Home

Context Navigation

source: palm/trunk/SOURCE/data_output_2d.f90 @ 3850

Last change on this file since 3850 was 3766, checked in by raasch, 6 years ago
unused_variables removed, bugfix in im_define_netcdf_grid argument list, trim added to avoid truncation compiler warnings, save attribute added to local targets to avoid outlive pointer target warning, first argument removed from module_interface_rrd_*, file module_interface reformatted with respect to coding standards, bugfix in surface_data_output_rrd_local (variable k removed)
Property svn:keywords set to `Id`
File size: 96.0 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 1997-2019 Leibniz Universitaet Hannover
18	!------------------------------------------------------------------------------!
19	!
20	! Current revisions:
21	! ------------------
22	!
23	!
24	! Former revisions:
25	! -----------------
26	! $Id: data_output_2d.f90 3766 2019-02-26 16:23:41Z raasch $
27	! unused variables removed
28	!
29	! 3655 2019-01-07 16:51:22Z knoop
30	! Bugfix: use time_since_reference_point instead of simulated_time (relevant
31	! when using wall/soil spinup)
32	!
33	! 3637 2018-12-20 01:51:36Z knoop
34	! Implementation of the PALM module interface
35	!
36	! 3597 2018-12-04 08:40:18Z maronga
37	! Added theta_2m
38	!
39	! 3589 2018-11-30 15:09:51Z suehring
40	! Move the control parameter "salsa" from salsa_mod to control_parameters
41	! (M. Kurppa)
42	!
43	! 3582 2018-11-29 19:16:36Z suehring
44	! Remove fill_value from bio_data_output_2d call
45	! dom_dwd_user, Schrempf:
46	! Clean up of biometeorology calls,
47	! remove uv exposure model code, this is now part of biometeorology_mod.
48	!
49	! 3554 2018-11-22 11:24:52Z gronemeier
50	! - add variable description
51	! - rename variable 'if' into 'ivar'
52	! - removed namelist LOCAL
53	! - removed variable rtext
54	!
55	! 3525 2018-11-14 16:06:14Z kanani
56	! Changes related to clean-up of biometeorology (dom_dwd_user)
57	!
58	! 3467 2018-10-30 19:05:21Z suehring
59	! Implementation of a new aerosol module salsa.
60	!
61	! 3448 2018-10-29 18:14:31Z kanani
62	! Add biometeorology
63	!
64	! 3421 2018-10-24 18:39:32Z gronemeier
65	! Renamed output variables
66	!
67	! 3419 2018-10-24 17:27:31Z gronemeier
68	! minor formatting (kanani)
69	! chem_data_output_2d subroutine added (basit)
70	!
71	! 3294 2018-10-01 02:37:10Z raasch
72	! changes concerning modularization of ocean option
73	!
74	! 3274 2018-09-24 15:42:55Z knoop
75	! Modularization of all bulk cloud physics code components
76	!
77	! 3241 2018-09-12 15:02:00Z raasch
78	! unused variables removed
79	!
80	! 3176 2018-07-26 17:12:48Z suehring
81	! Remove output of latent heat flux at urban-surfaces and set fill values
82	! instead
83	!
84	! 3052 2018-05-31 06:11:20Z raasch
85	! Do not open FORTRAN binary files in case of parallel netCDF I/O
86	!
87	! 3049 2018-05-29 13:52:36Z Giersch
88	! Error messages revised
89	!
90	! 3045 2018-05-28 07:55:41Z Giersch
91	! Error messages revised
92	!
93	! 3014 2018-05-09 08:42:38Z maronga
94	! Added nzb_do and nzt_do for some modules for 2d output
95	!
96	! 3004 2018-04-27 12:33:25Z Giersch
97	! precipitation_rate removed, case prr*_xy removed, to_be_resorted have to point
98	! to ql_vp_av and not to ql_vp, allocation checks implemented (averaged data
99	! will be assigned to fill values if no allocation happened so far)
100	!
101	! 2963 2018-04-12 14:47:44Z suehring
102	! Introduce index for vegetation/wall, pavement/green-wall and water/window
103	! surfaces, for clearer access of surface fraction, albedo, emissivity, etc. .
104	!
105	! 2817 2018-02-19 16:32:21Z knoop
106	! Preliminary gust module interface implemented
107	!
108	! 2805 2018-02-14 17:00:09Z suehring
109	! Consider also default-type surfaces for surface temperature output.
110	!
111	! 2797 2018-02-08 13:24:35Z suehring
112	! Enable output of ground-heat flux also at urban surfaces.
113	!
114	! 2743 2018-01-12 16:03:39Z suehring
115	! In case of natural- and urban-type surfaces output surfaces fluxes in W/m2.
116	!
117	! 2742 2018-01-12 14:59:47Z suehring
118	! Enable output of surface temperature
119	!
120	! 2735 2018-01-11 12:01:27Z suehring
121	! output of r_a moved from land-surface to consider also urban-type surfaces
122	!
123	! 2718 2018-01-02 08:49:38Z maronga
124	! Corrected "Former revisions" section
125	!
126	! 2696 2017-12-14 17:12:51Z kanani
127	! Change in file header (GPL part)
128	! Implementation of uv exposure model (FK)
129	! Implementation of turbulence_closure_mod (TG)
130	! Set fill values at topography grid points or e.g. non-natural-type surface
131	! in case of LSM output (MS)
132	!
133	! 2512 2017-10-04 08:26:59Z raasch
134	! upper bounds of cross section output changed from nx+1,ny+1 to nx,ny
135	! no output of ghost layer data
136	!
137	! 2292 2017-06-20 09:51:42Z schwenkel
138	! Implementation of new microphysic scheme: cloud_scheme = 'morrison'
139	! includes two more prognostic equations for cloud drop concentration (nc)
140	! and cloud water content (qc).
141	!
142	! 2277 2017-06-12 10:47:51Z kanani
143	! Removed unused variables do2d_xy_n, do2d_xz_n, do2d_yz_n
144	!
145	! 2233 2017-05-30 18:08:54Z suehring
146	!
147	! 2232 2017-05-30 17:47:52Z suehring
148	! Adjustments to new surface concept
149	!
150	!
151	! 2190 2017-03-21 12:16:43Z raasch
152	! bugfix for r2031: string rho replaced by rho_ocean
153	!
154	! 2031 2016-10-21 15:11:58Z knoop
155	! renamed variable rho to rho_ocean and rho_av to rho_ocean_av
156	!
157	! 2000 2016-08-20 18:09:15Z knoop
158	! Forced header and separation lines into 80 columns
159	!
160	! 1980 2016-07-29 15:51:57Z suehring
161	! Bugfix, in order to steer user-defined output, setting flag found explicitly
162	! to .F.
163	!
164	! 1976 2016-07-27 13:28:04Z maronga
165	! Output of radiation quantities is now done directly in the respective module
166	!
167	! 1972 2016-07-26 07:52:02Z maronga
168	! Output of land surface quantities is now done directly in the respective
169	! module
170	!
171	! 1960 2016-07-12 16:34:24Z suehring
172	! Scalar surface flux added
173	! Rename INTEGER variable s into s_ind, as s is already assigned to scalar
174	!
175	! 1849 2016-04-08 11:33:18Z hoffmann
176	! precipitation_amount, precipitation_rate, prr moved to arrays_3d
177	!
