Home

Context Navigation

source: palm/trunk/SOURCE/data_output_2d.f90 @ 4410

Last change on this file since 4410 was 4360, checked in by suehring, 5 years ago
Bugfix in output of time-averaged plant-canopy quanities; Output of plant-canopy data only where tall canopy is defined; land-surface model: fix wrong location strings; tests: update urban test case; all source code files: copyright update
Property svn:keywords set to `Id`
File size: 88.6 KB

Line
1	!> @file data_output_2d.f90
2	!------------------------------------------------------------------------------!
3	! This file is part of the PALM model system.
4	!
5	! PALM is free software: you can redistribute it and/or modify it under the
6	! terms of the GNU General Public License as published by the Free Software
7	! Foundation, either version 3 of the License, or (at your option) any later
8	! version.
9	!
10	! PALM is distributed in the hope that it will be useful, but WITHOUT ANY
11	! WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
12	! A PARTICULAR PURPOSE. See the GNU General Public License for more details.
13	!
14	! You should have received a copy of the GNU General Public License along with
15	! PALM. If not, see <http://www.gnu.org/licenses/>.
16	!
17	! Copyright 1997-2020 Leibniz Universitaet Hannover
18	!------------------------------------------------------------------------------!
19	!
20	! Current revisions:
21	! ------------------
22	!
23	!
24	! Former revisions:
25	! -----------------
26	! $Id: data_output_2d.f90 4360 2020-01-07 11:25:50Z Giersch $
27	! Introduction of wall_flags_total_0, which currently sets bits based on static
28	! topography information used in wall_flags_static_0
29	!
30	! 4331 2019-12-10 18:25:02Z suehring
31	! Move 2-m potential temperature output to diagnostic_output_quantities
32	!
33	! 4329 2019-12-10 15:46:36Z motisi
34	! Renamed wall_flags_0 to wall_flags_static_0
35	!
36	! 4182 2019-08-22 15:20:23Z scharf
37	! Corrected "Former revisions" section
38	!
39	! 4048 2019-06-21 21:00:21Z knoop
40	! Removed turbulence_closure_mod dependency
41	!
42	! 4039 2019-06-18 10:32:41Z suehring
43	! modularize diagnostic output
44	!
45	! 3994 2019-05-22 18:08:09Z suehring
46	! output of turbulence intensity added
47	!
48	! 3987 2019-05-22 09:52:13Z kanani
49	! Introduce alternative switch for debug output during timestepping
50	!
51	! 3943 2019-05-02 09:50:41Z maronga
52	! Added output of qsws for green roofs.
53	!
54	! 3885 2019-04-11 11:29:34Z kanani
55	! Changes related to global restructuring of location messages and introduction
56	! of additional debug messages
57	!
58	! 3766 2019-02-26 16:23:41Z raasch
59	! unused variables removed
60	!
61	! 3655 2019-01-07 16:51:22Z knoop
62	! Bugfix: use time_since_reference_point instead of simulated_time (relevant
63	! when using wall/soil spinup)
64	!
65	! Revision 1.1 1997/08/11 06:24:09 raasch
66	! Initial revision
67	!
68	!
69	! Description:
70	! ------------
71	!> Data output of cross-sections in netCDF format or binary format
72	!> to be later converted to NetCDF by helper routine combine_plot_fields.
73	!> Attention: The position of the sectional planes is still not always computed
74	!> --------- correctly. (zu is used always)!
75	!------------------------------------------------------------------------------!
76	SUBROUTINE data_output_2d( mode, av )
77
78
79	USE arrays_3d, &
80	ONLY: dzw, d_exner, e, heatflux_output_conversion, p, pt, q, ql, ql_c, ql_v, s, tend, u, &
81	v, vpt, w, waterflux_output_conversion, zu, zw
82
83	USE averaging
84
85	USE basic_constants_and_equations_mod, &
86	ONLY: c_p, lv_d_cp, l_v
87
88	USE bulk_cloud_model_mod, &
89	ONLY: bulk_cloud_model
90
91	USE control_parameters, &
92	ONLY: data_output_2d_on_each_pe, &
93	data_output_xy, data_output_xz, data_output_yz, &
94	debug_output_timestep, &
95	do2d, &
96	do2d_xy_last_time, do2d_xy_time_count, &
97	do2d_xz_last_time, do2d_xz_time_count, &
98	do2d_yz_last_time, do2d_yz_time_count, &
99	ibc_uv_b, io_blocks, io_group, message_string, &
100	ntdim_2d_xy, ntdim_2d_xz, ntdim_2d_yz, &
101	psolver, section, &
102	time_since_reference_point
103
104	USE cpulog, &
105	ONLY: cpu_log, log_point
106
107	USE indices, &
108	ONLY: nbgp, nx, nxl, nxlg, nxr, nxrg, ny, nyn, nyng, nys, nysg, &
109	nzb, nzt, wall_flags_total_0
110
111	USE kinds
112
113	USE land_surface_model_mod, &
114	ONLY: zs
115
116	USE module_interface, &
117	ONLY: module_interface_data_output_2d
118
119	#if defined( __netcdf )
120	USE NETCDF
121	#endif
122
123	USE netcdf_interface, &
124	ONLY: fill_value, id_set_xy, id_set_xz, id_set_yz, id_var_do2d, &
125	id_var_time_xy, id_var_time_xz, id_var_time_yz, nc_stat, &
126	netcdf_data_format, netcdf_handle_error
127
128	USE particle_attributes, &
129	ONLY: grid_particles, number_of_particles, particle_advection_start, &
130	particles, prt_count
131
132	USE pegrid
133
134	USE surface_mod, &
135	ONLY: ind_pav_green, ind_veg_wall, ind_wat_win, surf_def_h, &
136	surf_lsm_h, surf_usm_h
137
138
139	IMPLICIT NONE
140
141	CHARACTER (LEN=2) :: do2d_mode !< output mode of variable ('xy', 'xz', 'yz')
142	CHARACTER (LEN=2) :: mode !< mode with which the routine is called ('xy', 'xz', 'yz')
143	CHARACTER (LEN=4) :: grid !< string defining the vertical grid
144
145	INTEGER(iwp) :: av !< flag for (non-)average output
146	INTEGER(iwp) :: ngp !< number of grid points of an output slice
147	INTEGER(iwp) :: file_id !< id of output files
148	INTEGER(iwp) :: flag_nr !< number of masking flag
149	INTEGER(iwp) :: i !< loop index
150	INTEGER(iwp) :: iis !< vertical index of a xy slice in array 'local_2d_sections'
151	INTEGER(iwp) :: is !< slice index
152	INTEGER(iwp) :: ivar !< variable index
153	INTEGER(iwp) :: j !< loop index
154	INTEGER(iwp) :: k !< loop index
155	INTEGER(iwp) :: l !< loop index
156	INTEGER(iwp) :: layer_xy !< vertical index of a xy slice in array 'local_pf'
157	INTEGER(iwp) :: m !< loop index
158	INTEGER(iwp) :: n !< loop index
159	INTEGER(iwp) :: nis !< number of vertical slices to be written via parallel NetCDF output
160	INTEGER(iwp) :: ns !< number of output slices
161	INTEGER(iwp) :: nzb_do !< lower limit of the data field (usually nzb)
162	INTEGER(iwp) :: nzt_do !< upper limit of the data field (usually nzt+1)
163	INTEGER(iwp) :: s_ind !< index of slice types (xy=1, xz=2, yz=3)
164	INTEGER(iwp) :: sender !< PE id of sending PE
165	INTEGER(iwp) :: ind(4) !< index limits (lower/upper bounds) of array 'local_2d'
166
167	LOGICAL :: found !< true if output variable was found
168	LOGICAL :: resorted !< true if variable is resorted
169	LOGICAL :: two_d !< true if variable is only two dimensional
170
171	REAL(wp) :: mean_r !< mean particle radius
172	REAL(wp) :: s_r2 !< sum( particle-radius**2 )
173	REAL(wp) :: s_r3 !< sum( particle-radius**3 )
174
175	REAL(wp), DIMENSION(:), ALLOCATABLE :: level_z !< z levels for output array
176	REAL(wp), DIMENSION(:,:), ALLOCATABLE :: local_2d !< local 2-dimensional array containing output values
177	REAL(wp), DIMENSION(:,:), ALLOCATABLE :: local_2d_l !< local 2-dimensional array containing output values
178
179	REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: local_pf !< output array
180	REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: local_2d_sections !< local array containing values at all slices
181	REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: local_2d_sections_l !< local array containing values at all slices
182
183	#if defined( __parallel )
184	REAL(wp), DIMENSION(:,:), ALLOCATABLE :: total_2d !< same as local_2d
185	#endif
186	REAL(wp), DIMENSION(:,:,:), POINTER :: to_be_resorted !< points to array which shall be output
187
188
189	IF ( debug_output_timestep ) CALL debug_message( 'data_output_2d', 'start' )
190	!
191	!-- Immediate return, if no output is requested (no respective sections
192	!-- found in parameter data_output)
193	IF ( mode == 'xy' .AND. .NOT. data_output_xy(av) ) RETURN
194	IF ( mode == 'xz' .AND. .NOT. data_output_xz(av) ) RETURN
195	IF ( mode == 'yz' .AND. .NOT. data_output_yz(av) ) RETURN
196
197	CALL cpu_log (log_point(3),'data_output_2d','start')
198
199	two_d = .FALSE. ! local variable to distinguish between output of pure 2D
200	! arrays and cross-sections of 3D arrays.
201
202	!
203	!-- Depending on the orientation of the cross-section, the respective output
204	!-- files have to be opened.
205	SELECT CASE ( mode )
206
207	CASE ( 'xy' )
208	s_ind = 1
209	ALLOCATE( level_z(nzb:nzt+1), local_2d(nxl:nxr,nys:nyn) )
210
211	IF ( netcdf_data_format > 4 ) THEN
212	ns = 1
213	DO WHILE ( section(ns,s_ind) /= -9999 .AND. ns <= 100 )
214	ns = ns + 1
215	ENDDO
216	ns = ns - 1
217	ALLOCATE( local_2d_sections(nxl:nxr,nys:nyn,1:ns) )
218	local_2d_sections = 0.0_wp
219	ENDIF
220
221	!
222	!-- Parallel netCDF4/HDF5 output is done on all PEs, all other on PE0 only
223	IF ( myid == 0 .OR. netcdf_data_format > 4 ) THEN
224	CALL check_open( 101+av*10 )
225	ENDIF
226	IF ( data_output_2d_on_each_pe .AND. netcdf_data_format < 5 ) THEN
227	CALL check_open( 21 )
228	ELSE
229	IF ( myid == 0 ) THEN
230	#if defined( __parallel )
231	ALLOCATE( total_2d(0:nx,0:ny) )
232	#endif
233	ENDIF
234	ENDIF
235
236	CASE ( 'xz' )
237	s_ind = 2
238	ALLOCATE( local_2d(nxl:nxr,nzb:nzt+1) )
239
240	IF ( netcdf_data_format > 4 ) THEN
241	ns = 1
242	DO WHILE ( section(ns,s_ind) /= -9999 .AND. ns <= 100 )
243	ns = ns + 1
244	ENDDO
245	ns = ns - 1
246	ALLOCATE( local_2d_sections(nxl:nxr,1:ns,nzb:nzt+1) )
247	ALLOCATE( local_2d_sections_l(nxl:nxr,1:ns,nzb:nzt+1) )
248	local_2d_sections = 0.0_wp; local_2d_sections_l = 0.0_wp
249	ENDIF
250
251	!
