 0 ) THEN
639	IF ( simulated_time >= particle_advection_start ) THEN
640	DO i = nxl, nxr
641	DO j = nys, nyn
642	DO k = nzb, nzt+1
643	number_of_particles = prt_count(k,j,i)
644	IF (number_of_particles <= 0) CYCLE
645	particles => grid_particles(k,j,i)%particles(1:number_of_particles)
646	s_r2 = 0.0_wp
647	s_r3 = 0.0_wp
648	DO n = 1, number_of_particles
649	IF ( particles(n)%particle_mask ) THEN
650	s_r2 = s_r2 + particles(n)%radius*2 &
651	particles(n)%weight_factor
652	s_r3 = s_r3 + particles(n)%radius*3 &
653	particles(n)%weight_factor
654	ENDIF
655	ENDDO
656	IF ( s_r2 > 0.0_wp ) THEN
657	mean_r = s_r3 / s_r2
658	ELSE
659	mean_r = 0.0_wp
660	ENDIF
661	tend(k,j,i) = mean_r
662	ENDDO
663	ENDDO
664	ENDDO
665	CALL exchange_horiz( tend, nbgp )
666	ELSE
667	tend = 0.0_wp
668	ENDIF
669	DO i = nxlg, nxrg
670	DO j = nysg, nyng
671	DO k = nzb, nzt+1
672	local_pf(i,j,k) = tend(k,j,i)
673	ENDDO
674	ENDDO
675	ENDDO
676	resorted = .TRUE.
677	ELSE
678	CALL exchange_horiz( pr_av, nbgp )
679	to_be_resorted => pr_av
680	ENDIF
681
682	CASE ( 'pra*_xy' ) ! 2d-array / integral quantity => no av
683	CALL exchange_horiz_2d( precipitation_amount )
684	DO i = nxlg, nxrg
685	DO j = nysg, nyng
686	local_pf(i,j,nzb+1) = precipitation_amount(j,i)
687	ENDDO
688	ENDDO
689	precipitation_amount = 0.0_wp ! reset for next integ. interval
690	resorted = .TRUE.
691	two_d = .TRUE.
692	level_z(nzb+1) = zu(nzb+1)
693
694	CASE ( 'prr*_xy' ) ! 2d-array
695	IF ( av == 0 ) THEN
696	CALL exchange_horiz_2d( prr(nzb+1,:,:) )
697	DO i = nxlg, nxrg
698	DO j = nysg, nyng
699	local_pf(i,j,nzb+1) = prr(nzb+1,j,i) * hyrho(nzb+1)
700	ENDDO
701	ENDDO
702	ELSE
703	CALL exchange_horiz_2d( prr_av(nzb+1,:,:) )
704	DO i = nxlg, nxrg
705	DO j = nysg, nyng
706	local_pf(i,j,nzb+1) = prr_av(nzb+1,j,i) * hyrho(nzb+1)
707	ENDDO
708	ENDDO
709	ENDIF
710	resorted = .TRUE.
711	two_d = .TRUE.
712	level_z(nzb+1) = zu(nzb+1)
713
714	CASE ( 'prr_xy', 'prr_xz', 'prr_yz' )
715	IF ( av == 0 ) THEN
716	CALL exchange_horiz( prr, nbgp )
717	DO i = nxlg, nxrg
718	DO j = nysg, nyng
719	DO k = nzb, nzt+1
720	local_pf(i,j,k) = prr(k,j,i) * hyrho(nzb+1)
721	ENDDO
722	ENDDO
723	ENDDO
724	ELSE
725	CALL exchange_horiz( prr_av, nbgp )
726	DO i = nxlg, nxrg
727	DO j = nysg, nyng
728	DO k = nzb, nzt+1
729	local_pf(i,j,k) = prr_av(k,j,i) * hyrho(nzb+1)
730	ENDDO
731	ENDDO
732	ENDDO
733	ENDIF
734	resorted = .TRUE.
735	IF ( mode == 'xy' ) level_z = zu
736
737	CASE ( 'pt_xy', 'pt_xz', 'pt_yz' )
738	IF ( av == 0 ) THEN
739	IF ( .NOT. cloud_physics ) THEN
740	to_be_resorted => pt
741	ELSE
742	DO i = nxlg, nxrg
743	DO j = nysg, nyng
744	DO k = nzb, nzt+1
745	local_pf(i,j,k) = pt(k,j,i) + l_d_cp * &
746	pt_d_t(k) * &
747	ql(k,j,i)
748	ENDDO
749	ENDDO
750	ENDDO
751	resorted = .TRUE.
752	ENDIF
753	ELSE
754	to_be_resorted => pt_av
755	ENDIF
756	IF ( mode == 'xy' ) level_z = zu
757
758	CASE ( 'q_xy', 'q_xz', 'q_yz' )
759	IF ( av == 0 ) THEN
760	to_be_resorted => q
761	ELSE
762	to_be_resorted => q_av
763	ENDIF
764	IF ( mode == 'xy' ) level_z = zu
765
766	CASE ( 'qc_xy', 'qc_xz', 'qc_yz' )
767	IF ( av == 0 ) THEN
768	to_be_resorted => qc
769	ELSE
770	to_be_resorted => qc_av
771	ENDIF
772	IF ( mode == 'xy' ) level_z = zu
773
774	CASE ( 'ql_xy', 'ql_xz', 'ql_yz' )
775	IF ( av == 0 ) THEN
776	to_be_resorted => ql
777	ELSE
778	to_be_resorted => ql_av
779	ENDIF
780	IF ( mode == 'xy' ) level_z = zu
781
782	CASE ( 'ql_c_xy', 'ql_c_xz', 'ql_c_yz' )
783	IF ( av == 0 ) THEN
784	to_be_resorted => ql_c
785	ELSE
786	to_be_resorted => ql_c_av
787	ENDIF
788	IF ( mode == 'xy' ) level_z = zu
789
790	CASE ( 'ql_v_xy', 'ql_v_xz', 'ql_v_yz' )
791	IF ( av == 0 ) THEN
792	to_be_resorted => ql_v
793	ELSE
794	to_be_resorted => ql_v_av
795	ENDIF
796	IF ( mode == 'xy' ) level_z = zu
797
798	CASE ( 'ql_vp_xy', 'ql_vp_xz', 'ql_vp_yz' )
799	IF ( av == 0 ) THEN
800	IF ( simulated_time >= particle_advection_start ) THEN
801	DO i = nxl, nxr
802	DO j = nys, nyn
803	DO k = nzb, nzt+1
804	number_of_particles = prt_count(k,j,i)
805	IF (number_of_particles <= 0) CYCLE
806	particles => grid_particles(k,j,i)%particles(1:number_of_particles)
807	DO n = 1, number_of_particles
808	IF ( particles(n)%particle_mask ) THEN
809	tend(k,j,i) = tend(k,j,i) + &
810	particles(n)%weight_factor / &
811	prt_count(k,j,i)
812	ENDIF
813	ENDDO
814	ENDDO
815	ENDDO
816	ENDDO
817	CALL exchange_horiz( tend, nbgp )
818	ELSE
819	tend = 0.0_wp
820	ENDIF
821	DO i = nxlg, nxrg
822	DO j = nysg, nyng
823	DO k = nzb, nzt+1
824	local_pf(i,j,k) = tend(k,j,i)
825	ENDDO
826	ENDDO
827	ENDDO
828	resorted = .TRUE.
829	ELSE
830	CALL exchange_horiz( ql_vp_av, nbgp )
831	to_be_resorted => ql_vp
832	ENDIF
833	IF ( mode == 'xy' ) level_z = zu
834
835	CASE ( 'qr_xy', 'qr_xz', 'qr_yz' )
836	IF ( av == 0 ) THEN
837	to_be_resorted => qr
838	ELSE
839	to_be_resorted => qr_av
840	ENDIF
841	IF ( mode == 'xy' ) level_z = zu
842
843	CASE ( 'qsws*_xy' ) ! 2d-array
844	IF ( av == 0 ) THEN
845	DO i = nxlg, nxrg
846	DO j = nysg, nyng
847	local_pf(i,j,nzb+1) = qsws(j,i)
848	ENDDO
849	ENDDO
850	ELSE
851	DO i = nxlg, nxrg
852	DO j = nysg, nyng
853	local_pf(i,j,nzb+1) = qsws_av(j,i)
854	ENDDO
855	ENDDO
856	ENDIF
857	resorted = .TRUE.