178	! 1822 2016-04-07 07:49:42Z hoffmann
179	! Output of bulk cloud physics simplified.
180	!
181	! 1788 2016-03-10 11:01:04Z maronga
182	! Added output of z0q
183	!
184	! 1783 2016-03-06 18:36:17Z raasch
185	! name change of netcdf routines and module + related changes
186	!
187	! 1745 2016-02-05 13:06:51Z gronemeier
188	! Bugfix: test if time axis limit exceeds moved to point after call of check_open
189	!
190	! 1703 2015-11-02 12:38:44Z raasch
191	! bugfix for output of single (*) xy-sections in case of parallel netcdf I/O
192	!
193	! 1701 2015-11-02 07:43:04Z maronga
194	! Bugfix in output of RRTGM data
195	!
196	! 1691 2015-10-26 16:17:44Z maronga
197	! Added output of Obukhov length (ol) and radiative heating rates for RRTMG.
198	! Formatting corrections.
199	!
200	! 1682 2015-10-07 23:56:08Z knoop
201	! Code annotations made doxygen readable
202	!
203	! 1585 2015-04-30 07:05:52Z maronga
204	! Added support for RRTMG
205	!
206	! 1555 2015-03-04 17:44:27Z maronga
207	! Added output of r_a and r_s
208	!
209	! 1551 2015-03-03 14:18:16Z maronga
210	! Added suppport for land surface model and radiation model output. In the course
211	! of this action, the limits for vertical loops have been changed (from nzb and
212	! nzt+1 to nzb_do and nzt_do, respectively in order to allow soil model output).
213	! Moreover, a new vertical grid zs was introduced.
214	!
215	! 1359 2014-04-11 17:15:14Z hoffmann
216	! New particle structure integrated.
217	!
218	! 1353 2014-04-08 15:21:23Z heinze
219	! REAL constants provided with KIND-attribute
220	!
221	! 1327 2014-03-21 11:00:16Z raasch
222	! parts concerning iso2d output removed,
223	! -netcdf output queries
224	!
225	! 1320 2014-03-20 08:40:49Z raasch
226	! ONLY-attribute added to USE-statements,
227	! kind-parameters added to all INTEGER and REAL declaration statements,
228	! kinds are defined in new module kinds,
229	! revision history before 2012 removed,
230	! comment fields (!:) to be used for variable explanations added to
231	! all variable declaration statements
232	!
233	! 1318 2014-03-17 13:35:16Z raasch
234	! barrier argument removed from cpu_log.
235	! module interfaces removed
236	!
237	! 1311 2014-03-14 12:13:39Z heinze
238	! bugfix: close #if defined( __netcdf )
239	!
240	! 1308 2014-03-13 14:58:42Z fricke
241	! +local_2d_sections, local_2d_sections_l, ns
242	! Check, if the limit of the time dimension is exceeded for parallel output
243	! To increase the performance for parallel output, the following is done:
244	! - Update of time axis is only done by PE0
245	! - Cross sections are first stored on a local array and are written
246	! collectively to the output file by all PEs.
247	!
248	! 1115 2013-03-26 18:16:16Z hoffmann
249	! ql is calculated by calc_liquid_water_content
250	!
251	! 1076 2012-12-05 08:30:18Z hoffmann
252	! Bugfix in output of ql
253	!
254	! 1065 2012-11-22 17:42:36Z hoffmann
255	! Bugfix: Output of cross sections of ql
256	!
257	! 1053 2012-11-13 17:11:03Z hoffmann
258	! +qr, nr, qc and cross sections
259	!
260	! 1036 2012-10-22 13:43:42Z raasch
261	! code put under GPL (PALM 3.9)
262	!
263	! 1031 2012-10-19 14:35:30Z raasch
264	! netCDF4 without parallel file support implemented
265	!
266	! 1007 2012-09-19 14:30:36Z franke
267	! Bugfix: missing calculation of ql_vp added
268	!
269	! 978 2012-08-09 08:28:32Z fricke
270	! +z0h
271	!
272	! Revision 1.1 1997/08/11 06:24:09 raasch
273	! Initial revision
274	!
275	!
276	! Description:
277	! ------------
278	!> Data output of cross-sections in netCDF format or binary format
279	!> to be later converted to NetCDF by helper routine combine_plot_fields.
280	!> Attention: The position of the sectional planes is still not always computed
281	!> --------- correctly. (zu is used always)!
282	!------------------------------------------------------------------------------!
283	SUBROUTINE data_output_2d( mode, av )
284
285
286	USE arrays_3d, &
287	ONLY: dzw, d_exner, e, heatflux_output_conversion, p, pt, q, ql, ql_c, ql_v, s, tend, u, &
288	v, vpt, w, waterflux_output_conversion, zu, zw
289
290	USE averaging
291
292	USE basic_constants_and_equations_mod, &
293	ONLY: c_p, lv_d_cp, l_v
294
295	USE bulk_cloud_model_mod, &
296	ONLY: bulk_cloud_model
297
298	USE control_parameters, &
299	ONLY: data_output_2d_on_each_pe, &
300	data_output_xy, data_output_xz, data_output_yz, do2d, &
301	do2d_xy_last_time, do2d_xy_time_count, &
302	do2d_xz_last_time, do2d_xz_time_count, &
303	do2d_yz_last_time, do2d_yz_time_count, &
304	ibc_uv_b, io_blocks, io_group, message_string, &
305	ntdim_2d_xy, ntdim_2d_xz, ntdim_2d_yz, &
306	psolver, section, &
307	time_since_reference_point
308
309	USE cpulog, &
310	ONLY: cpu_log, log_point
311
312	USE indices, &
313	ONLY: nbgp, nx, nxl, nxlg, nxr, nxrg, ny, nyn, nyng, nys, nysg, &
314	nzb, nzt, wall_flags_0
315
316	USE kinds
317
318	USE land_surface_model_mod, &
319	ONLY: zs
320
321	USE module_interface, &
322	ONLY: module_interface_data_output_2d
323
324	#if defined( __netcdf )
325	USE NETCDF
326	#endif
327
328	USE netcdf_interface, &
329	ONLY: fill_value, id_set_xy, id_set_xz, id_set_yz, id_var_do2d, &
330	id_var_time_xy, id_var_time_xz, id_var_time_yz, nc_stat, &
331	netcdf_data_format, netcdf_handle_error
332
333	USE particle_attributes, &
334	ONLY: grid_particles, number_of_particles, particle_advection_start, &
335	particles, prt_count
336
337	USE pegrid
338
339	USE surface_mod, &
340	ONLY: ind_pav_green, ind_veg_wall, ind_wat_win, surf_def_h, &
341	surf_lsm_h, surf_usm_h
342
343	USE turbulence_closure_mod, &
344	ONLY: tcm_data_output_2d
345
346
347	IMPLICIT NONE
348
349	CHARACTER (LEN=2) :: do2d_mode !< output mode of variable ('xy', 'xz', 'yz')
350	CHARACTER (LEN=2) :: mode !< mode with which the routine is called ('xy', 'xz', 'yz')