252	!-- Parallel netCDF4/HDF5 output is done on all PEs, all other on PE0 only
253	IF ( myid == 0 .OR. netcdf_data_format > 4 ) THEN
254	CALL check_open( 102+av*10 )
255	ENDIF
256
257	IF ( data_output_2d_on_each_pe .AND. netcdf_data_format < 5 ) THEN
258	CALL check_open( 22 )
259	ELSE
260	IF ( myid == 0 ) THEN
261	#if defined( __parallel )
262	ALLOCATE( total_2d(0:nx,nzb:nzt+1) )
263	#endif
264	ENDIF
265	ENDIF
266
267	CASE ( 'yz' )
268	s_ind = 3
269	ALLOCATE( local_2d(nys:nyn,nzb:nzt+1) )
270
271	IF ( netcdf_data_format > 4 ) THEN
272	ns = 1
273	DO WHILE ( section(ns,s_ind) /= -9999 .AND. ns <= 100 )
274	ns = ns + 1
275	ENDDO
276	ns = ns - 1
277	ALLOCATE( local_2d_sections(1:ns,nys:nyn,nzb:nzt+1) )
278	ALLOCATE( local_2d_sections_l(1:ns,nys:nyn,nzb:nzt+1) )
279	local_2d_sections = 0.0_wp; local_2d_sections_l = 0.0_wp
280	ENDIF
281
282	!
283	!-- Parallel netCDF4/HDF5 output is done on all PEs, all other on PE0 only
284	IF ( myid == 0 .OR. netcdf_data_format > 4 ) THEN
285	CALL check_open( 103+av*10 )
286	ENDIF
287
288	IF ( data_output_2d_on_each_pe .AND. netcdf_data_format < 5 ) THEN
289	CALL check_open( 23 )
290	ELSE
291	IF ( myid == 0 ) THEN
292	#if defined( __parallel )
293	ALLOCATE( total_2d(0:ny,nzb:nzt+1) )
294	#endif
295	ENDIF
296	ENDIF
297
298	CASE DEFAULT
299	message_string = 'unknown cross-section: ' // TRIM( mode )
300	CALL message( 'data_output_2d', 'PA0180', 1, 2, 0, 6, 0 )
301
302	END SELECT
303
304	!
305	!-- For parallel netcdf output the time axis must be limited. Return, if this
306	!-- limit is exceeded. This could be the case, if the simulated time exceeds
307	!-- the given end time by the length of the given output interval.
308	IF ( netcdf_data_format > 4 ) THEN
309	IF ( mode == 'xy' .AND. do2d_xy_time_count(av) + 1 > &
310	ntdim_2d_xy(av) ) THEN
311	WRITE ( message_string, * ) 'Output of xy cross-sections is not ', &
312	'given at t=', time_since_reference_point, 's because the', &
313	' maximum number of output time levels is exceeded.'
314	CALL message( 'data_output_2d', 'PA0384', 0, 1, 0, 6, 0 )
315	CALL cpu_log( log_point(3), 'data_output_2d', 'stop' )
316	RETURN
317	ENDIF
318	IF ( mode == 'xz' .AND. do2d_xz_time_count(av) + 1 > &
319	ntdim_2d_xz(av) ) THEN
320	WRITE ( message_string, * ) 'Output of xz cross-sections is not ', &
321	'given at t=', time_since_reference_point, 's because the', &
322	' maximum number of output time levels is exceeded.'
323	CALL message( 'data_output_2d', 'PA0385', 0, 1, 0, 6, 0 )
324	CALL cpu_log( log_point(3), 'data_output_2d', 'stop' )
325	RETURN
326	ENDIF
327	IF ( mode == 'yz' .AND. do2d_yz_time_count(av) + 1 > &
328	ntdim_2d_yz(av) ) THEN
329	WRITE ( message_string, * ) 'Output of yz cross-sections is not ', &
330	'given at t=', time_since_reference_point, 's because the', &
331	' maximum number of output time levels is exceeded.'
332	CALL message( 'data_output_2d', 'PA0386', 0, 1, 0, 6, 0 )
333	CALL cpu_log( log_point(3), 'data_output_2d', 'stop' )
334	RETURN
335	ENDIF
336	ENDIF
337
338	!
339	!-- Allocate a temporary array for resorting (kji -> ijk).
340	ALLOCATE( local_pf(nxl:nxr,nys:nyn,nzb:nzt+1) )
341	local_pf = 0.0
342
343	!
344	!-- Loop of all variables to be written.
345	!-- Output dimensions chosen
346	ivar = 1
347	l = MAX( 2, LEN_TRIM( do2d(av,ivar) ) )
348	do2d_mode = do2d(av,ivar)(l-1:l)
349
350	DO WHILE ( do2d(av,ivar)(1:1) /= ' ' )
351
352	IF ( do2d_mode == mode ) THEN
353	!
354	!-- Set flag to steer output of radiation, land-surface, or user-defined
355	!-- quantities
356	found = .FALSE.
357
358	nzb_do = nzb
359	nzt_do = nzt+1
360	!
361	!-- Before each output, set array local_pf to fill value
362	local_pf = fill_value
363	!
364	!-- Set masking flag for topography for not resorted arrays
365	flag_nr = 0
366
367	!
368	!-- Store the array chosen on the temporary array.
369	resorted = .FALSE.
370	SELECT CASE ( TRIM( do2d(av,ivar) ) )
371	CASE ( 'e_xy', 'e_xz', 'e_yz' )
372	IF ( av == 0 ) THEN
373	to_be_resorted => e
374	ELSE
375	IF ( .NOT. ALLOCATED( e_av ) ) THEN
376	ALLOCATE( e_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
377	e_av = REAL( fill_value, KIND = wp )
378	ENDIF
379	to_be_resorted => e_av
380	ENDIF
381	IF ( mode == 'xy' ) level_z = zu
382
383	CASE ( 'thetal_xy', 'thetal_xz', 'thetal_yz' )
384	IF ( av == 0 ) THEN
385	to_be_resorted => pt
386	ELSE
387	IF ( .NOT. ALLOCATED( lpt_av ) ) THEN
388	ALLOCATE( lpt_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
389	lpt_av = REAL( fill_value, KIND = wp )
390	ENDIF
391	to_be_resorted => lpt_av
392	ENDIF
393	IF ( mode == 'xy' ) level_z = zu
394
395	CASE ( 'lwp*_xy' ) ! 2d-array
396	IF ( av == 0 ) THEN
397	DO i = nxl, nxr
398	DO j = nys, nyn
399	local_pf(i,j,nzb+1) = SUM( ql(nzb:nzt,j,i) * &
400	dzw(1:nzt+1) )
401	ENDDO
402	ENDDO
403	ELSE
404	IF ( .NOT. ALLOCATED( lwp_av ) ) THEN
405	ALLOCATE( lwp_av(nysg:nyng,nxlg:nxrg) )
406	lwp_av = REAL( fill_value, KIND = wp )
407	ENDIF
408	DO i = nxl, nxr
409	DO j = nys, nyn
410	local_pf(i,j,nzb+1) = lwp_av(j,i)
411	ENDDO
412	ENDDO
413	ENDIF
414	resorted = .TRUE.
415	two_d = .TRUE.
416	level_z(nzb+1) = zu(nzb+1)
417
418	CASE ( 'ghf*_xy' ) ! 2d-array
419	IF ( av == 0 ) THEN
420	DO m = 1, surf_lsm_h%ns
421	i = surf_lsm_h%i(m)
422	j = surf_lsm_h%j(m)
423	local_pf(i,j,nzb+1) = surf_lsm_h%ghf(m)
424	ENDDO
425	DO m = 1, surf_usm_h%ns
426	i = surf_usm_h%i(m)
427	j = surf_usm_h%j(m)
428	local_pf(i,j,nzb+1) = surf_usm_h%frac(ind_veg_wall,m) * &
429	surf_usm_h%wghf_eb(m) + &
430	surf_usm_h%frac(ind_pav_green,m) * &
431	surf_usm_h%wghf_eb_green(m) + &
432	surf_usm_h%frac(ind_wat_win,m) * &
433	surf_usm_h%wghf_eb_window(m)
434	ENDDO
435	ELSE
436	IF ( .NOT. ALLOCATED( ghf_av ) ) THEN
437	ALLOCATE( ghf_av(nysg:nyng,nxlg:nxrg) )
438	ghf_av = REAL( fill_value, KIND = wp )
439	ENDIF
440	DO i = nxl, nxr
441	DO j = nys, nyn
442	local_pf(i,j,nzb+1) = ghf_av(j,i)
443	ENDDO
444	ENDDO
445	ENDIF
446
447	resorted = .TRUE.
448	two_d = .TRUE.
449	level_z(nzb+1) = zu(nzb+1)
450
451	CASE ( 'ol*_xy' ) ! 2d-array
452	IF ( av == 0 ) THEN
453	DO m = 1, surf_def_h(0)%ns
454	i = surf_def_h(0)%i(m)
455	j = surf_def_h(0)%j(m)
456	local_pf(i,j,nzb+1) = surf_def_h(0)%ol(m)
457	ENDDO
458	DO m = 1, surf_lsm_h%ns
459	i = surf_lsm_h%i(m)
460	j = surf_lsm_h%j(m)
461	local_pf(i,j,nzb+1) = surf_lsm_h%ol(m)
462	ENDDO
463	DO m = 1, surf_usm_h%ns
464	i = surf_usm_h%i(m)
465	j = surf_usm_h%j(m)
466	local_pf(i,j,nzb+1) = surf_usm_h%ol(m)
467	ENDDO
468	ELSE
469	IF ( .NOT. ALLOCATED( ol_av ) ) THEN
470	ALLOCATE( ol_av(nysg:nyng,nxlg:nxrg) )
471	ol_av = REAL( fill_value, KIND = wp )
472	ENDIF
473	DO i = nxl, nxr
474	DO j = nys, nyn
475	local_pf(i,j,nzb+1) = ol_av(j,i)
476	ENDDO
477	ENDDO
478	ENDIF
479	resorted = .TRUE.
480	two_d = .TRUE.