858	two_d = .TRUE.
859	level_z(nzb+1) = zu(nzb+1)
860
861	CASE ( 'qsws_eb*_xy' ) ! 2d-array
862	IF ( av == 0 ) THEN
863	DO i = nxlg, nxrg
864	DO j = nysg, nyng
865	local_pf(i,j,nzb+1) = qsws_eb(j,i)
866	ENDDO
867	ENDDO
868	ELSE
869	DO i = nxlg, nxrg
870	DO j = nysg, nyng
871	local_pf(i,j,nzb+1) = qsws_eb_av(j,i)
872	ENDDO
873	ENDDO
874	ENDIF
875	resorted = .TRUE.
876	two_d = .TRUE.
877	level_z(nzb+1) = zu(nzb+1)
878
879	CASE ( 'qsws_liq_eb*_xy' ) ! 2d-array
880	IF ( av == 0 ) THEN
881	DO i = nxlg, nxrg
882	DO j = nysg, nyng
883	local_pf(i,j,nzb+1) = qsws_liq_eb(j,i)
884	ENDDO
885	ENDDO
886	ELSE
887	DO i = nxlg, nxrg
888	DO j = nysg, nyng
889	local_pf(i,j,nzb+1) = qsws_liq_eb_av(j,i)
890	ENDDO
891	ENDDO
892	ENDIF
893	resorted = .TRUE.
894	two_d = .TRUE.
895	level_z(nzb+1) = zu(nzb+1)
896
897	CASE ( 'qsws_soil_eb*_xy' ) ! 2d-array
898	IF ( av == 0 ) THEN
899	DO i = nxlg, nxrg
900	DO j = nysg, nyng
901	local_pf(i,j,nzb+1) = qsws_soil_eb(j,i)
902	ENDDO
903	ENDDO
904	ELSE
905	DO i = nxlg, nxrg
906	DO j = nysg, nyng
907	local_pf(i,j,nzb+1) = qsws_soil_eb_av(j,i)
908	ENDDO
909	ENDDO
910	ENDIF
911	resorted = .TRUE.
912	two_d = .TRUE.
913	level_z(nzb+1) = zu(nzb+1)
914
915	CASE ( 'qsws_veg_eb*_xy' ) ! 2d-array
916	IF ( av == 0 ) THEN
917	DO i = nxlg, nxrg
918	DO j = nysg, nyng
919	local_pf(i,j,nzb+1) = qsws_veg_eb(j,i)
920	ENDDO
921	ENDDO
922	ELSE
923	DO i = nxlg, nxrg
924	DO j = nysg, nyng
925	local_pf(i,j,nzb+1) = qsws_veg_eb_av(j,i)
926	ENDDO
927	ENDDO
928	ENDIF
929	resorted = .TRUE.
930	two_d = .TRUE.
931	level_z(nzb+1) = zu(nzb+1)
932
933	CASE ( 'qv_xy', 'qv_xz', 'qv_yz' )
934	IF ( av == 0 ) THEN
935	DO i = nxlg, nxrg
936	DO j = nysg, nyng
937	DO k = nzb, nzt+1
938	local_pf(i,j,k) = q(k,j,i) - ql(k,j,i)
939	ENDDO
940	ENDDO
941	ENDDO
942	resorted = .TRUE.
943	ELSE
944	to_be_resorted => qv_av
945	ENDIF
946	IF ( mode == 'xy' ) level_z = zu
947
948	CASE ( 'rad_net*_xy' ) ! 2d-array
949	IF ( av == 0 ) THEN
950	DO i = nxlg, nxrg
951	DO j = nysg, nyng
952	local_pf(i,j,nzb+1) = rad_net(j,i)
953	ENDDO
954	ENDDO
955	ELSE
956	DO i = nxlg, nxrg
957	DO j = nysg, nyng
958	local_pf(i,j,nzb+1) = rad_net_av(j,i)
959	ENDDO
960	ENDDO
961	ENDIF
962	resorted = .TRUE.
963	two_d = .TRUE.
964	level_z(nzb+1) = zu(nzb+1)
965
966
967	CASE ( 'rad_lw_in_xy', 'rad_lw_in_xz', 'rad_lw_in_yz' )
968	IF ( av == 0 ) THEN
969	to_be_resorted => rad_lw_in
970	ELSE
971	to_be_resorted => rad_lw_in_av
972	ENDIF
973	IF ( mode == 'xy' ) level_z = zu
974
975	CASE ( 'rad_lw_out_xy', 'rad_lw_out_xz', 'rad_lw_out_yz' )
976	IF ( av == 0 ) THEN
977	to_be_resorted => rad_lw_out
978	ELSE
979	to_be_resorted => rad_lw_out_av
980	ENDIF
981	IF ( mode == 'xy' ) level_z = zu
982
983	CASE ( 'rad_lw_cs_hr_xy', 'rad_lw_cs_hr_xz', 'rad_lw_cs_hr_yz' )
984	IF ( av == 0 ) THEN
985	to_be_resorted => rad_lw_cs_hr
986	ELSE
987	to_be_resorted => rad_lw_cs_hr_av
988	ENDIF
989	IF ( mode == 'xy' ) level_z = zw
990
991	CASE ( 'rad_lw_hr_xy', 'rad_lw_hr_xz', 'rad_lw_hr_yz' )
992	IF ( av == 0 ) THEN
993	to_be_resorted => rad_lw_hr
994	ELSE
995	to_be_resorted => rad_lw_hr_av
996	ENDIF
997	IF ( mode == 'xy' ) level_z = zw
998
999	CASE ( 'rad_sw_in_xy', 'rad_sw_in_xz', 'rad_sw_in_yz' )
1000	IF ( av == 0 ) THEN
1001	to_be_resorted => rad_sw_in
1002	ELSE
1003	to_be_resorted => rad_sw_in_av
1004	ENDIF
1005	IF ( mode == 'xy' ) level_z = zu
1006
1007	CASE ( 'rad_sw_out_xy', 'rad_sw_out_xz', 'rad_sw_out_yz' )
1008	IF ( av == 0 ) THEN
1009	to_be_resorted => rad_sw_out
1010	ELSE
1011	to_be_resorted => rad_sw_out_av
1012	ENDIF
1013	IF ( mode == 'xy' ) level_z = zu
1014
1015	CASE ( 'rad_sw_cs_hr_xy', 'rad_sw_cs_hr_xz', 'rad_sw_cs_hr_yz' )
1016	IF ( av == 0 ) THEN
1017	to_be_resorted => rad_sw_cs_hr
1018	ELSE
1019	to_be_resorted => rad_sw_cs_hr_av
1020	ENDIF
1021	IF ( mode == 'xy' ) level_z = zw
1022
1023	CASE ( 'rad_sw_hr_xy', 'rad_sw_hr_xz', 'rad_sw_hr_yz' )
1024	IF ( av == 0 ) THEN
1025	to_be_resorted => rad_sw_hr
1026	ELSE
1027	to_be_resorted => rad_sw_hr_av
1028	ENDIF
1029	IF ( mode == 'xy' ) level_z = zw
1030
1031	CASE ( 'rho_xy', 'rho_xz', 'rho_yz' )
1032	IF ( av == 0 ) THEN
1033	to_be_resorted => rho
1034	ELSE
1035	to_be_resorted => rho_av
1036	ENDIF
1037
1038	CASE ( 'r_a*_xy' ) ! 2d-array
1039	IF ( av == 0 ) THEN
1040	DO i = nxlg, nxrg
1041	DO j = nysg, nyng
1042	local_pf(i,j,nzb+1) = r_a(j,i)
1043	ENDDO
1044	ENDDO
1045	ELSE
1046	DO i = nxlg, nxrg
1047	DO j = nysg, nyng
1048	local_pf(i,j,nzb+1) = r_a_av(j,i)
1049	ENDDO
1050	ENDDO
1051	ENDIF
1052	resorted = .TRUE.