351	CHARACTER (LEN=4) :: grid !< string defining the vertical grid
352
353	INTEGER(iwp) :: av !< flag for (non-)average output
354	INTEGER(iwp) :: ngp !< number of grid points of an output slice
355	INTEGER(iwp) :: file_id !< id of output files
356	INTEGER(iwp) :: flag_nr !< number of masking flag
357	INTEGER(iwp) :: i !< loop index
358	INTEGER(iwp) :: iis !< vertical index of a xy slice in array 'local_2d_sections'
359	INTEGER(iwp) :: is !< slice index
360	INTEGER(iwp) :: ivar !< variable index
361	INTEGER(iwp) :: j !< loop index
362	INTEGER(iwp) :: k !< loop index
363	INTEGER(iwp) :: l !< loop index
364	INTEGER(iwp) :: layer_xy !< vertical index of a xy slice in array 'local_pf'
365	INTEGER(iwp) :: m !< loop index
366	INTEGER(iwp) :: n !< loop index
367	INTEGER(iwp) :: nis !< number of vertical slices to be written via parallel NetCDF output
368	INTEGER(iwp) :: ns !< number of output slices
369	INTEGER(iwp) :: nzb_do !< lower limit of the data field (usually nzb)
370	INTEGER(iwp) :: nzt_do !< upper limit of the data field (usually nzt+1)
371	INTEGER(iwp) :: s_ind !< index of slice types (xy=1, xz=2, yz=3)
372	INTEGER(iwp) :: sender !< PE id of sending PE
373	INTEGER(iwp) :: ind(4) !< index limits (lower/upper bounds) of array 'local_2d'
374
375	LOGICAL :: found !< true if output variable was found
376	LOGICAL :: resorted !< true if variable is resorted
377	LOGICAL :: two_d !< true if variable is only two dimensional
378
379	REAL(wp) :: mean_r !< mean particle radius
380	REAL(wp) :: s_r2 !< sum( particle-radius**2 )
381	REAL(wp) :: s_r3 !< sum( particle-radius**3 )
382
383	REAL(wp), DIMENSION(:), ALLOCATABLE :: level_z !< z levels for output array
384	REAL(wp), DIMENSION(:,:), ALLOCATABLE :: local_2d !< local 2-dimensional array containing output values
385	REAL(wp), DIMENSION(:,:), ALLOCATABLE :: local_2d_l !< local 2-dimensional array containing output values
386
387	REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: local_pf !< output array
388	REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: local_2d_sections !< local array containing values at all slices
389	REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: local_2d_sections_l !< local array containing values at all slices
390
391	#if defined( __parallel )
392	REAL(wp), DIMENSION(:,:), ALLOCATABLE :: total_2d !< same as local_2d
393	#endif
394	REAL(wp), DIMENSION(:,:,:), POINTER :: to_be_resorted !< points to array which shall be output
395
396	!
397	!-- Immediate return, if no output is requested (no respective sections
398	!-- found in parameter data_output)
399	IF ( mode == 'xy' .AND. .NOT. data_output_xy(av) ) RETURN
400	IF ( mode == 'xz' .AND. .NOT. data_output_xz(av) ) RETURN
401	IF ( mode == 'yz' .AND. .NOT. data_output_yz(av) ) RETURN
402
403	CALL cpu_log (log_point(3),'data_output_2d','start')
404
405	two_d = .FALSE. ! local variable to distinguish between output of pure 2D
406	! arrays and cross-sections of 3D arrays.
407
408	!
409	!-- Depending on the orientation of the cross-section, the respective output
410	!-- files have to be opened.
411	SELECT CASE ( mode )
412
413	CASE ( 'xy' )
414	s_ind = 1
415	ALLOCATE( level_z(nzb:nzt+1), local_2d(nxl:nxr,nys:nyn) )
416
417	IF ( netcdf_data_format > 4 ) THEN
418	ns = 1
419	DO WHILE ( section(ns,s_ind) /= -9999 .AND. ns <= 100 )
420	ns = ns + 1
421	ENDDO
422	ns = ns - 1
423	ALLOCATE( local_2d_sections(nxl:nxr,nys:nyn,1:ns) )
424	local_2d_sections = 0.0_wp
425	ENDIF
426
427	!
428	!-- Parallel netCDF4/HDF5 output is done on all PEs, all other on PE0 only
429	IF ( myid == 0 .OR. netcdf_data_format > 4 ) THEN
430	CALL check_open( 101+av*10 )
431	ENDIF
432	IF ( data_output_2d_on_each_pe .AND. netcdf_data_format < 5 ) THEN
433	CALL check_open( 21 )
434	ELSE
435	IF ( myid == 0 ) THEN
436	#if defined( __parallel )
437	ALLOCATE( total_2d(0:nx,0:ny) )
438	#endif
439	ENDIF
440	ENDIF
441
442	CASE ( 'xz' )
443	s_ind = 2
444	ALLOCATE( local_2d(nxl:nxr,nzb:nzt+1) )
445
446	IF ( netcdf_data_format > 4 ) THEN
447	ns = 1
448	DO WHILE ( section(ns,s_ind) /= -9999 .AND. ns <= 100 )
449	ns = ns + 1
450	ENDDO
451	ns = ns - 1
452	ALLOCATE( local_2d_sections(nxl:nxr,1:ns,nzb:nzt+1) )
453	ALLOCATE( local_2d_sections_l(nxl:nxr,1:ns,nzb:nzt+1) )
454	local_2d_sections = 0.0_wp; local_2d_sections_l = 0.0_wp
455	ENDIF
456
457	!
458	!-- Parallel netCDF4/HDF5 output is done on all PEs, all other on PE0 only
459	IF ( myid == 0 .OR. netcdf_data_format > 4 ) THEN
460	CALL check_open( 102+av*10 )
461	ENDIF
462
463	IF ( data_output_2d_on_each_pe .AND. netcdf_data_format < 5 ) THEN
464	CALL check_open( 22 )
465	ELSE
466	IF ( myid == 0 ) THEN
467	#if defined( __parallel )
468	ALLOCATE( total_2d(0:nx,nzb:nzt+1) )
469	#endif
470	ENDIF
471	ENDIF
472
473	CASE ( 'yz' )
474	s_ind = 3
475	ALLOCATE( local_2d(nys:nyn,nzb:nzt+1) )
476
477	IF ( netcdf_data_format > 4 ) THEN
478	ns = 1
479	DO WHILE ( section(ns,s_ind) /= -9999 .AND. ns <= 100 )
480	ns = ns + 1
481	ENDDO
482	ns = ns - 1
483	ALLOCATE( local_2d_sections(1:ns,nys:nyn,nzb:nzt+1) )
484	ALLOCATE( local_2d_sections_l(1:ns,nys:nyn,nzb:nzt+1) )
485	local_2d_sections = 0.0_wp; local_2d_sections_l = 0.0_wp
486	ENDIF
487
488	!
489	!-- Parallel netCDF4/HDF5 output is done on all PEs, all other on PE0 only
490	IF ( myid == 0 .OR. netcdf_data_format > 4 ) THEN
491	CALL check_open( 103+av*10 )
492	ENDIF
493
494	IF ( data_output_2d_on_each_pe .AND. netcdf_data_format < 5 ) THEN
495	CALL check_open( 23 )
496	ELSE
497	IF ( myid == 0 ) THEN
498	#if defined( __parallel )
499	ALLOCATE( total_2d(0:ny,nzb:nzt+1) )
500	#endif
501	ENDIF
502	ENDIF
503
504	CASE DEFAULT
505	message_string = 'unknown cross-section: ' // TRIM( mode )
506	CALL message( 'data_output_2d', 'PA0180', 1, 2, 0, 6, 0 )
507
508	END SELECT
509
510	!