481	level_z(nzb+1) = zu(nzb+1)
482
483	CASE ( 'p_xy', 'p_xz', 'p_yz' )
484	IF ( av == 0 ) THEN
485	IF ( psolver /= 'sor' ) CALL exchange_horiz( p, nbgp )
486	to_be_resorted => p
487	ELSE
488	IF ( .NOT. ALLOCATED( p_av ) ) THEN
489	ALLOCATE( p_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
490	p_av = REAL( fill_value, KIND = wp )
491	ENDIF
492	IF ( psolver /= 'sor' ) CALL exchange_horiz( p_av, nbgp )
493	to_be_resorted => p_av
494	ENDIF
495	IF ( mode == 'xy' ) level_z = zu
496
497	CASE ( 'pc_xy', 'pc_xz', 'pc_yz' ) ! particle concentration
498	IF ( av == 0 ) THEN
499	IF ( time_since_reference_point >= particle_advection_start ) THEN
500	tend = prt_count
501	! CALL exchange_horiz( tend, nbgp )
502	ELSE
503	tend = 0.0_wp
504	ENDIF
505	DO i = nxl, nxr
506	DO j = nys, nyn
507	DO k = nzb, nzt+1
508	local_pf(i,j,k) = tend(k,j,i)
509	ENDDO
510	ENDDO
511	ENDDO
512	resorted = .TRUE.
513	ELSE
514	IF ( .NOT. ALLOCATED( pc_av ) ) THEN
515	ALLOCATE( pc_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
516	pc_av = REAL( fill_value, KIND = wp )
517	ENDIF
518	! CALL exchange_horiz( pc_av, nbgp )
519	to_be_resorted => pc_av
520	ENDIF
521
522	CASE ( 'pr_xy', 'pr_xz', 'pr_yz' ) ! mean particle radius (effective radius)
523	IF ( av == 0 ) THEN
524	IF ( time_since_reference_point >= particle_advection_start ) THEN
525	DO i = nxl, nxr
526	DO j = nys, nyn
527	DO k = nzb, nzt+1
528	number_of_particles = prt_count(k,j,i)
529	IF (number_of_particles <= 0) CYCLE
530	particles => grid_particles(k,j,i)%particles(1:number_of_particles)
531	s_r2 = 0.0_wp
532	s_r3 = 0.0_wp
533	DO n = 1, number_of_particles
534	IF ( particles(n)%particle_mask ) THEN
535	s_r2 = s_r2 + particles(n)%radius*2 &
536	particles(n)%weight_factor
537	s_r3 = s_r3 + particles(n)%radius*3 &
538	particles(n)%weight_factor
539	ENDIF
540	ENDDO
541	IF ( s_r2 > 0.0_wp ) THEN
542	mean_r = s_r3 / s_r2
543	ELSE
544	mean_r = 0.0_wp
545	ENDIF
546	tend(k,j,i) = mean_r
547	ENDDO
548	ENDDO
549	ENDDO
550	! CALL exchange_horiz( tend, nbgp )
551	ELSE
552	tend = 0.0_wp
553	ENDIF
554	DO i = nxl, nxr
555	DO j = nys, nyn
556	DO k = nzb, nzt+1
557	local_pf(i,j,k) = tend(k,j,i)
558	ENDDO
559	ENDDO
560	ENDDO
561	resorted = .TRUE.
562	ELSE
563	IF ( .NOT. ALLOCATED( pr_av ) ) THEN
564	ALLOCATE( pr_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
565	pr_av = REAL( fill_value, KIND = wp )
566	ENDIF
567	! CALL exchange_horiz( pr_av, nbgp )
568	to_be_resorted => pr_av
569	ENDIF
570
571	CASE ( 'theta_xy', 'theta_xz', 'theta_yz' )
572	IF ( av == 0 ) THEN
573	IF ( .NOT. bulk_cloud_model ) THEN
574	to_be_resorted => pt
575	ELSE
576	DO i = nxl, nxr
577	DO j = nys, nyn
578	DO k = nzb, nzt+1
579	local_pf(i,j,k) = pt(k,j,i) + lv_d_cp * &
580	d_exner(k) * &
581	ql(k,j,i)
582	ENDDO
583	ENDDO
584	ENDDO
585	resorted = .TRUE.
586	ENDIF
587	ELSE
588	IF ( .NOT. ALLOCATED( pt_av ) ) THEN
589	ALLOCATE( pt_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
590	pt_av = REAL( fill_value, KIND = wp )
591	ENDIF
592	to_be_resorted => pt_av
593	ENDIF
594	IF ( mode == 'xy' ) level_z = zu
595
596	CASE ( 'q_xy', 'q_xz', 'q_yz' )
597	IF ( av == 0 ) THEN
598	to_be_resorted => q
599	ELSE
600	IF ( .NOT. ALLOCATED( q_av ) ) THEN
601	ALLOCATE( q_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
602	q_av = REAL( fill_value, KIND = wp )
603	ENDIF
604	to_be_resorted => q_av
605	ENDIF
606	IF ( mode == 'xy' ) level_z = zu
607
608	CASE ( 'ql_xy', 'ql_xz', 'ql_yz' )
609	IF ( av == 0 ) THEN
610	to_be_resorted => ql
611	ELSE
612	IF ( .NOT. ALLOCATED( ql_av ) ) THEN
613	ALLOCATE( ql_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
614	ql_av = REAL( fill_value, KIND = wp )
615	ENDIF
616	to_be_resorted => ql_av
617	ENDIF
618	IF ( mode == 'xy' ) level_z = zu
619
620	CASE ( 'ql_c_xy', 'ql_c_xz', 'ql_c_yz' )
621	IF ( av == 0 ) THEN
622	to_be_resorted => ql_c
623	ELSE
624	IF ( .NOT. ALLOCATED( ql_c_av ) ) THEN
625	ALLOCATE( ql_c_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
626	ql_c_av = REAL( fill_value, KIND = wp )
627	ENDIF
628	to_be_resorted => ql_c_av
629	ENDIF
630	IF ( mode == 'xy' ) level_z = zu
631
632	CASE ( 'ql_v_xy', 'ql_v_xz', 'ql_v_yz' )
633	IF ( av == 0 ) THEN
634	to_be_resorted => ql_v
635	ELSE
636	IF ( .NOT. ALLOCATED( ql_v_av ) ) THEN
637	ALLOCATE( ql_v_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
638	ql_v_av = REAL( fill_value, KIND = wp )
639	ENDIF
640	to_be_resorted => ql_v_av
641	ENDIF
642	IF ( mode == 'xy' ) level_z = zu
643
644	CASE ( 'ql_vp_xy', 'ql_vp_xz', 'ql_vp_yz' )
645	IF ( av == 0 ) THEN
646	IF ( time_since_reference_point >= particle_advection_start ) THEN
647	DO i = nxl, nxr
648	DO j = nys, nyn
649	DO k = nzb, nzt+1
650	number_of_particles = prt_count(k,j,i)
651	IF (number_of_particles <= 0) CYCLE
652	particles => grid_particles(k,j,i)%particles(1:number_of_particles)
653	DO n = 1, number_of_particles
654	IF ( particles(n)%particle_mask ) THEN
655	tend(k,j,i) = tend(k,j,i) + &
656	particles(n)%weight_factor / &
657	prt_count(k,j,i)
658	ENDIF
659	ENDDO
660	ENDDO
661	ENDDO
662	ENDDO
663	! CALL exchange_horiz( tend, nbgp )
664	ELSE
665	tend = 0.0_wp
666	ENDIF
667	DO i = nxl, nxr
668	DO j = nys, nyn
669	DO k = nzb, nzt+1
670	local_pf(i,j,k) = tend(k,j,i)
671	ENDDO
672	ENDDO
673	ENDDO
674	resorted = .TRUE.
675	ELSE
676	IF ( .NOT. ALLOCATED( ql_vp_av ) ) THEN
677	ALLOCATE( ql_vp_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
678	ql_vp_av = REAL( fill_value, KIND = wp )
679	ENDIF
680	! CALL exchange_horiz( ql_vp_av, nbgp )
681	to_be_resorted => ql_vp_av
682	ENDIF
683	IF ( mode == 'xy' ) level_z = zu
684
685	CASE ( 'qsws*_xy' ) ! 2d-array
686	IF ( av == 0 ) THEN
687	local_pf(:,:,nzb+1) = REAL( fill_value, KIND = wp )
688	!
689	!-- In case of default surfaces, clean-up flux by density.
690	!-- In case of land-surfaces, convert fluxes into
691	!-- dynamic units
692	DO m = 1, surf_def_h(0)%ns
693	i = surf_def_h(0)%i(m)
694	j = surf_def_h(0)%j(m)
695	k = surf_def_h(0)%k(m)
696	local_pf(i,j,nzb+1) = surf_def_h(0)%qsws(m) * &
697	waterflux_output_conversion(k)
698	ENDDO
699	DO m = 1, surf_lsm_h%ns
700	i = surf_lsm_h%i(m)
701	j = surf_lsm_h%j(m)
702	k = surf_lsm_h%k(m)
703	local_pf(i,j,nzb+1) = surf_lsm_h%qsws(m) * l_v
704	ENDDO
705	DO m = 1, surf_usm_h%ns
706	i = surf_usm_h%i(m)
707	j = surf_usm_h%j(m)
708	k = surf_usm_h%k(m)
709	local_pf(i,j,nzb+1) = surf_usm_h%qsws(m) * l_v
710	ENDDO
711	ELSE
712	IF ( .NOT. ALLOCATED( qsws_av ) ) THEN
713	ALLOCATE( qsws_av(nysg:nyng,nxlg:nxrg) )
714	qsws_av = REAL( fill_value, KIND = wp )
715	ENDIF
716	DO i = nxl, nxr
717	DO j = nys, nyn
718	local_pf(i,j,nzb+1) = qsws_av(j,i)
719	ENDDO
720	ENDDO
721	ENDIF
722	resorted = .TRUE.
723	two_d = .TRUE.
724	level_z(nzb+1) = zu(nzb+1)
725
726	CASE ( 'qv_xy', 'qv_xz', 'qv_yz' )
727	IF ( av == 0 ) THEN
728	DO i = nxl, nxr
729	DO j = nys, nyn
730	DO k = nzb, nzt+1
731	local_pf(i,j,k) = q(k,j,i) - ql(k,j,i)
732	ENDDO
733	ENDDO
734	ENDDO
735	resorted = .TRUE.
736	ELSE
737	IF ( .NOT. ALLOCATED( qv_av ) ) THEN
738	ALLOCATE( qv_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
739	qv_av = REAL( fill_value, KIND = wp )
740	ENDIF
741	to_be_resorted => qv_av
742	ENDIF
743	IF ( mode == 'xy' ) level_z = zu
744
745	CASE ( 'r_a*_xy' ) ! 2d-array
746	IF ( av == 0 ) THEN
747	DO m = 1, surf_lsm_h%ns
748	i = surf_lsm_h%i(m)
749	j = surf_lsm_h%j(m)
750	local_pf(i,j,nzb+1) = surf_lsm_h%r_a(m)
751	ENDDO
752
753	DO m = 1, surf_usm_h%ns
754	i = surf_usm_h%i(m)
755	j = surf_usm_h%j(m)
756	local_pf(i,j,nzb+1) = &
757	( surf_usm_h%frac(ind_veg_wall,m) * &
758	surf_usm_h%r_a(m) + &
759	surf_usm_h%frac(ind_pav_green,m) * &
760	surf_usm_h%r_a_green(m) + &
761	surf_usm_h%frac(ind_wat_win,m) * &
762	surf_usm_h%r_a_window(m) )
763	ENDDO
764	ELSE
765	IF ( .NOT. ALLOCATED( r_a_av ) ) THEN
766	ALLOCATE( r_a_av(nysg:nyng,nxlg:nxrg) )
767	r_a_av = REAL( fill_value, KIND = wp )
768	ENDIF
769	DO i = nxl, nxr
770	DO j = nys, nyn
771	local_pf(i,j,nzb+1) = r_a_av(j,i)
772	ENDDO
773	ENDDO
774	ENDIF
775	resorted = .TRUE.