1053	two_d = .TRUE.
1054	level_z(nzb+1) = zu(nzb+1)
1055
1056	CASE ( 'r_s*_xy' ) ! 2d-array
1057	IF ( av == 0 ) THEN
1058	DO i = nxlg, nxrg
1059	DO j = nysg, nyng
1060	local_pf(i,j,nzb+1) = r_s(j,i)
1061	ENDDO
1062	ENDDO
1063	ELSE
1064	DO i = nxlg, nxrg
1065	DO j = nysg, nyng
1066	local_pf(i,j,nzb+1) = r_s_av(j,i)
1067	ENDDO
1068	ENDDO
1069	ENDIF
1070	resorted = .TRUE.
1071	two_d = .TRUE.
1072	level_z(nzb+1) = zu(nzb+1)
1073
1074	CASE ( 's_xy', 's_xz', 's_yz' )
1075	IF ( av == 0 ) THEN
1076	to_be_resorted => s
1077	ELSE
1078	to_be_resorted => s_av
1079	ENDIF
1080
1081	CASE ( 'sa_xy', 'sa_xz', 'sa_yz' )
1082	IF ( av == 0 ) THEN
1083	to_be_resorted => sa
1084	ELSE
1085	to_be_resorted => sa_av
1086	ENDIF
1087
1088	CASE ( 'shf*_xy' ) ! 2d-array
1089	IF ( av == 0 ) THEN
1090	DO i = nxlg, nxrg
1091	DO j = nysg, nyng
1092	local_pf(i,j,nzb+1) = shf(j,i)
1093	ENDDO
1094	ENDDO
1095	ELSE
1096	DO i = nxlg, nxrg
1097	DO j = nysg, nyng
1098	local_pf(i,j,nzb+1) = shf_av(j,i)
1099	ENDDO
1100	ENDDO
1101	ENDIF
1102	resorted = .TRUE.
1103	two_d = .TRUE.
1104	level_z(nzb+1) = zu(nzb+1)
1105
1106	CASE ( 'shf_eb*_xy' ) ! 2d-array
1107	IF ( av == 0 ) THEN
1108	DO i = nxlg, nxrg
1109	DO j = nysg, nyng
1110	local_pf(i,j,nzb+1) = shf_eb(j,i)
1111	ENDDO
1112	ENDDO
1113	ELSE
1114	DO i = nxlg, nxrg
1115	DO j = nysg, nyng
1116	local_pf(i,j,nzb+1) = shf_eb_av(j,i)
1117	ENDDO
1118	ENDDO
1119	ENDIF
1120	resorted = .TRUE.
1121	two_d = .TRUE.
1122	level_z(nzb+1) = zu(nzb+1)
1123
1124	CASE ( 'ssws*_xy' ) ! 2d-array
1125	IF ( av == 0 ) THEN
1126	DO i = nxlg, nxrg
1127	DO j = nysg, nyng
1128	local_pf(i,j,nzb+1) = ssws(j,i)
1129	ENDDO
1130	ENDDO
1131	ELSE
1132	DO i = nxlg, nxrg
1133	DO j = nysg, nyng
1134	local_pf(i,j,nzb+1) = ssws_av(j,i)
1135	ENDDO
1136	ENDDO
1137	ENDIF
1138	resorted = .TRUE.
1139	two_d = .TRUE.
1140	level_z(nzb+1) = zu(nzb+1)
1141
1142	CASE ( 't*_xy' ) ! 2d-array
1143	IF ( av == 0 ) THEN
1144	DO i = nxlg, nxrg
1145	DO j = nysg, nyng
1146	local_pf(i,j,nzb+1) = ts(j,i)
1147	ENDDO
1148	ENDDO
1149	ELSE
1150	DO i = nxlg, nxrg
1151	DO j = nysg, nyng
1152	local_pf(i,j,nzb+1) = ts_av(j,i)
1153	ENDDO
1154	ENDDO
1155	ENDIF
1156	resorted = .TRUE.
1157	two_d = .TRUE.
1158	level_z(nzb+1) = zu(nzb+1)
1159
1160	CASE ( 't_soil_xy', 't_soil_xz', 't_soil_yz' )
1161	nzb_do = nzb_soil
1162	nzt_do = nzt_soil
1163	IF ( av == 0 ) THEN
1164	to_be_resorted => t_soil
1165	ELSE
1166	to_be_resorted => t_soil_av
1167	ENDIF
1168	IF ( mode == 'xy' ) level_z = zs
1169
1170	CASE ( 'u_xy', 'u_xz', 'u_yz' )
1171	IF ( av == 0 ) THEN
1172	to_be_resorted => u
1173	ELSE
1174	to_be_resorted => u_av
1175	ENDIF
1176	IF ( mode == 'xy' ) level_z = zu
1177	!
1178	!-- Substitute the values generated by "mirror" boundary condition
1179	!-- at the bottom boundary by the real surface values.
1180	IF ( do2d(av,if) == 'u_xz' .OR. do2d(av,if) == 'u_yz' ) THEN
1181	IF ( ibc_uv_b == 0 ) local_pf(:,:,nzb) = 0.0_wp
1182	ENDIF
1183
1184	CASE ( 'u*_xy' ) ! 2d-array
1185	IF ( av == 0 ) THEN
1186	DO i = nxlg, nxrg
1187	DO j = nysg, nyng
1188	local_pf(i,j,nzb+1) = us(j,i)
1189	ENDDO
1190	ENDDO
1191	ELSE
1192	DO i = nxlg, nxrg
1193	DO j = nysg, nyng
1194	local_pf(i,j,nzb+1) = us_av(j,i)
1195	ENDDO
1196	ENDDO
1197	ENDIF
1198	resorted = .TRUE.
1199	two_d = .TRUE.
1200	level_z(nzb+1) = zu(nzb+1)
1201
1202	CASE ( 'v_xy', 'v_xz', 'v_yz' )
1203	IF ( av == 0 ) THEN
1204	to_be_resorted => v
1205	ELSE
1206	to_be_resorted => v_av
1207	ENDIF
1208	IF ( mode == 'xy' ) level_z = zu
1209	!
1210	!-- Substitute the values generated by "mirror" boundary condition
1211	!-- at the bottom boundary by the real surface values.
1212	IF ( do2d(av,if) == 'v_xz' .OR. do2d(av,if) == 'v_yz' ) THEN
1213	IF ( ibc_uv_b == 0 ) local_pf(:,:,nzb) = 0.0_wp
1214	ENDIF
1215
1216	CASE ( 'vpt_xy', 'vpt_xz', 'vpt_yz' )
1217	IF ( av == 0 ) THEN
1218	to_be_resorted => vpt
1219	ELSE
1220	to_be_resorted => vpt_av
1221	ENDIF
1222	IF ( mode == 'xy' ) level_z = zu
1223
1224	CASE ( 'w_xy', 'w_xz', 'w_yz' )
1225	IF ( av == 0 ) THEN
1226	to_be_resorted => w
1227	ELSE
1228	to_be_resorted => w_av
1229	ENDIF
1230	IF ( mode == 'xy' ) level_z = zw
1231
1232	CASE ( 'z0*_xy' ) ! 2d-array
1233	IF ( av == 0 ) THEN
1234	DO i = nxlg, nxrg
1235	DO j = nysg, nyng
1236	local_pf(i,j,nzb+1) = z0(j,i)
1237	ENDDO
1238	ENDDO
1239	ELSE
1240	DO i = nxlg, nxrg
1241	DO j = nysg, nyng
1242	local_pf(i,j,nzb+1) = z0_av(j,i)
1243	ENDDO
1244	ENDDO
1245	ENDIF
1246	resorted = .TRUE.
1247	two_d = .TRUE.