511	!-- For parallel netcdf output the time axis must be limited. Return, if this
512	!-- limit is exceeded. This could be the case, if the simulated time exceeds
513	!-- the given end time by the length of the given output interval.
514	IF ( netcdf_data_format > 4 ) THEN
515	IF ( mode == 'xy' .AND. do2d_xy_time_count(av) + 1 > &
516	ntdim_2d_xy(av) ) THEN
517	WRITE ( message_string, * ) 'Output of xy cross-sections is not ', &
518	'given at t=', time_since_reference_point, 's because the', &
519	' maximum number of output time levels is exceeded.'
520	CALL message( 'data_output_2d', 'PA0384', 0, 1, 0, 6, 0 )
521	CALL cpu_log( log_point(3), 'data_output_2d', 'stop' )
522	RETURN
523	ENDIF
524	IF ( mode == 'xz' .AND. do2d_xz_time_count(av) + 1 > &
525	ntdim_2d_xz(av) ) THEN
526	WRITE ( message_string, * ) 'Output of xz cross-sections is not ', &
527	'given at t=', time_since_reference_point, 's because the', &
528	' maximum number of output time levels is exceeded.'
529	CALL message( 'data_output_2d', 'PA0385', 0, 1, 0, 6, 0 )
530	CALL cpu_log( log_point(3), 'data_output_2d', 'stop' )
531	RETURN
532	ENDIF
533	IF ( mode == 'yz' .AND. do2d_yz_time_count(av) + 1 > &
534	ntdim_2d_yz(av) ) THEN
535	WRITE ( message_string, * ) 'Output of yz cross-sections is not ', &
536	'given at t=', time_since_reference_point, 's because the', &
537	' maximum number of output time levels is exceeded.'
538	CALL message( 'data_output_2d', 'PA0386', 0, 1, 0, 6, 0 )
539	CALL cpu_log( log_point(3), 'data_output_2d', 'stop' )
540	RETURN
541	ENDIF
542	ENDIF
543
544	!
545	!-- Allocate a temporary array for resorting (kji -> ijk).
546	ALLOCATE( local_pf(nxl:nxr,nys:nyn,nzb:nzt+1) )
547	local_pf = 0.0
548
549	!
550	!-- Loop of all variables to be written.
551	!-- Output dimensions chosen
552	ivar = 1
553	l = MAX( 2, LEN_TRIM( do2d(av,ivar) ) )
554	do2d_mode = do2d(av,ivar)(l-1:l)
555
556	DO WHILE ( do2d(av,ivar)(1:1) /= ' ' )
557
558	IF ( do2d_mode == mode ) THEN
559	!
560	!-- Set flag to steer output of radiation, land-surface, or user-defined
561	!-- quantities
562	found = .FALSE.
563
564	nzb_do = nzb
565	nzt_do = nzt+1
566	!
567	!-- Before each output, set array local_pf to fill value
568	local_pf = fill_value
569	!
570	!-- Set masking flag for topography for not resorted arrays
571	flag_nr = 0
572
573	!
574	!-- Store the array chosen on the temporary array.
575	resorted = .FALSE.
576	SELECT CASE ( TRIM( do2d(av,ivar) ) )
577	CASE ( 'e_xy', 'e_xz', 'e_yz' )
578	IF ( av == 0 ) THEN
579	to_be_resorted => e
580	ELSE
581	IF ( .NOT. ALLOCATED( e_av ) ) THEN
582	ALLOCATE( e_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
583	e_av = REAL( fill_value, KIND = wp )
584	ENDIF
585	to_be_resorted => e_av
586	ENDIF
587	IF ( mode == 'xy' ) level_z = zu
588
589	CASE ( 'thetal_xy', 'thetal_xz', 'thetal_yz' )
590	IF ( av == 0 ) THEN
591	to_be_resorted => pt
592	ELSE
593	IF ( .NOT. ALLOCATED( lpt_av ) ) THEN
594	ALLOCATE( lpt_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
595	lpt_av = REAL( fill_value, KIND = wp )
596	ENDIF
597	to_be_resorted => lpt_av
598	ENDIF
599	IF ( mode == 'xy' ) level_z = zu
600
601	CASE ( 'lwp*_xy' ) ! 2d-array
602	IF ( av == 0 ) THEN
603	DO i = nxl, nxr
604	DO j = nys, nyn
605	local_pf(i,j,nzb+1) = SUM( ql(nzb:nzt,j,i) * &
606	dzw(1:nzt+1) )
607	ENDDO
608	ENDDO
609	ELSE
610	IF ( .NOT. ALLOCATED( lwp_av ) ) THEN
611	ALLOCATE( lwp_av(nysg:nyng,nxlg:nxrg) )
612	lwp_av = REAL( fill_value, KIND = wp )
613	ENDIF
614	DO i = nxl, nxr
615	DO j = nys, nyn
616	local_pf(i,j,nzb+1) = lwp_av(j,i)
617	ENDDO
618	ENDDO
619	ENDIF
620	resorted = .TRUE.
621	two_d = .TRUE.
622	level_z(nzb+1) = zu(nzb+1)
623
624	CASE ( 'ghf*_xy' ) ! 2d-array
625	IF ( av == 0 ) THEN
626	DO m = 1, surf_lsm_h%ns
627	i = surf_lsm_h%i(m)
628	j = surf_lsm_h%j(m)
629	local_pf(i,j,nzb+1) = surf_lsm_h%ghf(m)
630	ENDDO
631	DO m = 1, surf_usm_h%ns
632	i = surf_usm_h%i(m)
633	j = surf_usm_h%j(m)
634	local_pf(i,j,nzb+1) = surf_usm_h%frac(ind_veg_wall,m) * &
635	surf_usm_h%wghf_eb(m) + &
636	surf_usm_h%frac(ind_pav_green,m) * &
637	surf_usm_h%wghf_eb_green(m) + &
638	surf_usm_h%frac(ind_wat_win,m) * &
639	surf_usm_h%wghf_eb_window(m)
640	ENDDO
641	ELSE
642	IF ( .NOT. ALLOCATED( ghf_av ) ) THEN
643	ALLOCATE( ghf_av(nysg:nyng,nxlg:nxrg) )
644	ghf_av = REAL( fill_value, KIND = wp )
645	ENDIF
646	DO i = nxl, nxr
647	DO j = nys, nyn
648	local_pf(i,j,nzb+1) = ghf_av(j,i)
649	ENDDO
650	ENDDO
651	ENDIF
652
653	resorted = .TRUE.
654	two_d = .TRUE.
655	level_z(nzb+1) = zu(nzb+1)
656
657	CASE ( 'ol*_xy' ) ! 2d-array
658	IF ( av == 0 ) THEN
659	DO m = 1, surf_def_h(0)%ns
660	i = surf_def_h(0)%i(m)
661	j = surf_def_h(0)%j(m)
662	local_pf(i,j,nzb+1) = surf_def_h(0)%ol(m)
663	ENDDO
664	DO m = 1, surf_lsm_h%ns
665	i = surf_lsm_h%i(m)
666	j = surf_lsm_h%j(m)
667	local_pf(i,j,nzb+1) = surf_lsm_h%ol(m)
668	ENDDO
669	DO m = 1, surf_usm_h%ns
670	i = surf_usm_h%i(m)
671	j = surf_usm_h%j(m)
672	local_pf(i,j,nzb+1) = surf_usm_h%ol(m)
673	ENDDO
674	ELSE
675	IF ( .NOT. ALLOCATED( ol_av ) ) THEN
676	ALLOCATE( ol_av(nysg:nyng,nxlg:nxrg) )
677	ol_av = REAL( fill_value, KIND = wp )
678	ENDIF
679	DO i = nxl, nxr
680	DO j = nys, nyn
681	local_pf(i,j,nzb+1) = ol_av(j,i)
682	ENDDO
683	ENDDO
684	ENDIF
685	resorted = .TRUE.