776	two_d = .TRUE.
777	level_z(nzb+1) = zu(nzb+1)
778
779	CASE ( 's_xy', 's_xz', 's_yz' )
780	IF ( av == 0 ) THEN
781	to_be_resorted => s
782	ELSE
783	IF ( .NOT. ALLOCATED( s_av ) ) THEN
784	ALLOCATE( s_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
785	s_av = REAL( fill_value, KIND = wp )
786	ENDIF
787	to_be_resorted => s_av
788	ENDIF
789
790	CASE ( 'shf*_xy' ) ! 2d-array
791	IF ( av == 0 ) THEN
792	!
793	!-- In case of default surfaces, clean-up flux by density.
794	!-- In case of land- and urban-surfaces, convert fluxes into
795	!-- dynamic units.
796	DO m = 1, surf_def_h(0)%ns
797	i = surf_def_h(0)%i(m)
798	j = surf_def_h(0)%j(m)
799	k = surf_def_h(0)%k(m)
800	local_pf(i,j,nzb+1) = surf_def_h(0)%shf(m) * &
801	heatflux_output_conversion(k)
802	ENDDO
803	DO m = 1, surf_lsm_h%ns
804	i = surf_lsm_h%i(m)
805	j = surf_lsm_h%j(m)
806	k = surf_lsm_h%k(m)
807	local_pf(i,j,nzb+1) = surf_lsm_h%shf(m) * c_p
808	ENDDO
809	DO m = 1, surf_usm_h%ns
810	i = surf_usm_h%i(m)
811	j = surf_usm_h%j(m)
812	k = surf_usm_h%k(m)
813	local_pf(i,j,nzb+1) = surf_usm_h%shf(m) * c_p
814	ENDDO
815	ELSE
816	IF ( .NOT. ALLOCATED( shf_av ) ) THEN
817	ALLOCATE( shf_av(nysg:nyng,nxlg:nxrg) )
818	shf_av = REAL( fill_value, KIND = wp )
819	ENDIF
820	DO i = nxl, nxr
821	DO j = nys, nyn
822	local_pf(i,j,nzb+1) = shf_av(j,i)
823	ENDDO
824	ENDDO
825	ENDIF
826	resorted = .TRUE.
827	two_d = .TRUE.
828	level_z(nzb+1) = zu(nzb+1)
829
830	CASE ( 'ssws*_xy' ) ! 2d-array
831	IF ( av == 0 ) THEN
832	DO m = 1, surf_def_h(0)%ns
833	i = surf_def_h(0)%i(m)
834	j = surf_def_h(0)%j(m)
835	local_pf(i,j,nzb+1) = surf_def_h(0)%ssws(m)
836	ENDDO
837	DO m = 1, surf_lsm_h%ns
838	i = surf_lsm_h%i(m)
839	j = surf_lsm_h%j(m)
840	local_pf(i,j,nzb+1) = surf_lsm_h%ssws(m)
841	ENDDO
842	DO m = 1, surf_usm_h%ns
843	i = surf_usm_h%i(m)
844	j = surf_usm_h%j(m)
845	local_pf(i,j,nzb+1) = surf_usm_h%ssws(m)
846	ENDDO
847	ELSE
848	IF ( .NOT. ALLOCATED( ssws_av ) ) THEN
849	ALLOCATE( ssws_av(nysg:nyng,nxlg:nxrg) )
850	ssws_av = REAL( fill_value, KIND = wp )
851	ENDIF
852	DO i = nxl, nxr
853	DO j = nys, nyn
854	local_pf(i,j,nzb+1) = ssws_av(j,i)
855	ENDDO
856	ENDDO
857	ENDIF
858	resorted = .TRUE.
859	two_d = .TRUE.
860	level_z(nzb+1) = zu(nzb+1)
861
862	CASE ( 't*_xy' ) ! 2d-array
863	IF ( av == 0 ) THEN
864	DO m = 1, surf_def_h(0)%ns
865	i = surf_def_h(0)%i(m)
866	j = surf_def_h(0)%j(m)
867	local_pf(i,j,nzb+1) = surf_def_h(0)%ts(m)
868	ENDDO
869	DO m = 1, surf_lsm_h%ns
870	i = surf_lsm_h%i(m)
871	j = surf_lsm_h%j(m)
872	local_pf(i,j,nzb+1) = surf_lsm_h%ts(m)
873	ENDDO
874	DO m = 1, surf_usm_h%ns
875	i = surf_usm_h%i(m)
876	j = surf_usm_h%j(m)
877	local_pf(i,j,nzb+1) = surf_usm_h%ts(m)
878	ENDDO
879	ELSE
880	IF ( .NOT. ALLOCATED( ts_av ) ) THEN
881	ALLOCATE( ts_av(nysg:nyng,nxlg:nxrg) )
882	ts_av = REAL( fill_value, KIND = wp )
883	ENDIF
884	DO i = nxl, nxr
885	DO j = nys, nyn
886	local_pf(i,j,nzb+1) = ts_av(j,i)
887	ENDDO
888	ENDDO
889	ENDIF
890	resorted = .TRUE.
891	two_d = .TRUE.
892	level_z(nzb+1) = zu(nzb+1)
893
894	CASE ( 'tsurf*_xy' ) ! 2d-array
895	IF ( av == 0 ) THEN
896	DO m = 1, surf_def_h(0)%ns
897	i = surf_def_h(0)%i(m)
898	j = surf_def_h(0)%j(m)
899	local_pf(i,j,nzb+1) = surf_def_h(0)%pt_surface(m)
900	ENDDO
901
902	DO m = 1, surf_lsm_h%ns
903	i = surf_lsm_h%i(m)
904	j = surf_lsm_h%j(m)
905	local_pf(i,j,nzb+1) = surf_lsm_h%pt_surface(m)
906	ENDDO
907
908	DO m = 1, surf_usm_h%ns
909	i = surf_usm_h%i(m)
910	j = surf_usm_h%j(m)
911	local_pf(i,j,nzb+1) = surf_usm_h%pt_surface(m)
912	ENDDO
913
914	ELSE
915	IF ( .NOT. ALLOCATED( tsurf_av ) ) THEN
916	ALLOCATE( tsurf_av(nysg:nyng,nxlg:nxrg) )
917	tsurf_av = REAL( fill_value, KIND = wp )
918	ENDIF
919	DO i = nxl, nxr
920	DO j = nys, nyn
921	local_pf(i,j,nzb+1) = tsurf_av(j,i)
922	ENDDO
923	ENDDO
924	ENDIF
925	resorted = .TRUE.
926	two_d = .TRUE.
927	level_z(nzb+1) = zu(nzb+1)
928
929	CASE ( 'u_xy', 'u_xz', 'u_yz' )
930	flag_nr = 1
931	IF ( av == 0 ) THEN
932	to_be_resorted => u
933	ELSE
934	IF ( .NOT. ALLOCATED( u_av ) ) THEN
935	ALLOCATE( u_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
936	u_av = REAL( fill_value, KIND = wp )
937	ENDIF
938	to_be_resorted => u_av
939	ENDIF
940	IF ( mode == 'xy' ) level_z = zu
941	!
942	!-- Substitute the values generated by "mirror" boundary condition
943	!-- at the bottom boundary by the real surface values.
944	IF ( do2d(av,ivar) == 'u_xz' .OR. do2d(av,ivar) == 'u_yz' ) THEN
945	IF ( ibc_uv_b == 0 ) local_pf(:,:,nzb) = 0.0_wp
946	ENDIF
947
948	CASE ( 'us*_xy' ) ! 2d-array
949	IF ( av == 0 ) THEN
950	DO m = 1, surf_def_h(0)%ns
951	i = surf_def_h(0)%i(m)
952	j = surf_def_h(0)%j(m)
953	local_pf(i,j,nzb+1) = surf_def_h(0)%us(m)
954	ENDDO
955	DO m = 1, surf_lsm_h%ns
956	i = surf_lsm_h%i(m)
957	j = surf_lsm_h%j(m)
958	local_pf(i,j,nzb+1) = surf_lsm_h%us(m)
959	ENDDO
960	DO m = 1, surf_usm_h%ns
961	i = surf_usm_h%i(m)
962	j = surf_usm_h%j(m)
963	local_pf(i,j,nzb+1) = surf_usm_h%us(m)
964	ENDDO
965	ELSE
966	IF ( .NOT. ALLOCATED( us_av ) ) THEN
967	ALLOCATE( us_av(nysg:nyng,nxlg:nxrg) )
968	us_av = REAL( fill_value, KIND = wp )
969	ENDIF
970	DO i = nxl, nxr
971	DO j = nys, nyn
972	local_pf(i,j,nzb+1) = us_av(j,i)
973	ENDDO
974	ENDDO
975	ENDIF
976	resorted = .TRUE.
977	two_d = .TRUE.
978	level_z(nzb+1) = zu(nzb+1)
979
980	CASE ( 'v_xy', 'v_xz', 'v_yz' )
981	flag_nr = 2
982	IF ( av == 0 ) THEN
983	to_be_resorted => v
984	ELSE
985	IF ( .NOT. ALLOCATED( v_av ) ) THEN
986	ALLOCATE( v_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
987	v_av = REAL( fill_value, KIND = wp )
988	ENDIF
989	to_be_resorted => v_av
990	ENDIF
991	IF ( mode == 'xy' ) level_z = zu
992	!
993	!-- Substitute the values generated by "mirror" boundary condition
994	!-- at the bottom boundary by the real surface values.