1248	level_z(nzb+1) = zu(nzb+1)
1249
1250	CASE ( 'z0h*_xy' ) ! 2d-array
1251	IF ( av == 0 ) THEN
1252	DO i = nxlg, nxrg
1253	DO j = nysg, nyng
1254	local_pf(i,j,nzb+1) = z0h(j,i)
1255	ENDDO
1256	ENDDO
1257	ELSE
1258	DO i = nxlg, nxrg
1259	DO j = nysg, nyng
1260	local_pf(i,j,nzb+1) = z0h_av(j,i)
1261	ENDDO
1262	ENDDO
1263	ENDIF
1264	resorted = .TRUE.
1265	two_d = .TRUE.
1266	level_z(nzb+1) = zu(nzb+1)
1267
1268	CASE ( 'z0q*_xy' ) ! 2d-array
1269	IF ( av == 0 ) THEN
1270	DO i = nxlg, nxrg
1271	DO j = nysg, nyng
1272	local_pf(i,j,nzb+1) = z0q(j,i)
1273	ENDDO
1274	ENDDO
1275	ELSE
1276	DO i = nxlg, nxrg
1277	DO j = nysg, nyng
1278	local_pf(i,j,nzb+1) = z0q_av(j,i)
1279	ENDDO
1280	ENDDO
1281	ENDIF
1282	resorted = .TRUE.
1283	two_d = .TRUE.
1284	level_z(nzb+1) = zu(nzb+1)
1285
1286	CASE DEFAULT
1287	!
1288	!-- User defined quantity
1289	CALL user_data_output_2d( av, do2d(av,if), found, grid, &
1290	local_pf, two_d, nzb_do, nzt_do )
1291	resorted = .TRUE.
1292
1293	IF ( grid == 'zu' ) THEN
1294	IF ( mode == 'xy' ) level_z = zu
1295	ELSEIF ( grid == 'zw' ) THEN
1296	IF ( mode == 'xy' ) level_z = zw
1297	ELSEIF ( grid == 'zu1' ) THEN
1298	IF ( mode == 'xy' ) level_z(nzb+1) = zu(nzb+1)
1299	ELSEIF ( grid == 'zs' ) THEN
1300	IF ( mode == 'xy' ) level_z = zs
1301	ENDIF
1302
1303	IF ( .NOT. found ) THEN
1304	message_string = 'no output provided for: ' // &
1305	TRIM( do2d(av,if) )
1306	CALL message( 'data_output_2d', 'PA0181', 0, 0, 0, 6, 0 )
1307	ENDIF
1308
1309	END SELECT
1310
1311	!
1312	!-- Resort the array to be output, if not done above
1313	IF ( .NOT. resorted ) THEN
1314	DO i = nxlg, nxrg
1315	DO j = nysg, nyng
1316	DO k = nzb_do, nzt_do
1317	local_pf(i,j,k) = to_be_resorted(k,j,i)
1318	ENDDO
1319	ENDDO
1320	ENDDO
1321	ENDIF
1322
1323	!
1324	!-- Output of the individual cross-sections, depending on the cross-
1325	!-- section mode chosen.
1326	is = 1
1327	loop1: DO WHILE ( section(is,s_ind) /= -9999 .OR. two_d )
1328
1329	SELECT CASE ( mode )
1330
1331	CASE ( 'xy' )
1332	!
1333	!-- Determine the cross section index
1334	IF ( two_d ) THEN
1335	layer_xy = nzb+1
1336	ELSE
1337	layer_xy = section(is,s_ind)
1338	ENDIF
1339
1340	!
1341	!-- Exit the loop for layers beyond the data output domain
1342	!-- (used for soil model)
1343	IF ( layer_xy > nzt_do ) THEN
1344	EXIT loop1
1345	ENDIF
1346
1347	!
1348	!-- Update the netCDF xy cross section time axis.
1349	!-- In case of parallel output, this is only done by PE0
1350	!-- to increase the performance.
1351	IF ( simulated_time /= do2d_xy_last_time(av) ) THEN
1352	do2d_xy_time_count(av) = do2d_xy_time_count(av) + 1
1353	do2d_xy_last_time(av) = simulated_time
1354	IF ( myid == 0 ) THEN
1355	IF ( .NOT. data_output_2d_on_each_pe &
1356	.OR. netcdf_data_format > 4 ) &
1357	THEN
1358	#if defined( __netcdf )
1359	nc_stat = NF90_PUT_VAR( id_set_xy(av), &
1360	id_var_time_xy(av), &
1361	(/ time_since_reference_point /), &
1362	start = (/ do2d_xy_time_count(av) /), &
1363	count = (/ 1 /) )
1364	CALL netcdf_handle_error( 'data_output_2d', 53 )
1365	#endif
1366	ENDIF
1367	ENDIF
1368	ENDIF
1369	!
1370	!-- If required, carry out averaging along z
1371	IF ( section(is,s_ind) == -1 .AND. .NOT. two_d ) THEN
1372
1373	local_2d = 0.0_wp
1374	!
1375	!-- Carry out the averaging (all data are on the PE)
1376	DO k = nzb_do, nzt_do
1377	DO j = nysg, nyng
1378	DO i = nxlg, nxrg
1379	local_2d(i,j) = local_2d(i,j) + local_pf(i,j,k)
1380	ENDDO
1381	ENDDO
1382	ENDDO
1383
1384	local_2d = local_2d / ( nzt_do - nzb_do + 1.0_wp)
1385
1386	ELSE
1387	!
1388	!-- Just store the respective section on the local array
1389	local_2d = local_pf(:,:,layer_xy)
1390
1391	ENDIF
1392
1393	#if defined( __parallel )
1394	IF ( netcdf_data_format > 4 ) THEN
1395	!
1396	!-- Parallel output in netCDF4/HDF5 format.
1397	IF ( two_d ) THEN
1398	iis = 1
1399	ELSE
1400	iis = is
1401	ENDIF
1402
1403	#if defined( __netcdf )
1404	!
1405	!-- For parallel output, all cross sections are first stored
1406	!-- here on a local array and will be written to the output
1407	!-- file afterwards to increase the performance.
1408	DO i = nxlg, nxrg
1409	DO j = nysg, nyng
1410	local_2d_sections(i,j,iis) = local_2d(i,j)
1411	ENDDO
1412	ENDDO
1413	#endif
1414	ELSE
1415
1416	IF ( data_output_2d_on_each_pe ) THEN
1417	!
1418	!-- Output of partial arrays on each PE
1419	#if defined( __netcdf )
1420	IF ( myid == 0 ) THEN
1421	WRITE ( 21 ) time_since_reference_point, &
1422	do2d_xy_time_count(av), av
1423	ENDIF
1424	#endif
1425	DO i = 0, io_blocks-1
1426	IF ( i == io_group ) THEN
1427	WRITE ( 21 ) nxlg, nxrg, nysg, nyng, nysg, nyng
1428	WRITE ( 21 ) local_2d
1429	ENDIF
1430	#if defined( __parallel )
1431	CALL MPI_BARRIER( comm2d, ierr )
1432	#endif
1433	ENDDO
1434
1435	ELSE
1436	!
1437	!-- PE0 receives partial arrays from all processors and
1438	!-- then outputs them. Here a barrier has to be set,
1439	!-- because otherwise "-MPI- FATAL: Remote protocol queue
1440	!-- full" may occur.
1441	CALL MPI_BARRIER( comm2d, ierr )
1442
1443	ngp = ( nxrg-nxlg+1 ) * ( nyng-nysg+1 )
1444	IF ( myid == 0 ) THEN
1445	!
1446	!-- Local array can be relocated directly.
1447	total_2d(nxlg:nxrg,nysg:nyng) = local_2d
1448	!
1449	!-- Receive data from all other PEs.
1450	DO n = 1, numprocs-1
1451	!
1452	!-- Receive index limits first, then array.
1453	!-- Index limits are received in arbitrary order from
1454	!-- the PEs.
1455	CALL MPI_RECV( ind(1), 4, MPI_INTEGER, &
1456	MPI_ANY_SOURCE, 0, comm2d, &
1457	status, ierr )
1458	sender = status(MPI_SOURCE)
1459	DEALLOCATE( local_2d )
1460	ALLOCATE( local_2d(ind(1):ind(2),ind(3):ind(4)) )
1461	CALL MPI_RECV( local_2d(ind(1),ind(3)), ngp, &
1462	MPI_REAL, sender, 1, comm2d, &
1463	status, ierr )
1464	total_2d(ind(1):ind(2),ind(3):ind(4)) = local_2d
1465	ENDDO
1466	!