686	two_d = .TRUE.
687	level_z(nzb+1) = zu(nzb+1)
688
689	CASE ( 'p_xy', 'p_xz', 'p_yz' )
690	IF ( av == 0 ) THEN
691	IF ( psolver /= 'sor' ) CALL exchange_horiz( p, nbgp )
692	to_be_resorted => p
693	ELSE
694	IF ( .NOT. ALLOCATED( p_av ) ) THEN
695	ALLOCATE( p_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
696	p_av = REAL( fill_value, KIND = wp )
697	ENDIF
698	IF ( psolver /= 'sor' ) CALL exchange_horiz( p_av, nbgp )
699	to_be_resorted => p_av
700	ENDIF
701	IF ( mode == 'xy' ) level_z = zu
702
703	CASE ( 'pc_xy', 'pc_xz', 'pc_yz' ) ! particle concentration
704	IF ( av == 0 ) THEN
705	IF ( time_since_reference_point >= particle_advection_start ) THEN
706	tend = prt_count
707	! CALL exchange_horiz( tend, nbgp )
708	ELSE
709	tend = 0.0_wp
710	ENDIF
711	DO i = nxl, nxr
712	DO j = nys, nyn
713	DO k = nzb, nzt+1
714	local_pf(i,j,k) = tend(k,j,i)
715	ENDDO
716	ENDDO
717	ENDDO
718	resorted = .TRUE.
719	ELSE
720	IF ( .NOT. ALLOCATED( pc_av ) ) THEN
721	ALLOCATE( pc_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
722	pc_av = REAL( fill_value, KIND = wp )
723	ENDIF
724	! CALL exchange_horiz( pc_av, nbgp )
725	to_be_resorted => pc_av
726	ENDIF
727
728	CASE ( 'pr_xy', 'pr_xz', 'pr_yz' ) ! mean particle radius (effective radius)
729	IF ( av == 0 ) THEN
730	IF ( time_since_reference_point >= particle_advection_start ) THEN
731	DO i = nxl, nxr
732	DO j = nys, nyn
733	DO k = nzb, nzt+1
734	number_of_particles = prt_count(k,j,i)
735	IF (number_of_particles <= 0) CYCLE
736	particles => grid_particles(k,j,i)%particles(1:number_of_particles)
737	s_r2 = 0.0_wp
738	s_r3 = 0.0_wp
739	DO n = 1, number_of_particles
740	IF ( particles(n)%particle_mask ) THEN
741	s_r2 = s_r2 + particles(n)%radius*2 &
742	particles(n)%weight_factor
743	s_r3 = s_r3 + particles(n)%radius*3 &
744	particles(n)%weight_factor
745	ENDIF
746	ENDDO
747	IF ( s_r2 > 0.0_wp ) THEN
748	mean_r = s_r3 / s_r2
749	ELSE
750	mean_r = 0.0_wp
751	ENDIF
752	tend(k,j,i) = mean_r
753	ENDDO
754	ENDDO
755	ENDDO
756	! CALL exchange_horiz( tend, nbgp )
757	ELSE
758	tend = 0.0_wp
759	ENDIF
760	DO i = nxl, nxr
761	DO j = nys, nyn
762	DO k = nzb, nzt+1
763	local_pf(i,j,k) = tend(k,j,i)
764	ENDDO
765	ENDDO
766	ENDDO
767	resorted = .TRUE.
768	ELSE
769	IF ( .NOT. ALLOCATED( pr_av ) ) THEN
770	ALLOCATE( pr_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
771	pr_av = REAL( fill_value, KIND = wp )
772	ENDIF
773	! CALL exchange_horiz( pr_av, nbgp )
774	to_be_resorted => pr_av
775	ENDIF
776
777	CASE ( 'theta_xy', 'theta_xz', 'theta_yz' )
778	IF ( av == 0 ) THEN
779	IF ( .NOT. bulk_cloud_model ) THEN
780	to_be_resorted => pt
781	ELSE
782	DO i = nxl, nxr
783	DO j = nys, nyn
784	DO k = nzb, nzt+1
785	local_pf(i,j,k) = pt(k,j,i) + lv_d_cp * &
786	d_exner(k) * &
787	ql(k,j,i)
788	ENDDO
789	ENDDO
790	ENDDO
791	resorted = .TRUE.
792	ENDIF
793	ELSE
794	IF ( .NOT. ALLOCATED( pt_av ) ) THEN
795	ALLOCATE( pt_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
796	pt_av = REAL( fill_value, KIND = wp )
797	ENDIF
798	to_be_resorted => pt_av
799	ENDIF
800	IF ( mode == 'xy' ) level_z = zu
801
802	CASE ( 'q_xy', 'q_xz', 'q_yz' )
803	IF ( av == 0 ) THEN
804	to_be_resorted => q
805	ELSE
806	IF ( .NOT. ALLOCATED( q_av ) ) THEN
807	ALLOCATE( q_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
808	q_av = REAL( fill_value, KIND = wp )
809	ENDIF
810	to_be_resorted => q_av
811	ENDIF
812	IF ( mode == 'xy' ) level_z = zu
813
814	CASE ( 'ql_xy', 'ql_xz', 'ql_yz' )
815	IF ( av == 0 ) THEN
816	to_be_resorted => ql
817	ELSE
818	IF ( .NOT. ALLOCATED( ql_av ) ) THEN
819	ALLOCATE( ql_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
820	ql_av = REAL( fill_value, KIND = wp )
821	ENDIF
822	to_be_resorted => ql_av
823	ENDIF
824	IF ( mode == 'xy' ) level_z = zu
825
826	CASE ( 'ql_c_xy', 'ql_c_xz', 'ql_c_yz' )
827	IF ( av == 0 ) THEN
828	to_be_resorted => ql_c
829	ELSE
830	IF ( .NOT. ALLOCATED( ql_c_av ) ) THEN
831	ALLOCATE( ql_c_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
832	ql_c_av = REAL( fill_value, KIND = wp )
833	ENDIF
834	to_be_resorted => ql_c_av
835	ENDIF
836	IF ( mode == 'xy' ) level_z = zu
837
838	CASE ( 'ql_v_xy', 'ql_v_xz', 'ql_v_yz' )
839	IF ( av == 0 ) THEN
840	to_be_resorted => ql_v
841	ELSE
842	IF ( .NOT. ALLOCATED( ql_v_av ) ) THEN
843	ALLOCATE( ql_v_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
844	ql_v_av = REAL( fill_value, KIND = wp )
845	ENDIF
846	to_be_resorted => ql_v_av
847	ENDIF
848	IF ( mode == 'xy' ) level_z = zu
849
850	CASE ( 'ql_vp_xy', 'ql_vp_xz', 'ql_vp_yz' )
851	IF ( av == 0 ) THEN
852	IF ( time_since_reference_point >= particle_advection_start ) THEN
853	DO i = nxl, nxr
854	DO j = nys, nyn
855	DO k = nzb, nzt+1
856	number_of_particles = prt_count(k,j,i)
857	IF (number_of_particles <= 0) CYCLE
858	particles => grid_particles(k,j,i)%particles(1:number_of_particles)
859	DO n = 1, number_of_particles
860	IF ( particles(n)%particle_mask ) THEN
861	tend(k,j,i) = tend(k,j,i) + &
862	particles(n)%weight_factor / &
863	prt_count(k,j,i)
864	ENDIF
865	ENDDO
866	ENDDO
867	ENDDO
868	ENDDO
869	! CALL exchange_horiz( tend, nbgp )
870	ELSE
871	tend = 0.0_wp
872	ENDIF
873	DO i = nxl, nxr
874	DO j = nys, nyn
875	DO k = nzb, nzt+1
876	local_pf(i,j,k) = tend(k,j,i)
877	ENDDO
878	ENDDO
879	ENDDO
880	resorted = .TRUE.