995	IF ( do2d(av,ivar) == 'v_xz' .OR. do2d(av,ivar) == 'v_yz' ) THEN
996	IF ( ibc_uv_b == 0 ) local_pf(:,:,nzb) = 0.0_wp
997	ENDIF
998
999	CASE ( 'thetav_xy', 'thetav_xz', 'thetav_yz' )
1000	IF ( av == 0 ) THEN
1001	to_be_resorted => vpt
1002	ELSE
1003	IF ( .NOT. ALLOCATED( vpt_av ) ) THEN
1004	ALLOCATE( vpt_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
1005	vpt_av = REAL( fill_value, KIND = wp )
1006	ENDIF
1007	to_be_resorted => vpt_av
1008	ENDIF
1009	IF ( mode == 'xy' ) level_z = zu
1010
1011	CASE ( 'w_xy', 'w_xz', 'w_yz' )
1012	flag_nr = 3
1013	IF ( av == 0 ) THEN
1014	to_be_resorted => w
1015	ELSE
1016	IF ( .NOT. ALLOCATED( w_av ) ) THEN
1017	ALLOCATE( w_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
1018	w_av = REAL( fill_value, KIND = wp )
1019	ENDIF
1020	to_be_resorted => w_av
1021	ENDIF
1022	IF ( mode == 'xy' ) level_z = zw
1023
1024	CASE ( 'z0*_xy' ) ! 2d-array
1025	IF ( av == 0 ) THEN
1026	DO m = 1, surf_def_h(0)%ns
1027	i = surf_def_h(0)%i(m)
1028	j = surf_def_h(0)%j(m)
1029	local_pf(i,j,nzb+1) = surf_def_h(0)%z0(m)
1030	ENDDO
1031	DO m = 1, surf_lsm_h%ns
1032	i = surf_lsm_h%i(m)
1033	j = surf_lsm_h%j(m)
1034	local_pf(i,j,nzb+1) = surf_lsm_h%z0(m)
1035	ENDDO
1036	DO m = 1, surf_usm_h%ns
1037	i = surf_usm_h%i(m)
1038	j = surf_usm_h%j(m)
1039	local_pf(i,j,nzb+1) = surf_usm_h%z0(m)
1040	ENDDO
1041	ELSE
1042	IF ( .NOT. ALLOCATED( z0_av ) ) THEN
1043	ALLOCATE( z0_av(nysg:nyng,nxlg:nxrg) )
1044	z0_av = REAL( fill_value, KIND = wp )
1045	ENDIF
1046	DO i = nxl, nxr
1047	DO j = nys, nyn
1048	local_pf(i,j,nzb+1) = z0_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 ( 'z0h*_xy' ) ! 2d-array
1057	IF ( av == 0 ) THEN
1058	DO m = 1, surf_def_h(0)%ns
1059	i = surf_def_h(0)%i(m)
1060	j = surf_def_h(0)%j(m)
1061	local_pf(i,j,nzb+1) = surf_def_h(0)%z0h(m)
1062	ENDDO
1063	DO m = 1, surf_lsm_h%ns
1064	i = surf_lsm_h%i(m)
1065	j = surf_lsm_h%j(m)
1066	local_pf(i,j,nzb+1) = surf_lsm_h%z0h(m)
1067	ENDDO
1068	DO m = 1, surf_usm_h%ns
1069	i = surf_usm_h%i(m)
1070	j = surf_usm_h%j(m)
1071	local_pf(i,j,nzb+1) = surf_usm_h%z0h(m)
1072	ENDDO
1073	ELSE
1074	IF ( .NOT. ALLOCATED( z0h_av ) ) THEN
1075	ALLOCATE( z0h_av(nysg:nyng,nxlg:nxrg) )
1076	z0h_av = REAL( fill_value, KIND = wp )
1077	ENDIF
1078	DO i = nxl, nxr
1079	DO j = nys, nyn
1080	local_pf(i,j,nzb+1) = z0h_av(j,i)
1081	ENDDO
1082	ENDDO
1083	ENDIF
1084	resorted = .TRUE.
1085	two_d = .TRUE.
1086	level_z(nzb+1) = zu(nzb+1)
1087
1088	CASE ( 'z0q*_xy' ) ! 2d-array
1089	IF ( av == 0 ) THEN
1090	DO m = 1, surf_def_h(0)%ns
1091	i = surf_def_h(0)%i(m)
1092	j = surf_def_h(0)%j(m)
1093	local_pf(i,j,nzb+1) = surf_def_h(0)%z0q(m)
1094	ENDDO
1095	DO m = 1, surf_lsm_h%ns
1096	i = surf_lsm_h%i(m)
1097	j = surf_lsm_h%j(m)
1098	local_pf(i,j,nzb+1) = surf_lsm_h%z0q(m)
1099	ENDDO
1100	DO m = 1, surf_usm_h%ns
1101	i = surf_usm_h%i(m)
1102	j = surf_usm_h%j(m)
1103	local_pf(i,j,nzb+1) = surf_usm_h%z0q(m)
1104	ENDDO
1105	ELSE
1106	IF ( .NOT. ALLOCATED( z0q_av ) ) THEN
1107	ALLOCATE( z0q_av(nysg:nyng,nxlg:nxrg) )
1108	z0q_av = REAL( fill_value, KIND = wp )
1109	ENDIF
1110	DO i = nxl, nxr
1111	DO j = nys, nyn
1112	local_pf(i,j,nzb+1) = z0q_av(j,i)
1113	ENDDO
1114	ENDDO
1115	ENDIF
1116	resorted = .TRUE.
1117	two_d = .TRUE.
1118	level_z(nzb+1) = zu(nzb+1)
1119
1120	CASE DEFAULT
1121
1122	!
1123	!-- Quantities of other modules
1124	IF ( .NOT. found ) THEN
1125	CALL module_interface_data_output_2d( &
1126	av, do2d(av,ivar), found, grid, mode, &
1127	local_pf, two_d, nzb_do, nzt_do, &
1128	fill_value &
1129	)
1130	ENDIF
1131
1132	resorted = .TRUE.
1133
1134	IF ( grid == 'zu' ) THEN
1135	IF ( mode == 'xy' ) level_z = zu
1136	ELSEIF ( grid == 'zw' ) THEN
1137	IF ( mode == 'xy' ) level_z = zw
1138	ELSEIF ( grid == 'zu1' ) THEN
1139	IF ( mode == 'xy' ) level_z(nzb+1) = zu(nzb+1)
1140	ELSEIF ( grid == 'zs' ) THEN
1141	IF ( mode == 'xy' ) level_z = zs
1142	ENDIF
1143
1144	IF ( .NOT. found ) THEN
1145	message_string = 'no output provided for: ' // &
1146	TRIM( do2d(av,ivar) )
1147	CALL message( 'data_output_2d', 'PA0181', 0, 0, 0, 6, 0 )
1148	ENDIF
1149
1150	END SELECT
1151
1152	!
1153	!-- Resort the array to be output, if not done above. Flag topography
1154	!-- grid points with fill values, using the corresponding maksing flag.
1155	IF ( .NOT. resorted ) THEN
1156	DO i = nxl, nxr
1157	DO j = nys, nyn
1158	DO k = nzb_do, nzt_do
1159	local_pf(i,j,k) = MERGE( to_be_resorted(k,j,i), &
1160	REAL( fill_value, KIND = wp ), &
1161	BTEST( wall_flags_total_0(k,j,i), &
1162	flag_nr ) )
1163	ENDDO
1164	ENDDO
1165	ENDDO
1166	ENDIF
1167
1168	!
1169	!-- Output of the individual cross-sections, depending on the cross-
1170	!-- section mode chosen.
1171	is = 1
1172	loop1: DO WHILE ( section(is,s_ind) /= -9999 .OR. two_d )
1173
1174	SELECT CASE ( mode )
1175
1176	CASE ( 'xy' )
1177	!
1178	!-- Determine the cross section index
1179	IF ( two_d ) THEN
1180	layer_xy = nzb+1
1181	ELSE
1182	layer_xy = section(is,s_ind)
1183	ENDIF
1184
1185	!
1186	!-- Exit the loop for layers beyond the data output domain
1187	!-- (used for soil model)
1188	IF ( layer_xy > nzt_do ) THEN
1189	EXIT loop1
1190	ENDIF
1191
1192	!
1193	!-- Update the netCDF xy cross section time axis.
1194	!-- In case of parallel output, this is only done by PE0
1195	!-- to increase the performance.
1196	IF ( time_since_reference_point /= do2d_xy_last_time(av) ) THEN
1197	do2d_xy_time_count(av) = do2d_xy_time_count(av) + 1
1198	do2d_xy_last_time(av) = time_since_reference_point
1199	IF ( myid == 0 ) THEN
1200	IF ( .NOT. data_output_2d_on_each_pe &
1201	.OR. netcdf_data_format > 4 ) &
1202	THEN
1203	#if defined( __netcdf )
1204	nc_stat = NF90_PUT_VAR( id_set_xy(av), &
1205	id_var_time_xy(av), &
1206	(/ time_since_reference_point /), &
1207	start = (/ do2d_xy_time_count(av) /), &
1208	count = (/ 1 /) )
1209	CALL netcdf_handle_error( 'data_output_2d', 53 )
1210	#endif
1211	ENDIF
1212	ENDIF
1213	ENDIF
1214	!
1215	!-- If required, carry out averaging along z
1216	IF ( section(is,s_ind) == -1 .AND. .NOT. two_d ) THEN
1217
1218	local_2d = 0.0_wp
1219	!
1220	!-- Carry out the averaging (all data are on the PE)
1221	DO k = nzb_do, nzt_do
1222	DO j = nys, nyn
1223	DO i = nxl, nxr
1224	local_2d(i,j) = local_2d(i,j) + local_pf(i,j,k)
1225	ENDDO
1226	ENDDO
1227	ENDDO
1228
1229	local_2d = local_2d / ( nzt_do - nzb_do + 1.0_wp)
1230
1231	ELSE
1232	!
1233	!-- Just store the respective section on the local array
1234	local_2d = local_pf(:,:,layer_xy)
1235
1236	ENDIF
1237
1238	#if defined( __parallel )
1239	IF ( netcdf_data_format > 4 ) THEN
1240	!
1241	!-- Parallel output in netCDF4/HDF5 format.
1242	IF ( two_d ) THEN
1243	iis = 1
1244	ELSE
1245	iis = is
1246	ENDIF
1247
1248	#if defined( __netcdf )
1249	!
1250	!-- For parallel output, all cross sections are first stored
1251	!-- here on a local array and will be written to the output
1252	!-- file afterwards to increase the performance.
1253	DO i = nxl, nxr
1254	DO j = nys, nyn
1255	local_2d_sections(i,j,iis) = local_2d(i,j)
1256	ENDDO
1257	ENDDO
1258	#endif
1259	ELSE
1260
1261	IF ( data_output_2d_on_each_pe ) THEN
1262	!
1263	!-- Output of partial arrays on each PE
1264	#if defined( __netcdf )
1265	IF ( myid == 0 ) THEN
1266	WRITE ( 21 ) time_since_reference_point, &
1267	do2d_xy_time_count(av), av
1268	ENDIF
1269	#endif
1270	DO i = 0, io_blocks-1
1271	IF ( i == io_group ) THEN
1272	WRITE ( 21 ) nxl, nxr, nys, nyn, nys, nyn
1273	WRITE ( 21 ) local_2d
1274	ENDIF
1275	#if defined( __parallel )
1276	CALL MPI_BARRIER( comm2d, ierr )
1277	#endif
1278	ENDDO
1279
1280	ELSE
1281	!
1282	!-- PE0 receives partial arrays from all processors and
1283	!-- then outputs them. Here a barrier has to be set,
1284	!-- because otherwise "-MPI- FATAL: Remote protocol queue
1285	!-- full" may occur.