1467	!-- Relocate the local array for the next loop increment
1468	DEALLOCATE( local_2d )
1469	ALLOCATE( local_2d(nxlg:nxrg,nysg:nyng) )
1470
1471	#if defined( __netcdf )
1472	IF ( two_d ) THEN
1473	nc_stat = NF90_PUT_VAR( id_set_xy(av), &
1474	id_var_do2d(av,if), &
1475	total_2d(0:nx+1,0:ny+1), &
1476	start = (/ 1, 1, 1, do2d_xy_time_count(av) /), &
1477	count = (/ nx+2, ny+2, 1, 1 /) )
1478	ELSE
1479	nc_stat = NF90_PUT_VAR( id_set_xy(av), &
1480	id_var_do2d(av,if), &
1481	total_2d(0:nx+1,0:ny+1), &
1482	start = (/ 1, 1, is, do2d_xy_time_count(av) /), &
1483	count = (/ nx+2, ny+2, 1, 1 /) )
1484	ENDIF
1485	CALL netcdf_handle_error( 'data_output_2d', 54 )
1486	#endif
1487
1488	ELSE
1489	!
1490	!-- First send the local index limits to PE0
1491	ind(1) = nxlg; ind(2) = nxrg
1492	ind(3) = nysg; ind(4) = nyng
1493	CALL MPI_SEND( ind(1), 4, MPI_INTEGER, 0, 0, &
1494	comm2d, ierr )
1495	!
1496	!-- Send data to PE0
1497	CALL MPI_SEND( local_2d(nxlg,nysg), ngp, &
1498	MPI_REAL, 0, 1, comm2d, ierr )
1499	ENDIF
1500	!
1501	!-- A barrier has to be set, because otherwise some PEs may
1502	!-- proceed too fast so that PE0 may receive wrong data on
1503	!-- tag 0
1504	CALL MPI_BARRIER( comm2d, ierr )
1505	ENDIF
1506
1507	ENDIF
1508	#else
1509	#if defined( __netcdf )
1510	IF ( two_d ) THEN
1511	nc_stat = NF90_PUT_VAR( id_set_xy(av), &
1512	id_var_do2d(av,if), &
1513	local_2d(nxl:nxr+1,nys:nyn+1), &
1514	start = (/ 1, 1, 1, do2d_xy_time_count(av) /), &
1515	count = (/ nx+2, ny+2, 1, 1 /) )
1516	ELSE
1517	nc_stat = NF90_PUT_VAR( id_set_xy(av), &
1518	id_var_do2d(av,if), &
1519	local_2d(nxl:nxr+1,nys:nyn+1), &
1520	start = (/ 1, 1, is, do2d_xy_time_count(av) /), &
1521	count = (/ nx+2, ny+2, 1, 1 /) )
1522	ENDIF
1523	CALL netcdf_handle_error( 'data_output_2d', 447 )
1524	#endif
1525	#endif
1526	do2d_xy_n = do2d_xy_n + 1
1527	!
1528	!-- For 2D-arrays (e.g. u*) only one cross-section is available.
1529	!-- Hence exit loop of output levels.
1530	IF ( two_d ) THEN
1531	IF ( netcdf_data_format < 5 ) two_d = .FALSE.
1532	EXIT loop1
1533	ENDIF
1534
1535	CASE ( 'xz' )
1536	!
1537	!-- Update the netCDF xz cross section time axis.
1538	!-- In case of parallel output, this is only done by PE0
1539	!-- to increase the performance.
1540	IF ( simulated_time /= do2d_xz_last_time(av) ) THEN
1541	do2d_xz_time_count(av) = do2d_xz_time_count(av) + 1
1542	do2d_xz_last_time(av) = simulated_time
1543	IF ( myid == 0 ) THEN
1544	IF ( .NOT. data_output_2d_on_each_pe &
1545	.OR. netcdf_data_format > 4 ) &
1546	THEN
1547	#if defined( __netcdf )
1548	nc_stat = NF90_PUT_VAR( id_set_xz(av), &
1549	id_var_time_xz(av), &
1550	(/ time_since_reference_point /), &
1551	start = (/ do2d_xz_time_count(av) /), &
1552	count = (/ 1 /) )
1553	CALL netcdf_handle_error( 'data_output_2d', 56 )
1554	#endif
1555	ENDIF
1556	ENDIF
1557	ENDIF
1558
1559	!
1560	!-- If required, carry out averaging along y
1561	IF ( section(is,s_ind) == -1 ) THEN
1562
1563	ALLOCATE( local_2d_l(nxlg:nxrg,nzb_do:nzt_do) )
1564	local_2d_l = 0.0_wp
1565	ngp = ( nxrg-nxlg + 1 ) * ( nzt_do-nzb_do + 1 )
1566	!
1567	!-- First local averaging on the PE
1568	DO k = nzb_do, nzt_do
1569	DO j = nys, nyn
1570	DO i = nxlg, nxrg
1571	local_2d_l(i,k) = local_2d_l(i,k) + &
1572	local_pf(i,j,k)
1573	ENDDO
1574	ENDDO
1575	ENDDO
1576	#if defined( __parallel )
1577	!
1578	!-- Now do the averaging over all PEs along y
1579	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
1580	CALL MPI_ALLREDUCE( local_2d_l(nxlg,nzb_do), &
1581	local_2d(nxlg,nzb_do), ngp, MPI_REAL, &
1582	MPI_SUM, comm1dy, ierr )
1583	#else
1584	local_2d = local_2d_l
1585	#endif
1586	local_2d = local_2d / ( ny + 1.0_wp )
1587
1588	DEALLOCATE( local_2d_l )
1589
1590	ELSE
1591	!
1592	!-- Just store the respective section on the local array
1593	!-- (but only if it is available on this PE!)
1594	IF ( section(is,s_ind) >= nys .AND. section(is,s_ind) <= nyn ) &
1595	THEN
1596	local_2d = local_pf(:,section(is,s_ind),nzb_do:nzt_do)
1597	ENDIF
1598
1599	ENDIF
1600
1601	#if defined( __parallel )
1602	IF ( netcdf_data_format > 4 ) THEN
1603	!
1604	!-- Output in netCDF4/HDF5 format.
1605	!-- Output only on those PEs where the respective cross
1606	!-- sections reside. Cross sections averaged along y are
1607	!-- output on the respective first PE along y (myidy=0).
1608	IF ( ( section(is,s_ind) >= nys .AND. &
1609	section(is,s_ind) <= nyn ) .OR. &
1610	( section(is,s_ind) == -1 .AND. myidy == 0 ) ) THEN
1611	#if defined( __netcdf )
1612	!
1613	!-- For parallel output, all cross sections are first
1614	!-- stored here on a local array and will be written to the
1615	!-- output file afterwards to increase the performance.
1616	DO i = nxlg, nxrg
1617	DO k = nzb_do, nzt_do
1618	local_2d_sections_l(i,is,k) = local_2d(i,k)
1619	ENDDO
1620	ENDDO
1621	#endif
1622	ENDIF
1623
1624	ELSE
1625
1626	IF ( data_output_2d_on_each_pe ) THEN
1627	!
1628	!-- Output of partial arrays on each PE. If the cross
1629	!-- section does not reside on the PE, output special
1630	!-- index values.
1631	#if defined( __netcdf )
1632	IF ( myid == 0 ) THEN
1633	WRITE ( 22 ) time_since_reference_point, &
1634	do2d_xz_time_count(av), av
1635	ENDIF
1636	#endif
1637	DO i = 0, io_blocks-1
1638	IF ( i == io_group ) THEN
1639	IF ( ( section(is,s_ind) >= nys .AND. &
1640	section(is,s_ind) <= nyn ) .OR. &
1641	( section(is,s_ind) == -1 .AND. &
1642	nys-1 == -1 ) ) &
1643	THEN
1644	WRITE (22) nxlg, nxrg, nzb_do, nzt_do, nzb, nzt+1
1645	WRITE (22) local_2d
1646	ELSE
1647	WRITE (22) -1, -1, -1, -1, -1, -1
1648	ENDIF
1649	ENDIF
1650	#if defined( __parallel )
1651	CALL MPI_BARRIER( comm2d, ierr )
1652	#endif
1653	ENDDO
1654
1655	ELSE
1656	!