881	ELSE
882	IF ( .NOT. ALLOCATED( ql_vp_av ) ) THEN
883	ALLOCATE( ql_vp_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
884	ql_vp_av = REAL( fill_value, KIND = wp )
885	ENDIF
886	! CALL exchange_horiz( ql_vp_av, nbgp )
887	to_be_resorted => ql_vp_av
888	ENDIF
889	IF ( mode == 'xy' ) level_z = zu
890
891	CASE ( 'qsws*_xy' ) ! 2d-array
892	IF ( av == 0 ) THEN
893	local_pf(:,:,nzb+1) = REAL( fill_value, KIND = wp )
894	!
895	!-- In case of default surfaces, clean-up flux by density.
896	!-- In case of land-surfaces, convert fluxes into
897	!-- dynamic units
898	DO m = 1, surf_def_h(0)%ns
899	i = surf_def_h(0)%i(m)
900	j = surf_def_h(0)%j(m)
901	k = surf_def_h(0)%k(m)
902	local_pf(i,j,nzb+1) = surf_def_h(0)%qsws(m) * &
903	waterflux_output_conversion(k)
904	ENDDO
905	DO m = 1, surf_lsm_h%ns
906	i = surf_lsm_h%i(m)
907	j = surf_lsm_h%j(m)
908	k = surf_lsm_h%k(m)
909	local_pf(i,j,nzb+1) = surf_lsm_h%qsws(m) * l_v
910	ENDDO
911	ELSE
912	IF ( .NOT. ALLOCATED( qsws_av ) ) THEN
913	ALLOCATE( qsws_av(nysg:nyng,nxlg:nxrg) )
914	qsws_av = REAL( fill_value, KIND = wp )
915	ENDIF
916	DO i = nxl, nxr
917	DO j = nys, nyn
918	local_pf(i,j,nzb+1) = qsws_av(j,i)
919	ENDDO
920	ENDDO
921	ENDIF
922	resorted = .TRUE.
923	two_d = .TRUE.
924	level_z(nzb+1) = zu(nzb+1)
925
926	CASE ( 'qv_xy', 'qv_xz', 'qv_yz' )
927	IF ( av == 0 ) THEN
928	DO i = nxl, nxr
929	DO j = nys, nyn
930	DO k = nzb, nzt+1
931	local_pf(i,j,k) = q(k,j,i) - ql(k,j,i)
932	ENDDO
933	ENDDO
934	ENDDO
935	resorted = .TRUE.
936	ELSE
937	IF ( .NOT. ALLOCATED( qv_av ) ) THEN
938	ALLOCATE( qv_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
939	qv_av = REAL( fill_value, KIND = wp )
940	ENDIF
941	to_be_resorted => qv_av
942	ENDIF
943	IF ( mode == 'xy' ) level_z = zu
944
945	CASE ( 'r_a*_xy' ) ! 2d-array
946	IF ( av == 0 ) THEN
947	DO m = 1, surf_lsm_h%ns
948	i = surf_lsm_h%i(m)
949	j = surf_lsm_h%j(m)
950	local_pf(i,j,nzb+1) = surf_lsm_h%r_a(m)
951	ENDDO
952
953	DO m = 1, surf_usm_h%ns
954	i = surf_usm_h%i(m)
955	j = surf_usm_h%j(m)
956	local_pf(i,j,nzb+1) = &
957	( surf_usm_h%frac(ind_veg_wall,m) * &
958	surf_usm_h%r_a(m) + &
959	surf_usm_h%frac(ind_pav_green,m) * &
960	surf_usm_h%r_a_green(m) + &
961	surf_usm_h%frac(ind_wat_win,m) * &
962	surf_usm_h%r_a_window(m) )
963	ENDDO
964	ELSE
965	IF ( .NOT. ALLOCATED( r_a_av ) ) THEN
966	ALLOCATE( r_a_av(nysg:nyng,nxlg:nxrg) )
967	r_a_av = REAL( fill_value, KIND = wp )
968	ENDIF
969	DO i = nxl, nxr
970	DO j = nys, nyn
971	local_pf(i,j,nzb+1) = r_a_av(j,i)
972	ENDDO
973	ENDDO
974	ENDIF
975	resorted = .TRUE.
976	two_d = .TRUE.
977	level_z(nzb+1) = zu(nzb+1)
978
979	CASE ( 's_xy', 's_xz', 's_yz' )
980	IF ( av == 0 ) THEN
981	to_be_resorted => s
982	ELSE
983	IF ( .NOT. ALLOCATED( s_av ) ) THEN
984	ALLOCATE( s_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
985	s_av = REAL( fill_value, KIND = wp )
986	ENDIF
987	to_be_resorted => s_av
988	ENDIF
989
990	CASE ( 'shf*_xy' ) ! 2d-array
991	IF ( av == 0 ) THEN
992	!
993	!-- In case of default surfaces, clean-up flux by density.
994	!-- In case of land- and urban-surfaces, convert fluxes into
995	!-- dynamic units.
996	DO m = 1, surf_def_h(0)%ns
997	i = surf_def_h(0)%i(m)
998	j = surf_def_h(0)%j(m)
999	k = surf_def_h(0)%k(m)
1000	local_pf(i,j,nzb+1) = surf_def_h(0)%shf(m) * &
1001	heatflux_output_conversion(k)
1002	ENDDO
1003	DO m = 1, surf_lsm_h%ns
1004	i = surf_lsm_h%i(m)
1005	j = surf_lsm_h%j(m)
1006	k = surf_lsm_h%k(m)
1007	local_pf(i,j,nzb+1) = surf_lsm_h%shf(m) * c_p
1008	ENDDO
1009	DO m = 1, surf_usm_h%ns
1010	i = surf_usm_h%i(m)
1011	j = surf_usm_h%j(m)
1012	k = surf_usm_h%k(m)
1013	local_pf(i,j,nzb+1) = surf_usm_h%shf(m) * c_p
1014	ENDDO
1015	ELSE
1016	IF ( .NOT. ALLOCATED( shf_av ) ) THEN
1017	ALLOCATE( shf_av(nysg:nyng,nxlg:nxrg) )
1018	shf_av = REAL( fill_value, KIND = wp )
1019	ENDIF
1020	DO i = nxl, nxr
1021	DO j = nys, nyn
1022	local_pf(i,j,nzb+1) = shf_av(j,i)
1023	ENDDO
1024	ENDDO
1025	ENDIF
1026	resorted = .TRUE.