1286	CALL MPI_BARRIER( comm2d, ierr )
1287
1288	ngp = ( nxr-nxl+1 ) * ( nyn-nys+1 )
1289	IF ( myid == 0 ) THEN
1290	!
1291	!-- Local array can be relocated directly.
1292	total_2d(nxl:nxr,nys:nyn) = local_2d
1293	!
1294	!-- Receive data from all other PEs.
1295	DO n = 1, numprocs-1
1296	!
1297	!-- Receive index limits first, then array.
1298	!-- Index limits are received in arbitrary order from
1299	!-- the PEs.
1300	CALL MPI_RECV( ind(1), 4, MPI_INTEGER, &
1301	MPI_ANY_SOURCE, 0, comm2d, &
1302	status, ierr )
1303	sender = status(MPI_SOURCE)
1304	DEALLOCATE( local_2d )
1305	ALLOCATE( local_2d(ind(1):ind(2),ind(3):ind(4)) )
1306	CALL MPI_RECV( local_2d(ind(1),ind(3)), ngp, &
1307	MPI_REAL, sender, 1, comm2d, &
1308	status, ierr )
1309	total_2d(ind(1):ind(2),ind(3):ind(4)) = local_2d
1310	ENDDO
1311	!
1312	!-- Relocate the local array for the next loop increment
1313	DEALLOCATE( local_2d )
1314	ALLOCATE( local_2d(nxl:nxr,nys:nyn) )
1315
1316	#if defined( __netcdf )
1317	IF ( two_d ) THEN
1318	nc_stat = NF90_PUT_VAR( id_set_xy(av), &
1319	id_var_do2d(av,ivar), &
1320	total_2d(0:nx,0:ny), &
1321	start = (/ 1, 1, 1, do2d_xy_time_count(av) /), &
1322	count = (/ nx+1, ny+1, 1, 1 /) )
1323	ELSE
1324	nc_stat = NF90_PUT_VAR( id_set_xy(av), &
1325	id_var_do2d(av,ivar), &
1326	total_2d(0:nx,0:ny), &
1327	start = (/ 1, 1, is, do2d_xy_time_count(av) /), &
1328	count = (/ nx+1, ny+1, 1, 1 /) )
1329	ENDIF
1330	CALL netcdf_handle_error( 'data_output_2d', 54 )
1331	#endif
1332
1333	ELSE
1334	!
1335	!-- First send the local index limits to PE0
1336	ind(1) = nxl; ind(2) = nxr
1337	ind(3) = nys; ind(4) = nyn
1338	CALL MPI_SEND( ind(1), 4, MPI_INTEGER, 0, 0, &
1339	comm2d, ierr )
1340	!
1341	!-- Send data to PE0
1342	CALL MPI_SEND( local_2d(nxl,nys), ngp, &
1343	MPI_REAL, 0, 1, comm2d, ierr )
1344	ENDIF
1345	!
1346	!-- A barrier has to be set, because otherwise some PEs may
1347	!-- proceed too fast so that PE0 may receive wrong data on
1348	!-- tag 0
1349	CALL MPI_BARRIER( comm2d, ierr )
1350	ENDIF
1351
1352	ENDIF
1353	#else
1354	#if defined( __netcdf )
1355	IF ( two_d ) THEN
1356	nc_stat = NF90_PUT_VAR( id_set_xy(av), &
1357	id_var_do2d(av,ivar), &
1358	local_2d(nxl:nxr,nys:nyn), &
1359	start = (/ 1, 1, 1, do2d_xy_time_count(av) /), &
1360	count = (/ nx+1, ny+1, 1, 1 /) )
1361	ELSE
1362	nc_stat = NF90_PUT_VAR( id_set_xy(av), &
1363	id_var_do2d(av,ivar), &
1364	local_2d(nxl:nxr,nys:nyn), &
1365	start = (/ 1, 1, is, do2d_xy_time_count(av) /), &
1366	count = (/ nx+1, ny+1, 1, 1 /) )
1367	ENDIF
1368	CALL netcdf_handle_error( 'data_output_2d', 447 )
1369	#endif
1370	#endif
1371
1372	!
1373	!-- For 2D-arrays (e.g. u*) only one cross-section is available.
1374	!-- Hence exit loop of output levels.
1375	IF ( two_d ) THEN
1376	IF ( netcdf_data_format < 5 ) two_d = .FALSE.
1377	EXIT loop1
1378	ENDIF
1379
1380	CASE ( 'xz' )
1381	!
1382	!-- Update the netCDF xz cross section time axis.
1383	!-- In case of parallel output, this is only done by PE0
1384	!-- to increase the performance.
1385	IF ( time_since_reference_point /= do2d_xz_last_time(av) ) THEN
1386	do2d_xz_time_count(av) = do2d_xz_time_count(av) + 1
1387	do2d_xz_last_time(av) = time_since_reference_point
1388	IF ( myid == 0 ) THEN
1389	IF ( .NOT. data_output_2d_on_each_pe &
1390	.OR. netcdf_data_format > 4 ) &
1391	THEN
1392	#if defined( __netcdf )
1393	nc_stat = NF90_PUT_VAR( id_set_xz(av), &
1394	id_var_time_xz(av), &
1395	(/ time_since_reference_point /), &
1396	start = (/ do2d_xz_time_count(av) /), &
1397	count = (/ 1 /) )
1398	CALL netcdf_handle_error( 'data_output_2d', 56 )
1399	#endif
1400	ENDIF
1401	ENDIF
1402	ENDIF
1403
1404	!
1405	!-- If required, carry out averaging along y
1406	IF ( section(is,s_ind) == -1 ) THEN
1407
1408	ALLOCATE( local_2d_l(nxl:nxr,nzb_do:nzt_do) )
1409	local_2d_l = 0.0_wp
1410	ngp = ( nxr-nxl + 1 ) * ( nzt_do-nzb_do + 1 )
1411	!
1412	!-- First local averaging on the PE
1413	DO k = nzb_do, nzt_do
1414	DO j = nys, nyn
1415	DO i = nxl, nxr
1416	local_2d_l(i,k) = local_2d_l(i,k) + &
1417	local_pf(i,j,k)
1418	ENDDO
1419	ENDDO
1420	ENDDO
1421	#if defined( __parallel )
1422	!
1423	!-- Now do the averaging over all PEs along y
1424	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
1425	CALL MPI_ALLREDUCE( local_2d_l(nxl,nzb_do), &
1426	local_2d(nxl,nzb_do), ngp, MPI_REAL, &
1427	MPI_SUM, comm1dy, ierr )
1428	#else
1429	local_2d = local_2d_l
1430	#endif
1431	local_2d = local_2d / ( ny + 1.0_wp )
1432
1433	DEALLOCATE( local_2d_l )
1434
1435	ELSE
1436	!
1437	!-- Just store the respective section on the local array
1438	!-- (but only if it is available on this PE!)
1439	IF ( section(is,s_ind) >= nys .AND. section(is,s_ind) <= nyn ) &
1440	THEN
1441	local_2d = local_pf(:,section(is,s_ind),nzb_do:nzt_do)
1442	ENDIF
1443
1444	ENDIF
1445
1446	#if defined( __parallel )
1447	IF ( netcdf_data_format > 4 ) THEN
1448	!
1449	!-- Output in netCDF4/HDF5 format.
1450	!-- Output only on those PEs where the respective cross
1451	!-- sections reside. Cross sections averaged along y are
1452	!-- output on the respective first PE along y (myidy=0).
1453	IF ( ( section(is,s_ind) >= nys .AND. &
1454	section(is,s_ind) <= nyn ) .OR. &
1455	( section(is,s_ind) == -1 .AND. myidy == 0 ) ) THEN
1456	#if defined( __netcdf )
1457	!
1458	!-- For parallel output, all cross sections are first
1459	!-- stored here on a local array and will be written to the
1460	!-- output file afterwards to increase the performance.
1461	DO i = nxl, nxr
1462	DO k = nzb_do, nzt_do
1463	local_2d_sections_l(i,is,k) = local_2d(i,k)
1464	ENDDO
1465	ENDDO
1466	#endif
1467	ENDIF
1468
1469	ELSE
1470
1471	IF ( data_output_2d_on_each_pe ) THEN
1472	!
1473	!-- Output of partial arrays on each PE. If the cross
1474	!-- section does not reside on the PE, output special
1475	!-- index values.
1476	#if defined( __netcdf )
1477	IF ( myid == 0 ) THEN
1478	WRITE ( 22 ) time_since_reference_point, &
1479	do2d_xz_time_count(av), av
1480	ENDIF
1481	#endif
1482	DO i = 0, io_blocks-1
1483	IF ( i == io_group ) THEN
1484	IF ( ( section(is,s_ind) >= nys .AND. &
1485	section(is,s_ind) <= nyn ) .OR. &
1486	( section(is,s_ind) == -1 .AND. &
1487	nys-1 == -1 ) ) &
1488	THEN
1489	WRITE (22) nxl, nxr, nzb_do, nzt_do, nzb, nzt+1
1490	WRITE (22) local_2d
1491	ELSE
1492	WRITE (22) -1, -1, -1, -1, -1, -1
1493	ENDIF
1494	ENDIF
1495	#if defined( __parallel )
1496	CALL MPI_BARRIER( comm2d, ierr )
1497	#endif
1498	ENDDO
1499
1500	ELSE
1501	!
1502	!-- PE0 receives partial arrays from all processors of the
1503	!-- respective cross section and outputs them. Here a
1504	!-- barrier has to be set, because otherwise
1505	!-- "-MPI- FATAL: Remote protocol queue full" may occur.
1506	CALL MPI_BARRIER( comm2d, ierr )
1507
1508	ngp = ( nxr-nxl + 1 ) * ( nzt_do-nzb_do + 1 )
1509	IF ( myid == 0 ) THEN
1510	!
1511	!-- Local array can be relocated directly.
1512	IF ( ( section(is,s_ind) >= nys .AND. &
1513	section(is,s_ind) <= nyn ) .OR. &
1514	( section(is,s_ind) == -1 .AND. &
1515	nys-1 == -1 ) ) THEN
1516	total_2d(nxl:nxr,nzb_do:nzt_do) = local_2d
1517	ENDIF
1518	!
1519	!-- Receive data from all other PEs.
1520	DO n = 1, numprocs-1
1521	!
1522	!-- Receive index limits first, then array.
1523	!-- Index limits are received in arbitrary order from
1524	!-- the PEs.
1525	CALL MPI_RECV( ind(1), 4, MPI_INTEGER, &
1526	MPI_ANY_SOURCE, 0, comm2d, &
1527	status, ierr )
1528	!
1529	!-- Not all PEs have data for XZ-cross-section.
1530	IF ( ind(1) /= -9999 ) THEN
1531	sender = status(MPI_SOURCE)
1532	DEALLOCATE( local_2d )
1533	ALLOCATE( local_2d(ind(1):ind(2), &
1534	ind(3):ind(4)) )
1535	CALL MPI_RECV( local_2d(ind(1),ind(3)), ngp, &
1536	MPI_REAL, sender, 1, comm2d, &
1537	status, ierr )
1538	total_2d(ind(1):ind(2),ind(3):ind(4)) = &
1539	local_2d
1540	ENDIF
1541	ENDDO
1542	!