1657	!-- PE0 receives partial arrays from all processors of the
1658	!-- respective cross section and outputs them. Here a
1659	!-- barrier has to be set, because otherwise
1660	!-- "-MPI- FATAL: Remote protocol queue full" may occur.
1661	CALL MPI_BARRIER( comm2d, ierr )
1662
1663	ngp = ( nxrg-nxlg + 1 ) * ( nzt_do-nzb_do + 1 )
1664	IF ( myid == 0 ) THEN
1665	!
1666	!-- Local array can be relocated directly.
1667	IF ( ( section(is,s_ind) >= nys .AND. &
1668	section(is,s_ind) <= nyn ) .OR. &
1669	( section(is,s_ind) == -1 .AND. &
1670	nys-1 == -1 ) ) THEN
1671	total_2d(nxlg:nxrg,nzb_do:nzt_do) = local_2d
1672	ENDIF
1673	!
1674	!-- Receive data from all other PEs.
1675	DO n = 1, numprocs-1
1676	!
1677	!-- Receive index limits first, then array.
1678	!-- Index limits are received in arbitrary order from
1679	!-- the PEs.
1680	CALL MPI_RECV( ind(1), 4, MPI_INTEGER, &
1681	MPI_ANY_SOURCE, 0, comm2d, &
1682	status, ierr )
1683	!
1684	!-- Not all PEs have data for XZ-cross-section.
1685	IF ( ind(1) /= -9999 ) THEN
1686	sender = status(MPI_SOURCE)
1687	DEALLOCATE( local_2d )
1688	ALLOCATE( local_2d(ind(1):ind(2), &
1689	ind(3):ind(4)) )
1690	CALL MPI_RECV( local_2d(ind(1),ind(3)), ngp, &
1691	MPI_REAL, sender, 1, comm2d, &
1692	status, ierr )
1693	total_2d(ind(1):ind(2),ind(3):ind(4)) = &
1694	local_2d
1695	ENDIF
1696	ENDDO
1697	!
1698	!-- Relocate the local array for the next loop increment
1699	DEALLOCATE( local_2d )
1700	ALLOCATE( local_2d(nxlg:nxrg,nzb_do:nzt_do) )
1701
1702	#if defined( __netcdf )
1703	nc_stat = NF90_PUT_VAR( id_set_xz(av), &
1704	id_var_do2d(av,if), &
1705	total_2d(0:nx+1,nzb_do:nzt_do),&
1706	start = (/ 1, is, 1, do2d_xz_time_count(av) /), &
1707	count = (/ nx+2, 1, nzt_do-nzb_do+1, 1 /) )
1708	CALL netcdf_handle_error( 'data_output_2d', 58 )
1709	#endif
1710
1711	ELSE
1712	!
1713	!-- If the cross section resides on the PE, send the
1714	!-- local index limits, otherwise send -9999 to PE0.
1715	IF ( ( section(is,s_ind) >= nys .AND. &
1716	section(is,s_ind) <= nyn ) .OR. &
1717	( section(is,s_ind) == -1 .AND. nys-1 == -1 ) ) &
1718	THEN
1719	ind(1) = nxlg; ind(2) = nxrg
1720	ind(3) = nzb_do; ind(4) = nzt_do
1721	ELSE
1722	ind(1) = -9999; ind(2) = -9999
1723	ind(3) = -9999; ind(4) = -9999
1724	ENDIF
1725	CALL MPI_SEND( ind(1), 4, MPI_INTEGER, 0, 0, &
1726	comm2d, ierr )
1727	!
1728	!-- If applicable, send data to PE0.
1729	IF ( ind(1) /= -9999 ) THEN
1730	CALL MPI_SEND( local_2d(nxlg,nzb_do), ngp, &
1731	MPI_REAL, 0, 1, comm2d, ierr )
1732	ENDIF
1733	ENDIF
1734	!
1735	!-- A barrier has to be set, because otherwise some PEs may
1736	!-- proceed too fast so that PE0 may receive wrong data on
1737	!-- tag 0
1738	CALL MPI_BARRIER( comm2d, ierr )
1739	ENDIF
1740
1741	ENDIF
1742	#else
1743	#if defined( __netcdf )
1744	nc_stat = NF90_PUT_VAR( id_set_xz(av), &
1745	id_var_do2d(av,if), &
1746	local_2d(nxl:nxr+1,nzb_do:nzt_do), &
1747	start = (/ 1, is, 1, do2d_xz_time_count(av) /), &
1748	count = (/ nx+2, 1, nzt_do-nzb_do+1, 1 /) )
1749	CALL netcdf_handle_error( 'data_output_2d', 451 )
1750	#endif
1751	#endif
1752	do2d_xz_n = do2d_xz_n + 1
1753
1754	CASE ( 'yz' )
1755	!
1756	!-- Update the netCDF yz cross section time axis.
1757	!-- In case of parallel output, this is only done by PE0
1758	!-- to increase the performance.
1759	IF ( simulated_time /= do2d_yz_last_time(av) ) THEN
1760	do2d_yz_time_count(av) = do2d_yz_time_count(av) + 1
1761	do2d_yz_last_time(av) = simulated_time
1762	IF ( myid == 0 ) THEN
1763	IF ( .NOT. data_output_2d_on_each_pe &
1764	.OR. netcdf_data_format > 4 ) &
1765	THEN
1766	#if defined( __netcdf )
1767	nc_stat = NF90_PUT_VAR( id_set_yz(av), &
1768	id_var_time_yz(av), &
1769	(/ time_since_reference_point /), &
1770	start = (/ do2d_yz_time_count(av) /), &
1771	count = (/ 1 /) )
1772	CALL netcdf_handle_error( 'data_output_2d', 59 )
1773	#endif
1774	ENDIF
1775	ENDIF
1776	ENDIF
1777
1778	!
1779	!-- If required, carry out averaging along x
1780	IF ( section(is,s_ind) == -1 ) THEN
1781
1782	ALLOCATE( local_2d_l(nysg:nyng,nzb_do:nzt_do) )
1783	local_2d_l = 0.0_wp
1784	ngp = ( nyng-nysg+1 ) * ( nzt_do-nzb_do+1 )
1785	!
1786	!-- First local averaging on the PE
1787	DO k = nzb_do, nzt_do
1788	DO j = nysg, nyng
1789	DO i = nxl, nxr
1790	local_2d_l(j,k) = local_2d_l(j,k) + &
1791	local_pf(i,j,k)
1792	ENDDO
1793	ENDDO
1794	ENDDO
1795	#if defined( __parallel )
1796	!
1797	!-- Now do the averaging over all PEs along x
1798	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
1799	CALL MPI_ALLREDUCE( local_2d_l(nysg,nzb_do), &
1800	local_2d(nysg,nzb_do), ngp, MPI_REAL, &
1801	MPI_SUM, comm1dx, ierr )
1802	#else
1803	local_2d = local_2d_l
1804	#endif
1805	local_2d = local_2d / ( nx + 1.0_wp )
1806
1807	DEALLOCATE( local_2d_l )
1808
1809	ELSE
1810	!
1811	!-- Just store the respective section on the local array
1812	!-- (but only if it is available on this PE!)
1813	IF ( section(is,s_ind) >= nxl .AND. section(is,s_ind) <= nxr ) &
1814	THEN
1815	local_2d = local_pf(section(is,s_ind),:,nzb_do:nzt_do)
1816	ENDIF
1817
1818	ENDIF
1819
1820	#if defined( __parallel )
1821	IF ( netcdf_data_format > 4 ) THEN
1822	!
1823	!-- Output in netCDF4/HDF5 format.
1824	!-- Output only on those PEs where the respective cross
1825	!-- sections reside. Cross sections averaged along x are
1826	!-- output on the respective first PE along x (myidx=0).