1027	two_d = .TRUE.
1028	level_z(nzb+1) = zu(nzb+1)
1029
1030	CASE ( 'ssws*_xy' ) ! 2d-array
1031	IF ( av == 0 ) THEN
1032	DO m = 1, surf_def_h(0)%ns
1033	i = surf_def_h(0)%i(m)
1034	j = surf_def_h(0)%j(m)
1035	local_pf(i,j,nzb+1) = surf_def_h(0)%ssws(m)
1036	ENDDO
1037	DO m = 1, surf_lsm_h%ns
1038	i = surf_lsm_h%i(m)
1039	j = surf_lsm_h%j(m)
1040	local_pf(i,j,nzb+1) = surf_lsm_h%ssws(m)
1041	ENDDO
1042	DO m = 1, surf_usm_h%ns
1043	i = surf_usm_h%i(m)
1044	j = surf_usm_h%j(m)
1045	local_pf(i,j,nzb+1) = surf_usm_h%ssws(m)
1046	ENDDO
1047	ELSE
1048	IF ( .NOT. ALLOCATED( ssws_av ) ) THEN
1049	ALLOCATE( ssws_av(nysg:nyng,nxlg:nxrg) )
1050	ssws_av = REAL( fill_value, KIND = wp )
1051	ENDIF
1052	DO i = nxl, nxr
1053	DO j = nys, nyn
1054	local_pf(i,j,nzb+1) = ssws_av(j,i)
1055	ENDDO
1056	ENDDO
1057	ENDIF
1058	resorted = .TRUE.
1059	two_d = .TRUE.
1060	level_z(nzb+1) = zu(nzb+1)
1061
1062	CASE ( 't*_xy' ) ! 2d-array
1063	IF ( av == 0 ) THEN
1064	DO m = 1, surf_def_h(0)%ns
1065	i = surf_def_h(0)%i(m)
1066	j = surf_def_h(0)%j(m)
1067	local_pf(i,j,nzb+1) = surf_def_h(0)%ts(m)
1068	ENDDO
1069	DO m = 1, surf_lsm_h%ns
1070	i = surf_lsm_h%i(m)
1071	j = surf_lsm_h%j(m)
1072	local_pf(i,j,nzb+1) = surf_lsm_h%ts(m)
1073	ENDDO
1074	DO m = 1, surf_usm_h%ns
1075	i = surf_usm_h%i(m)
1076	j = surf_usm_h%j(m)
1077	local_pf(i,j,nzb+1) = surf_usm_h%ts(m)
1078	ENDDO
1079	ELSE
1080	IF ( .NOT. ALLOCATED( ts_av ) ) THEN
1081	ALLOCATE( ts_av(nysg:nyng,nxlg:nxrg) )
1082	ts_av = REAL( fill_value, KIND = wp )
1083	ENDIF
1084	DO i = nxl, nxr
1085	DO j = nys, nyn
1086	local_pf(i,j,nzb+1) = ts_av(j,i)
1087	ENDDO
1088	ENDDO
1089	ENDIF
1090	resorted = .TRUE.
1091	two_d = .TRUE.
1092	level_z(nzb+1) = zu(nzb+1)
1093
1094	CASE ( 'tsurf*_xy' ) ! 2d-array
1095	IF ( av == 0 ) THEN
1096	DO m = 1, surf_def_h(0)%ns
1097	i = surf_def_h(0)%i(m)
1098	j = surf_def_h(0)%j(m)
1099	local_pf(i,j,nzb+1) = surf_def_h(0)%pt_surface(m)
1100	ENDDO
1101
1102	DO m = 1, surf_lsm_h%ns
1103	i = surf_lsm_h%i(m)
1104	j = surf_lsm_h%j(m)
1105	local_pf(i,j,nzb+1) = surf_lsm_h%pt_surface(m)
1106	ENDDO
1107
1108	DO m = 1, surf_usm_h%ns
1109	i = surf_usm_h%i(m)
1110	j = surf_usm_h%j(m)
1111	local_pf(i,j,nzb+1) = surf_usm_h%pt_surface(m)
1112	ENDDO
1113
1114	ELSE
1115	IF ( .NOT. ALLOCATED( tsurf_av ) ) THEN
1116	ALLOCATE( tsurf_av(nysg:nyng,nxlg:nxrg) )
1117	tsurf_av = REAL( fill_value, KIND = wp )
1118	ENDIF
1119	DO i = nxl, nxr
1120	DO j = nys, nyn
1121	local_pf(i,j,nzb+1) = tsurf_av(j,i)
1122	ENDDO
1123	ENDDO
1124	ENDIF
1125	resorted = .TRUE.
1126	two_d = .TRUE.
1127	level_z(nzb+1) = zu(nzb+1)
1128
1129	CASE ( 'u_xy', 'u_xz', 'u_yz' )
1130	flag_nr = 1
1131	IF ( av == 0 ) THEN
1132	to_be_resorted => u
1133	ELSE
1134	IF ( .NOT. ALLOCATED( u_av ) ) THEN
1135	ALLOCATE( u_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
1136	u_av = REAL( fill_value, KIND = wp )
1137	ENDIF
1138	to_be_resorted => u_av
1139	ENDIF
1140	IF ( mode == 'xy' ) level_z = zu
1141	!
1142	!-- Substitute the values generated by "mirror" boundary condition
1143	!-- at the bottom boundary by the real surface values.
1144	IF ( do2d(av,ivar) == 'u_xz' .OR. do2d(av,ivar) == 'u_yz' ) THEN
1145	IF ( ibc_uv_b == 0 ) local_pf(:,:,nzb) = 0.0_wp
1146	ENDIF
1147
1148	CASE ( 'us*_xy' ) ! 2d-array
1149	IF ( av == 0 ) THEN
1150	DO m = 1, surf_def_h(0)%ns
1151	i = surf_def_h(0)%i(m)
1152	j = surf_def_h(0)%j(m)
1153	local_pf(i,j,nzb+1) = surf_def_h(0)%us(m)
1154	ENDDO
1155	DO m = 1, surf_lsm_h%ns
1156	i = surf_lsm_h%i(m)
1157	j = surf_lsm_h%j(m)
1158	local_pf(i,j,nzb+1) = surf_lsm_h%us(m)
1159	ENDDO
1160	DO m = 1, surf_usm_h%ns
1161	i = surf_usm_h%i(m)
1162	j = surf_usm_h%j(m)
1163	local_pf(i,j,nzb+1) = surf_usm_h%us(m)
1164	ENDDO
1165	ELSE
1166	IF ( .NOT. ALLOCATED( us_av ) ) THEN
1167	ALLOCATE( us_av(nysg:nyng,nxlg:nxrg) )
1168	us_av = REAL( fill_value, KIND = wp )
1169	ENDIF
1170	DO i = nxl, nxr
1171	DO j = nys, nyn
1172	local_pf(i,j,nzb+1) = us_av(j,i)
1173	ENDDO
1174	ENDDO
1175	ENDIF
1176	resorted = .TRUE.
1177	two_d = .TRUE.