1543	!-- Relocate the local array for the next loop increment
1544	DEALLOCATE( local_2d )
1545	ALLOCATE( local_2d(nxl:nxr,nzb_do:nzt_do) )
1546
1547	#if defined( __netcdf )
1548	nc_stat = NF90_PUT_VAR( id_set_xz(av), &
1549	id_var_do2d(av,ivar), &
1550	total_2d(0:nx,nzb_do:nzt_do), &
1551	start = (/ 1, is, 1, do2d_xz_time_count(av) /), &
1552	count = (/ nx+1, 1, nzt_do-nzb_do+1, 1 /) )
1553	CALL netcdf_handle_error( 'data_output_2d', 58 )
1554	#endif
1555
1556	ELSE
1557	!
1558	!-- If the cross section resides on the PE, send the
1559	!-- local index limits, otherwise send -9999 to PE0.
1560	IF ( ( section(is,s_ind) >= nys .AND. &
1561	section(is,s_ind) <= nyn ) .OR. &
1562	( section(is,s_ind) == -1 .AND. nys-1 == -1 ) ) &
1563	THEN
1564	ind(1) = nxl; ind(2) = nxr
1565	ind(3) = nzb_do; ind(4) = nzt_do
1566	ELSE
1567	ind(1) = -9999; ind(2) = -9999
1568	ind(3) = -9999; ind(4) = -9999
1569	ENDIF
1570	CALL MPI_SEND( ind(1), 4, MPI_INTEGER, 0, 0, &
1571	comm2d, ierr )
1572	!
1573	!-- If applicable, send data to PE0.
1574	IF ( ind(1) /= -9999 ) THEN
1575	CALL MPI_SEND( local_2d(nxl,nzb_do), ngp, &
1576	MPI_REAL, 0, 1, comm2d, ierr )
1577	ENDIF
1578	ENDIF
1579	!
1580	!-- A barrier has to be set, because otherwise some PEs may
1581	!-- proceed too fast so that PE0 may receive wrong data on
1582	!-- tag 0
1583	CALL MPI_BARRIER( comm2d, ierr )
1584	ENDIF
1585
1586	ENDIF
1587	#else
1588	#if defined( __netcdf )
1589	nc_stat = NF90_PUT_VAR( id_set_xz(av), &
1590	id_var_do2d(av,ivar), &
1591	local_2d(nxl:nxr,nzb_do:nzt_do), &
1592	start = (/ 1, is, 1, do2d_xz_time_count(av) /), &
1593	count = (/ nx+1, 1, nzt_do-nzb_do+1, 1 /) )
1594	CALL netcdf_handle_error( 'data_output_2d', 451 )
1595	#endif
1596	#endif
1597
1598	CASE ( 'yz' )
1599	!
1600	!-- Update the netCDF yz cross section time axis.
1601	!-- In case of parallel output, this is only done by PE0
1602	!-- to increase the performance.
1603	IF ( time_since_reference_point /= do2d_yz_last_time(av) ) THEN
1604	do2d_yz_time_count(av) = do2d_yz_time_count(av) + 1
1605	do2d_yz_last_time(av) = time_since_reference_point
1606	IF ( myid == 0 ) THEN
1607	IF ( .NOT. data_output_2d_on_each_pe &
1608	.OR. netcdf_data_format > 4 ) &
1609	THEN
1610	#if defined( __netcdf )
1611	nc_stat = NF90_PUT_VAR( id_set_yz(av), &
1612	id_var_time_yz(av), &
1613	(/ time_since_reference_point /), &
1614	start = (/ do2d_yz_time_count(av) /), &
1615	count = (/ 1 /) )
1616	CALL netcdf_handle_error( 'data_output_2d', 59 )
1617	#endif
1618	ENDIF
1619	ENDIF
1620	ENDIF
1621
1622	!
1623	!-- If required, carry out averaging along x
1624	IF ( section(is,s_ind) == -1 ) THEN
1625
1626	ALLOCATE( local_2d_l(nys:nyn,nzb_do:nzt_do) )
1627	local_2d_l = 0.0_wp
1628	ngp = ( nyn-nys+1 ) * ( nzt_do-nzb_do+1 )
1629	!
1630	!-- First local averaging on the PE
1631	DO k = nzb_do, nzt_do
1632	DO j = nys, nyn
1633	DO i = nxl, nxr
1634	local_2d_l(j,k) = local_2d_l(j,k) + &
1635	local_pf(i,j,k)
1636	ENDDO
1637	ENDDO
1638	ENDDO
1639	#if defined( __parallel )
1640	!
1641	!-- Now do the averaging over all PEs along x
1642	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
1643	CALL MPI_ALLREDUCE( local_2d_l(nys,nzb_do), &
1644	local_2d(nys,nzb_do), ngp, MPI_REAL, &
1645	MPI_SUM, comm1dx, ierr )
1646	#else
1647	local_2d = local_2d_l
1648	#endif
1649	local_2d = local_2d / ( nx + 1.0_wp )
1650
1651	DEALLOCATE( local_2d_l )
1652
1653	ELSE
1654	!
1655	!-- Just store the respective section on the local array
1656	!-- (but only if it is available on this PE!)
1657	IF ( section(is,s_ind) >= nxl .AND. section(is,s_ind) <= nxr ) &
1658	THEN
1659	local_2d = local_pf(section(is,s_ind),:,nzb_do:nzt_do)
1660	ENDIF
1661
1662	ENDIF
1663
1664	#if defined( __parallel )
1665	IF ( netcdf_data_format > 4 ) THEN
1666	!
1667	!-- Output in netCDF4/HDF5 format.
1668	!-- Output only on those PEs where the respective cross
1669	!-- sections reside. Cross sections averaged along x are
1670	!-- output on the respective first PE along x (myidx=0).
1671	IF ( ( section(is,s_ind) >= nxl .AND. &
1672	section(is,s_ind) <= nxr ) .OR. &
1673	( section(is,s_ind) == -1 .AND. myidx == 0 ) ) THEN
1674	#if defined( __netcdf )
1675	!
1676	!-- For parallel output, all cross sections are first
1677	!-- stored here on a local array and will be written to the
1678	!-- output file afterwards to increase the performance.
1679	DO j = nys, nyn
1680	DO k = nzb_do, nzt_do
1681	local_2d_sections_l(is,j,k) = local_2d(j,k)
1682	ENDDO
1683	ENDDO
1684	#endif
1685	ENDIF
1686
1687	ELSE
1688
1689	IF ( data_output_2d_on_each_pe ) THEN
1690	!
1691	!-- Output of partial arrays on each PE. If the cross
1692	!-- section does not reside on the PE, output special
1693	!-- index values.
1694	#if defined( __netcdf )
1695	IF ( myid == 0 ) THEN
1696	WRITE ( 23 ) time_since_reference_point, &
1697	do2d_yz_time_count(av), av
1698	ENDIF
1699	#endif
1700	DO i = 0, io_blocks-1
1701	IF ( i == io_group ) THEN
1702	IF ( ( section(is,s_ind) >= nxl .AND. &
1703	section(is,s_ind) <= nxr ) .OR. &
1704	( section(is,s_ind) == -1 .AND. &
1705	nxl-1 == -1 ) ) &
1706	THEN
1707	WRITE (23) nys, nyn, nzb_do, nzt_do, nzb, nzt+1
1708	WRITE (23) local_2d
1709	ELSE
1710	WRITE (23) -1, -1, -1, -1, -1, -1
1711	ENDIF
1712	ENDIF
1713	#if defined( __parallel )
1714	CALL MPI_BARRIER( comm2d, ierr )
1715	#endif
1716	ENDDO
1717
1718	ELSE
1719	!
1720	!-- PE0 receives partial arrays from all processors of the
1721	!-- respective cross section and outputs them. Here a
1722	!-- barrier has to be set, because otherwise
1723	!-- "-MPI- FATAL: Remote protocol queue full" may occur.
1724	CALL MPI_BARRIER( comm2d, ierr )
1725
1726	ngp = ( nyn-nys+1 ) * ( nzt_do-nzb_do+1 )
1727	IF ( myid == 0 ) THEN
1728	!
1729	!-- Local array can be relocated directly.
1730	IF ( ( section(is,s_ind) >= nxl .AND. &
1731	section(is,s_ind) <= nxr ) .OR. &
1732	( section(is,s_ind) == -1 .AND. nxl-1 == -1 ) ) &
1733	THEN
1734	total_2d(nys:nyn,nzb_do:nzt_do) = local_2d
1735	ENDIF
1736	!
1737	!-- Receive data from all other PEs.
1738	DO n = 1, numprocs-1
1739	!
1740	!-- Receive index limits first, then array.
1741	!-- Index limits are received in arbitrary order from
1742	!-- the PEs.
1743	CALL MPI_RECV( ind(1), 4, MPI_INTEGER, &
1744	MPI_ANY_SOURCE, 0, comm2d, &
1745	status, ierr )
1746	!
1747	!-- Not all PEs have data for YZ-cross-section.
1748	IF ( ind(1) /= -9999 ) THEN
1749	sender = status(MPI_SOURCE)
1750	DEALLOCATE( local_2d )
1751	ALLOCATE( local_2d(ind(1):ind(2), &
1752	ind(3):ind(4)) )
1753	CALL MPI_RECV( local_2d(ind(1),ind(3)), ngp, &
1754	MPI_REAL, sender, 1, comm2d, &
1755	status, ierr )
1756	total_2d(ind(1):ind(2),ind(3):ind(4)) = &
1757	local_2d
1758	ENDIF
1759	ENDDO
1760	!
1761	!-- Relocate the local array for the next loop increment
1762	DEALLOCATE( local_2d )
1763	ALLOCATE( local_2d(nys:nyn,nzb_do:nzt_do) )
1764
1765	#if defined( __netcdf )
1766	nc_stat = NF90_PUT_VAR( id_set_yz(av), &
1767	id_var_do2d(av,ivar), &
1768	total_2d(0:ny,nzb_do:nzt_do), &
1769	start = (/ is, 1, 1, do2d_yz_time_count(av) /), &
1770	count = (/ 1, ny+1, nzt_do-nzb_do+1, 1 /) )
1771	CALL netcdf_handle_error( 'data_output_2d', 61 )
1772	#endif
1773
1774	ELSE
1775	!
1776	!-- If the cross section resides on the PE, send the
1777	!-- local index limits, otherwise send -9999 to PE0.
1778	IF ( ( section(is,s_ind) >= nxl .AND. &
1779	section(is,s_ind) <= nxr ) .OR. &
1780	( section(is,s_ind) == -1 .AND. nxl-1 == -1 ) ) &
1781	THEN
1782	ind(1) = nys; ind(2) = nyn
1783	ind(3) = nzb_do; ind(4) = nzt_do
1784	ELSE
1785	ind(1) = -9999; ind(2) = -9999
1786	ind(3) = -9999; ind(4) = -9999
1787	ENDIF
1788	CALL MPI_SEND( ind(1), 4, MPI_INTEGER, 0, 0, &
1789	comm2d, ierr )
1790	!