1827	IF ( ( section(is,s_ind) >= nxl .AND. &
1828	section(is,s_ind) <= nxr ) .OR. &
1829	( section(is,s_ind) == -1 .AND. myidx == 0 ) ) THEN
1830	#if defined( __netcdf )
1831	!
1832	!-- For parallel output, all cross sections are first
1833	!-- stored here on a local array and will be written to the
1834	!-- output file afterwards to increase the performance.
1835	DO j = nysg, nyng
1836	DO k = nzb_do, nzt_do
1837	local_2d_sections_l(is,j,k) = local_2d(j,k)
1838	ENDDO
1839	ENDDO
1840	#endif
1841	ENDIF
1842
1843	ELSE
1844
1845	IF ( data_output_2d_on_each_pe ) THEN
1846	!
1847	!-- Output of partial arrays on each PE. If the cross
1848	!-- section does not reside on the PE, output special
1849	!-- index values.
1850	#if defined( __netcdf )
1851	IF ( myid == 0 ) THEN
1852	WRITE ( 23 ) time_since_reference_point, &
1853	do2d_yz_time_count(av), av
1854	ENDIF
1855	#endif
1856	DO i = 0, io_blocks-1
1857	IF ( i == io_group ) THEN
1858	IF ( ( section(is,s_ind) >= nxl .AND. &
1859	section(is,s_ind) <= nxr ) .OR. &
1860	( section(is,s_ind) == -1 .AND. &
1861	nxl-1 == -1 ) ) &
1862	THEN
1863	WRITE (23) nysg, nyng, nzb_do, nzt_do, nzb, nzt+1
1864	WRITE (23) local_2d
1865	ELSE
1866	WRITE (23) -1, -1, -1, -1, -1, -1
1867	ENDIF
1868	ENDIF
1869	#if defined( __parallel )
1870	CALL MPI_BARRIER( comm2d, ierr )
1871	#endif
1872	ENDDO
1873
1874	ELSE
1875	!
1876	!-- PE0 receives partial arrays from all processors of the
1877	!-- respective cross section and outputs them. Here a
1878	!-- barrier has to be set, because otherwise
1879	!-- "-MPI- FATAL: Remote protocol queue full" may occur.
1880	CALL MPI_BARRIER( comm2d, ierr )
1881
1882	ngp = ( nyng-nysg+1 ) * ( nzt_do-nzb_do+1 )
1883	IF ( myid == 0 ) THEN
1884	!
1885	!-- Local array can be relocated directly.
1886	IF ( ( section(is,s_ind) >= nxl .AND. &
1887	section(is,s_ind) <= nxr ) .OR. &
1888	( section(is,s_ind) == -1 .AND. nxl-1 == -1 ) ) &
1889	THEN
1890	total_2d(nysg:nyng,nzb_do:nzt_do) = local_2d
1891	ENDIF
1892	!
1893	!-- Receive data from all other PEs.
1894	DO n = 1, numprocs-1
1895	!
1896	!-- Receive index limits first, then array.
1897	!-- Index limits are received in arbitrary order from
1898	!-- the PEs.
1899	CALL MPI_RECV( ind(1), 4, MPI_INTEGER, &
1900	MPI_ANY_SOURCE, 0, comm2d, &
1901	status, ierr )
1902	!
1903	!-- Not all PEs have data for YZ-cross-section.
1904	IF ( ind(1) /= -9999 ) THEN
1905	sender = status(MPI_SOURCE)
1906	DEALLOCATE( local_2d )
1907	ALLOCATE( local_2d(ind(1):ind(2), &
1908	ind(3):ind(4)) )
1909	CALL MPI_RECV( local_2d(ind(1),ind(3)), ngp, &
1910	MPI_REAL, sender, 1, comm2d, &
1911	status, ierr )
1912	total_2d(ind(1):ind(2),ind(3):ind(4)) = &
1913	local_2d
1914	ENDIF
1915	ENDDO
1916	!
1917	!-- Relocate the local array for the next loop increment
1918	DEALLOCATE( local_2d )
1919	ALLOCATE( local_2d(nysg:nyng,nzb_do:nzt_do) )
1920
1921	#if defined( __netcdf )
1922	nc_stat = NF90_PUT_VAR( id_set_yz(av), &
1923	id_var_do2d(av,if), &
1924	total_2d(0:ny+1,nzb_do:nzt_do),&
1925	start = (/ is, 1, 1, do2d_yz_time_count(av) /), &
1926	count = (/ 1, ny+2, nzt_do-nzb_do+1, 1 /) )
1927	CALL netcdf_handle_error( 'data_output_2d', 61 )
1928	#endif
1929
1930	ELSE
1931	!
1932	!-- If the cross section resides on the PE, send the
1933	!-- local index limits, otherwise send -9999 to PE0.
1934	IF ( ( section(is,s_ind) >= nxl .AND. &
1935	section(is,s_ind) <= nxr ) .OR. &
1936	( section(is,s_ind) == -1 .AND. nxl-1 == -1 ) ) &
1937	THEN
1938	ind(1) = nysg; ind(2) = nyng
1939	ind(3) = nzb_do; ind(4) = nzt_do
1940	ELSE
1941	ind(1) = -9999; ind(2) = -9999
1942	ind(3) = -9999; ind(4) = -9999
1943	ENDIF
1944	CALL MPI_SEND( ind(1), 4, MPI_INTEGER, 0, 0, &
1945	comm2d, ierr )
1946	!
1947	!-- If applicable, send data to PE0.
1948	IF ( ind(1) /= -9999 ) THEN
1949	CALL MPI_SEND( local_2d(nysg,nzb_do), ngp, &
1950	MPI_REAL, 0, 1, comm2d, ierr )
1951	ENDIF
1952	ENDIF
1953	!
1954	!-- A barrier has to be set, because otherwise some PEs may
1955	!-- proceed too fast so that PE0 may receive wrong data on
1956	!-- tag 0
1957	CALL MPI_BARRIER( comm2d, ierr )
1958	ENDIF
1959
1960	ENDIF
1961	#else
1962	#if defined( __netcdf )
1963	nc_stat = NF90_PUT_VAR( id_set_yz(av), &
1964	id_var_do2d(av,if), &
1965	local_2d(nys:nyn+1,nzb_do:nzt_do), &
1966	start = (/ is, 1, 1, do2d_xz_time_count(av) /), &
1967	count = (/ 1, ny+2, nzt_do-nzb_do+1, 1 /) )
1968	CALL netcdf_handle_error( 'data_output_2d', 452 )
1969	#endif
1970	#endif
1971	do2d_yz_n = do2d_yz_n + 1
1972
1973	END SELECT
1974
1975	is = is + 1
1976	ENDDO loop1
1977
1978	!
1979	!-- For parallel output, all data were collected before on a local array
1980	!-- and are written now to the netcdf file. This must be done to increase
1981	!-- the performance of the parallel output.
1982	#if defined( __netcdf )
1983	IF ( netcdf_data_format > 4 ) THEN
1984
1985	SELECT CASE ( mode )
1986
1987	CASE ( 'xy' )
1988	IF ( two_d ) THEN
1989	nis = 1
1990	two_d = .FALSE.
1991	ELSE
1992	nis = ns
1993	ENDIF
1994	!
1995	!-- Do not output redundant ghost point data except for the
1996	!-- boundaries of the total domain.
1997	IF ( nxr == nx .AND. nyn /= ny ) THEN
1998	nc_stat = NF90_PUT_VAR( id_set_xy(av), &
1999	id_var_do2d(av,if), &
2000	local_2d_sections(nxl:nxr+1, &
2001	nys:nyn,1:nis), &
2002	start = (/ nxl+1, nys+1, 1, &
2003	do2d_xy_time_count(av) /), &
2004	count = (/ nxr-nxl+2, &
2005	nyn-nys+1, nis, 1 &
2006	/) )
2007	ELSEIF ( nxr /= nx .AND. nyn == ny ) THEN
2008	nc_stat = NF90_PUT_VAR( id_set_xy(av), &
2009	id_var_do2d(av,if), &
2010	local_2d_sections(nxl:nxr, &
2011	nys:nyn+1,1:nis), &
2012	start = (/ nxl+1, nys+1, 1, &
2013	do2d_xy_time_count(av) /), &
2014	count = (/ nxr-nxl+1, &
2015	nyn-nys+2, nis, 1 &
2016	/) )
2017	ELSEIF ( nxr == nx .AND. nyn == ny ) THEN
2018	nc_stat = NF90_PUT_VAR( id_set_xy(av), &
2019	id_var_do2d(av,if), &
2020	local_2d_sections(nxl:nxr+1, &
2021	nys:nyn+1,1:nis), &
2022	start = (/ nxl+1, nys+1, 1, &
2023	do2d_xy_time_count(av) /), &
2024	count = (/ nxr-nxl+2, &
2025	nyn-nys+2, nis, 1 &
2026	/) )
2027	ELSE
2028	nc_stat = NF90_PUT_VAR( id_set_xy(av), &
2029	id_var_do2d(av,if), &
2030	local_2d_sections(nxl:nxr, &
2031	nys:nyn,1:nis), &
2032	start = (/ nxl+1, nys+1, 1, &
2033	do2d_xy_time_count(av) /), &
2034	count = (/ nxr-nxl+1, &
2035	nyn-nys+1, nis, 1 &
2036	/) )
2037	ENDIF
2038
2039	CALL netcdf_handle_error( 'data_output_2d', 55 )
2040
2041	CASE ( 'xz' )
2042	!
2043	!-- First, all PEs get the information of all cross-sections.
2044	!-- Then the data are written to the output file by all PEs
2045	!-- while NF90_COLLECTIVE is set in subroutine
2046	!-- define_netcdf_header. Although redundant information are
2047	!-- written to the output file in that case, the performance
2048	!-- is significantly better compared to the case where only
2049	!-- the first row of PEs in x-direction (myidx = 0) is given
2050	!-- the output while NF90_INDEPENDENT is set.
2051	IF ( npey /= 1 ) THEN
2052
2053	#if defined( __parallel )
2054	!
2055	!-- Distribute data over all PEs along y
2056	ngp = ( nxrg-nxlg+1 ) * ( nzt_do-nzb_do+1 ) * ns
2057	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
2058	CALL MPI_ALLREDUCE( local_2d_sections_l(nxlg,1,nzb_do), &
2059	local_2d_sections(nxlg,1,nzb_do), &
2060	ngp, MPI_REAL, MPI_SUM, comm1dy, &
2061	ierr )
2062	#else
2063	local_2d_sections = local_2d_sections_l
2064	#endif
2065	ENDIF
2066	!
2067	!-- Do not output redundant ghost point data except for the
2068	!-- boundaries of the total domain.
2069	IF ( nxr == nx ) THEN
2070	nc_stat = NF90_PUT_VAR( id_set_xz(av), &
2071	id_var_do2d(av,if), &
2072	local_2d_sections(nxl:nxr+1,1:ns, &
2073	nzb_do:nzt_do), &
2074	start = (/ nxl+1, 1, 1, &
2075	do2d_xz_time_count(av) /), &
2076	count = (/ nxr-nxl+2, ns, nzt_do-nzb_do+1, &
2077	1 /) )
2078	ELSE
2079	nc_stat = NF90_PUT_VAR( id_set_xz(av), &
2080	id_var_do2d(av,if), &
2081	local_2d_sections(nxl:nxr,1:ns, &
2082	nzb_do:nzt_do), &
2083	start = (/ nxl+1, 1, 1, &
2084	do2d_xz_time_count(av) /), &
2085	count = (/ nxr-nxl+1, ns, nzt_do-nzb_do+1, &
2086	1 /) )
2087	ENDIF
2088
2089	CALL netcdf_handle_error( 'data_output_2d', 57 )
2090
2091	CASE ( 'yz' )
2092	!
2093	!-- First, all PEs get the information of all cross-sections.
2094	!-- Then the data are written to the output file by all PEs
2095	!-- while NF90_COLLECTIVE is set in subroutine
2096	!-- define_netcdf_header. Although redundant information are
2097	!-- written to the output file in that case, the performance
2098	!-- is significantly better compared to the case where only
2099	!-- the first row of PEs in y-direction (myidy = 0) is given
2100	!-- the output while NF90_INDEPENDENT is set.
2101	IF ( npex /= 1 ) THEN
2102
2103	#if defined( __parallel )
2104	!
2105	!-- Distribute data over all PEs along x
2106	ngp = ( nyng-nysg+1 ) * ( nzt-nzb + 2 ) * ns
2107	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
2108	CALL MPI_ALLREDUCE( local_2d_sections_l(1,nysg,nzb_do), &
2109	local_2d_sections(1,nysg,nzb_do), &
2110	ngp, MPI_REAL, MPI_SUM, comm1dx, &
2111	ierr )
2112	#else
2113	local_2d_sections = local_2d_sections_l
2114	#endif
2115	ENDIF
2116	!
2117	!-- Do not output redundant ghost point data except for the
2118	!-- boundaries of the total domain.
2119	IF ( nyn == ny ) THEN
2120	nc_stat = NF90_PUT_VAR( id_set_yz(av), &
2121	id_var_do2d(av,if), &
2122	local_2d_sections(1:ns, &
2123	nys:nyn+1,nzb_do:nzt_do), &
2124	start = (/ 1, nys+1, 1, &
2125	do2d_yz_time_count(av) /), &
2126	count = (/ ns, nyn-nys+2, &
2127	nzt_do-nzb_do+1, 1 /) )
2128	ELSE
2129	nc_stat = NF90_PUT_VAR( id_set_yz(av), &
2130	id_var_do2d(av,if), &
2131	local_2d_sections(1:ns,nys:nyn, &
2132	nzb_do:nzt_do), &
2133	start = (/ 1, nys+1, 1, &
2134	do2d_yz_time_count(av) /), &
2135	count = (/ ns, nyn-nys+1, &
2136	nzt_do-nzb_do+1, 1 /) )
2137	ENDIF
2138
2139	CALL netcdf_handle_error( 'data_output_2d', 60 )
2140
2141	CASE DEFAULT
2142	message_string = 'unknown cross-section: ' // TRIM( mode )
2143	CALL message( 'data_output_2d', 'PA0180', 1, 2, 0, 6, 0 )
2144
2145	END SELECT
2146
2147	ENDIF
2148	#endif
2149	ENDIF
2150
2151	if = if + 1
2152	l = MAX( 2, LEN_TRIM( do2d(av,if) ) )
2153	do2d_mode = do2d(av,if)(l-1:l)
2154
2155	ENDDO
2156
2157	!
2158	!-- Deallocate temporary arrays.
2159	IF ( ALLOCATED( level_z ) ) DEALLOCATE( level_z )
2160	IF ( netcdf_data_format > 4 ) THEN
2161	DEALLOCATE( local_pf, local_2d, local_2d_sections )
2162	IF( mode == 'xz' .OR. mode == 'yz' ) DEALLOCATE( local_2d_sections_l )
2163	ENDIF
2164	#if defined( __parallel )
2165	IF ( .NOT. data_output_2d_on_each_pe .AND. myid == 0 ) THEN
2166	DEALLOCATE( total_2d )
2167	ENDIF
2168	#endif
2169
2170	!
2171	!-- Close plot output file.
2172	file_id = 20 + s_ind
2173
2174	IF ( data_output_2d_on_each_pe ) THEN
2175	DO i = 0, io_blocks-1
2176	IF ( i == io_group ) THEN
2177	CALL close_file( file_id )
2178	ENDIF
2179	#if defined( __parallel )
2180	CALL MPI_BARRIER( comm2d, ierr )
2181	#endif
2182	ENDDO
2183	ELSE
2184	IF ( myid == 0 ) CALL close_file( file_id )
2185	ENDIF
2186
2187	CALL cpu_log( log_point(3), 'data_output_2d', 'stop' )
2188
2189	END SUBROUTINE data_output_2d

Note: See TracBrowser for help on using the repository browser.

Download in other formats:

| Impressum | ©Leibniz Universität Hannover |