1791	!-- If applicable, send data to PE0.
1792	IF ( ind(1) /= -9999 ) THEN
1793	CALL MPI_SEND( local_2d(nys,nzb_do), ngp, &
1794	MPI_REAL, 0, 1, comm2d, ierr )
1795	ENDIF
1796	ENDIF
1797	!
1798	!-- A barrier has to be set, because otherwise some PEs may
1799	!-- proceed too fast so that PE0 may receive wrong data on
1800	!-- tag 0
1801	CALL MPI_BARRIER( comm2d, ierr )
1802	ENDIF
1803
1804	ENDIF
1805	#else
1806	#if defined( __netcdf )
1807	nc_stat = NF90_PUT_VAR( id_set_yz(av), &
1808	id_var_do2d(av,ivar), &
1809	local_2d(nys:nyn,nzb_do:nzt_do), &
1810	start = (/ is, 1, 1, do2d_xz_time_count(av) /), &
1811	count = (/ 1, ny+1, nzt_do-nzb_do+1, 1 /) )
1812	CALL netcdf_handle_error( 'data_output_2d', 452 )
1813	#endif
1814	#endif
1815
1816	END SELECT
1817
1818	is = is + 1
1819	ENDDO loop1
1820
1821	!
1822	!-- For parallel output, all data were collected before on a local array
1823	!-- and are written now to the netcdf file. This must be done to increase
1824	!-- the performance of the parallel output.
1825	#if defined( __netcdf )
1826	IF ( netcdf_data_format > 4 ) THEN
1827
1828	SELECT CASE ( mode )
1829
1830	CASE ( 'xy' )
1831	IF ( two_d ) THEN
1832	nis = 1
1833	two_d = .FALSE.
1834	ELSE
1835	nis = ns
1836	ENDIF
1837	!
1838	!-- Do not output redundant ghost point data except for the
1839	!-- boundaries of the total domain.
1840	! IF ( nxr == nx .AND. nyn /= ny ) THEN
1841	! nc_stat = NF90_PUT_VAR( id_set_xy(av), &
1842	! id_var_do2d(av,ivar), &
1843	! local_2d_sections(nxl:nxr+1, &
1844	! nys:nyn,1:nis), &
1845	! start = (/ nxl+1, nys+1, 1, &
1846	! do2d_xy_time_count(av) /), &
1847	! count = (/ nxr-nxl+2, &
1848	! nyn-nys+1, nis, 1 &
1849	! /) )
1850	! ELSEIF ( nxr /= nx .AND. nyn == ny ) THEN
1851	! nc_stat = NF90_PUT_VAR( id_set_xy(av), &
1852	! id_var_do2d(av,ivar), &
1853	! local_2d_sections(nxl:nxr, &
1854	! nys:nyn+1,1:nis), &
1855	! start = (/ nxl+1, nys+1, 1, &
1856	! do2d_xy_time_count(av) /), &
1857	! count = (/ nxr-nxl+1, &
1858	! nyn-nys+2, nis, 1 &
1859	! /) )
1860	! ELSEIF ( nxr == nx .AND. nyn == ny ) THEN
1861	! nc_stat = NF90_PUT_VAR( id_set_xy(av), &
1862	! id_var_do2d(av,ivar), &
1863	! local_2d_sections(nxl:nxr+1, &
1864	! nys:nyn+1,1:nis), &
1865	! start = (/ nxl+1, nys+1, 1, &
1866	! do2d_xy_time_count(av) /), &
1867	! count = (/ nxr-nxl+2, &
1868	! nyn-nys+2, nis, 1 &
1869	! /) )
1870	! ELSE
1871	nc_stat = NF90_PUT_VAR( id_set_xy(av), &
1872	id_var_do2d(av,ivar), &
1873	local_2d_sections(nxl:nxr, &
1874	nys:nyn,1:nis), &
1875	start = (/ nxl+1, nys+1, 1, &
1876	do2d_xy_time_count(av) /), &
1877	count = (/ nxr-nxl+1, &
1878	nyn-nys+1, nis, 1 &
1879	/) )
1880	! ENDIF
1881
1882	CALL netcdf_handle_error( 'data_output_2d', 55 )
1883
1884	CASE ( 'xz' )
1885	!
1886	!-- First, all PEs get the information of all cross-sections.
1887	!-- Then the data are written to the output file by all PEs
1888	!-- while NF90_COLLECTIVE is set in subroutine
1889	!-- define_netcdf_header. Although redundant information are
1890	!-- written to the output file in that case, the performance
1891	!-- is significantly better compared to the case where only
1892	!-- the first row of PEs in x-direction (myidx = 0) is given
1893	!-- the output while NF90_INDEPENDENT is set.
1894	IF ( npey /= 1 ) THEN
1895
1896	#if defined( __parallel )
1897	!
1898	!-- Distribute data over all PEs along y
1899	ngp = ( nxr-nxl+1 ) * ( nzt_do-nzb_do+1 ) * ns
1900	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
1901	CALL MPI_ALLREDUCE( local_2d_sections_l(nxl,1,nzb_do), &
1902	local_2d_sections(nxl,1,nzb_do), &
1903	ngp, MPI_REAL, MPI_SUM, comm1dy, &
1904	ierr )
1905	#else
1906	local_2d_sections = local_2d_sections_l
1907	#endif
1908	ENDIF
1909	!
1910	!-- Do not output redundant ghost point data except for the
1911	!-- boundaries of the total domain.
1912	! IF ( nxr == nx ) THEN
1913	! nc_stat = NF90_PUT_VAR( id_set_xz(av), &
1914	! id_var_do2d(av,ivar), &
1915	! local_2d_sections(nxl:nxr+1,1:ns, &
1916	! nzb_do:nzt_do), &
1917	! start = (/ nxl+1, 1, 1, &
1918	! do2d_xz_time_count(av) /), &
1919	! count = (/ nxr-nxl+2, ns, nzt_do-nzb_do+1, &
1920	! 1 /) )
1921	! ELSE
1922	nc_stat = NF90_PUT_VAR( id_set_xz(av), &
1923	id_var_do2d(av,ivar), &
1924	local_2d_sections(nxl:nxr,1:ns, &
1925	nzb_do:nzt_do), &
1926	start = (/ nxl+1, 1, 1, &
1927	do2d_xz_time_count(av) /), &
1928	count = (/ nxr-nxl+1, ns, nzt_do-nzb_do+1, &
1929	1 /) )
1930	! ENDIF
1931
1932	CALL netcdf_handle_error( 'data_output_2d', 57 )
1933
1934	CASE ( 'yz' )
1935	!
1936	!-- First, all PEs get the information of all cross-sections.
1937	!-- Then the data are written to the output file by all PEs
1938	!-- while NF90_COLLECTIVE is set in subroutine
1939	!-- define_netcdf_header. Although redundant information are
1940	!-- written to the output file in that case, the performance
1941	!-- is significantly better compared to the case where only
1942	!-- the first row of PEs in y-direction (myidy = 0) is given
1943	!-- the output while NF90_INDEPENDENT is set.
1944	IF ( npex /= 1 ) THEN
1945
1946	#if defined( __parallel )
1947	!
1948	!-- Distribute data over all PEs along x
1949	ngp = ( nyn-nys+1 ) * ( nzt-nzb + 2 ) * ns
1950	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
1951	CALL MPI_ALLREDUCE( local_2d_sections_l(1,nys,nzb_do), &
1952	local_2d_sections(1,nys,nzb_do), &
1953	ngp, MPI_REAL, MPI_SUM, comm1dx, &
1954	ierr )
1955	#else
1956	local_2d_sections = local_2d_sections_l
1957	#endif
1958	ENDIF
1959	!
1960	!-- Do not output redundant ghost point data except for the
1961	!-- boundaries of the total domain.
1962	! IF ( nyn == ny ) THEN
1963	! nc_stat = NF90_PUT_VAR( id_set_yz(av), &
1964	! id_var_do2d(av,ivar), &
1965	! local_2d_sections(1:ns, &
1966	! nys:nyn+1,nzb_do:nzt_do), &
1967	! start = (/ 1, nys+1, 1, &
1968	! do2d_yz_time_count(av) /), &
1969	! count = (/ ns, nyn-nys+2, &
1970	! nzt_do-nzb_do+1, 1 /) )
1971	! ELSE
1972	nc_stat = NF90_PUT_VAR( id_set_yz(av), &
1973	id_var_do2d(av,ivar), &
1974	local_2d_sections(1:ns,nys:nyn, &
1975	nzb_do:nzt_do), &
1976	start = (/ 1, nys+1, 1, &
1977	do2d_yz_time_count(av) /), &
1978	count = (/ ns, nyn-nys+1, &
1979	nzt_do-nzb_do+1, 1 /) )
1980	! ENDIF
1981
1982	CALL netcdf_handle_error( 'data_output_2d', 60 )
1983
1984	CASE DEFAULT
1985	message_string = 'unknown cross-section: ' // TRIM( mode )
1986	CALL message( 'data_output_2d', 'PA0180', 1, 2, 0, 6, 0 )
1987
1988	END SELECT
1989
1990	ENDIF
1991	#endif
1992	ENDIF
1993
1994	ivar = ivar + 1
1995	l = MAX( 2, LEN_TRIM( do2d(av,ivar) ) )
1996	do2d_mode = do2d(av,ivar)(l-1:l)
1997
1998	ENDDO
1999
2000	!
2001	!-- Deallocate temporary arrays.
2002	IF ( ALLOCATED( level_z ) ) DEALLOCATE( level_z )
2003	IF ( netcdf_data_format > 4 ) THEN
2004	DEALLOCATE( local_pf, local_2d, local_2d_sections )
2005	IF( mode == 'xz' .OR. mode == 'yz' ) DEALLOCATE( local_2d_sections_l )
2006	ENDIF
2007	#if defined( __parallel )
2008	IF ( .NOT. data_output_2d_on_each_pe .AND. myid == 0 ) THEN
2009	DEALLOCATE( total_2d )
2010	ENDIF
2011	#endif
2012
2013	!
2014	!-- Close plot output file.
2015	file_id = 20 + s_ind
2016
2017	IF ( data_output_2d_on_each_pe ) THEN
2018	DO i = 0, io_blocks-1
2019	IF ( i == io_group ) THEN
2020	CALL close_file( file_id )
2021	ENDIF
2022	#if defined( __parallel )
2023	CALL MPI_BARRIER( comm2d, ierr )
2024	#endif
2025	ENDDO
2026	ELSE
2027	IF ( myid == 0 ) CALL close_file( file_id )
2028	ENDIF
2029
2030	CALL cpu_log( log_point(3), 'data_output_2d', 'stop' )
2031
2032	IF ( debug_output_timestep ) CALL debug_message( 'data_output_2d', 'end' )
2033
2034
2035	END SUBROUTINE data_output_2d

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

Download in other formats:

| Impressum | ©Leibniz Universität Hannover |