1178	level_z(nzb+1) = zu(nzb+1)
1179
1180	CASE ( 'v_xy', 'v_xz', 'v_yz' )
1181	flag_nr = 2
1182	IF ( av == 0 ) THEN
1183	to_be_resorted => v
1184	ELSE
1185	IF ( .NOT. ALLOCATED( v_av ) ) THEN
1186	ALLOCATE( v_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
1187	v_av = REAL( fill_value, KIND = wp )
1188	ENDIF
1189	to_be_resorted => v_av
1190	ENDIF
1191	IF ( mode == 'xy' ) level_z = zu
1192	!
1193	!-- Substitute the values generated by "mirror" boundary condition
1194	!-- at the bottom boundary by the real surface values.
1195	IF ( do2d(av,ivar) == 'v_xz' .OR. do2d(av,ivar) == 'v_yz' ) THEN
1196	IF ( ibc_uv_b == 0 ) local_pf(:,:,nzb) = 0.0_wp
1197	ENDIF
1198
1199	CASE ( 'thetav_xy', 'thetav_xz', 'thetav_yz' )
1200	IF ( av == 0 ) THEN
1201	to_be_resorted => vpt
1202	ELSE
1203	IF ( .NOT. ALLOCATED( vpt_av ) ) THEN
1204	ALLOCATE( vpt_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
1205	vpt_av = REAL( fill_value, KIND = wp )
1206	ENDIF
1207	to_be_resorted => vpt_av
1208	ENDIF
1209	IF ( mode == 'xy' ) level_z = zu
1210
1211	CASE ( 'theta_2m*_xy' ) ! 2d-array
1212	IF ( av == 0 ) THEN
1213	DO m = 1, surf_def_h(0)%ns
1214	i = surf_def_h(0)%i(m)
1215	j = surf_def_h(0)%j(m)
1216	local_pf(i,j,nzb+1) = surf_def_h(0)%pt_2m(m)
1217	ENDDO
1218	DO m = 1, surf_lsm_h%ns
1219	i = surf_lsm_h%i(m)
1220	j = surf_lsm_h%j(m)
1221	local_pf(i,j,nzb+1) = surf_lsm_h%pt_2m(m)
1222	ENDDO
1223	DO m = 1, surf_usm_h%ns
1224	i = surf_usm_h%i(m)
1225	j = surf_usm_h%j(m)
1226	local_pf(i,j,nzb+1) = surf_usm_h%pt_2m(m)
1227	ENDDO
1228	ELSE
1229	IF ( .NOT. ALLOCATED( pt_2m_av ) ) THEN
1230	ALLOCATE( pt_2m_av(nysg:nyng,nxlg:nxrg) )
1231	pt_2m_av = REAL( fill_value, KIND = wp )
1232	ENDIF
1233	DO i = nxl, nxr
1234	DO j = nys, nyn
1235	local_pf(i,j,nzb+1) = pt_2m_av(j,i)
1236	ENDDO
1237	ENDDO
1238	ENDIF
1239	resorted = .TRUE.
1240	two_d = .TRUE.
1241	level_z(nzb+1) = zu(nzb+1)
1242
1243
1244
1245	CASE ( 'w_xy', 'w_xz', 'w_yz' )
1246	flag_nr = 3
1247	IF ( av == 0 ) THEN
1248	to_be_resorted => w
1249	ELSE
1250	IF ( .NOT. ALLOCATED( w_av ) ) THEN
1251	ALLOCATE( w_av(nzb:nzt+1,nysg:nyng,nxlg:nxrg) )
1252	w_av = REAL( fill_value, KIND = wp )
1253	ENDIF
1254	to_be_resorted => w_av
1255	ENDIF
1256	IF ( mode == 'xy' ) level_z = zw
1257
1258	CASE ( 'z0*_xy' ) ! 2d-array
1259	IF ( av == 0 ) THEN
1260	DO m = 1, surf_def_h(0)%ns
1261	i = surf_def_h(0)%i(m)
1262	j = surf_def_h(0)%j(m)
1263	local_pf(i,j,nzb+1) = surf_def_h(0)%z0(m)
1264	ENDDO
1265	DO m = 1, surf_lsm_h%ns
1266	i = surf_lsm_h%i(m)
1267	j = surf_lsm_h%j(m)
1268	local_pf(i,j,nzb+1) = surf_lsm_h%z0(m)
1269	ENDDO
1270	DO m = 1, surf_usm_h%ns
1271	i = surf_usm_h%i(m)
1272	j = surf_usm_h%j(m)
1273	local_pf(i,j,nzb+1) = surf_usm_h%z0(m)
1274	ENDDO
1275	ELSE
1276	IF ( .NOT. ALLOCATED( z0_av ) ) THEN
1277	ALLOCATE( z0_av(nysg:nyng,nxlg:nxrg) )
1278	z0_av = REAL( fill_value, KIND = wp )
1279	ENDIF
1280	DO i = nxl, nxr
1281	DO j = nys, nyn
1282	local_pf(i,j,nzb+1) = z0_av(j,i)
1283	ENDDO
1284	ENDDO
1285	ENDIF
1286	resorted = .TRUE.
1287	two_d = .TRUE.
1288	level_z(nzb+1) = zu(nzb+1)
1289
1290	CASE ( 'z0h*_xy' ) ! 2d-array
1291	IF ( av == 0 ) THEN
1292	DO m = 1, surf_def_h(0)%ns
1293	i = surf_def_h(0)%i(m)
1294	j = surf_def_h(0)%j(m)
1295	local_pf(i,j,nzb+1) = surf_def_h(0)%z0h(m)
1296	ENDDO
1297	DO m = 1, surf_lsm_h%ns
1298	i = surf_lsm_h%i(m)
1299	j = surf_lsm_h%j(m)
1300	local_pf(i,j,nzb+1) = surf_lsm_h%z0h(m)
1301	ENDDO
1302	DO m = 1, surf_usm_h%ns
1303	i = surf_usm_h%i(m)
1304	j = surf_usm_h%j(m)
1305	local_pf(i,j,nzb+1) = surf_usm_h%z0h(m)
1306	ENDDO
1307	ELSE
1308	IF ( .NOT. ALLOCATED( z0h_av ) ) THEN
1309	ALLOCATE( z0h_av(nysg:nyng,nxlg:nxrg) )
1310	z0h_av = REAL( fill_value, KIND = wp )
1311	ENDIF
1312	DO i = nxl, nxr
1313	DO j = nys, nyn
1314	local_pf(i,j,nzb+1) = z0h_av(j,i)
1315	ENDDO
1316	ENDDO
1317	ENDIF
1318	resorted = .TRUE.
1319	two_d = .TRUE.
1320	level_z(nzb+1) = zu(nzb+1)
1321
1322	CASE ( 'z0q*_xy' ) ! 2d-array
1323	IF ( av == 0 ) THEN
1324	DO m = 1, surf_def_h(0)%ns
1325	i = surf_def_h(0)%i(m)
1326	j = surf_def_h(0)%j(m)
1327	local_pf(i,j,nzb+1) = surf_def_h(0)%z0q(m)
1328	ENDDO
1329	DO m = 1, surf_lsm_h%ns
1330	i = surf_lsm_h%i(m)
1331	j = surf_lsm_h%j(m)
1332	local_pf(i,j,nzb+1) = surf_lsm_h%z0q(m)
1333	ENDDO
1334	DO m = 1, surf_usm_h%ns
1335	i = surf_usm_h%i(m)
1336	j = surf_usm_h%j(m)
1337	local_pf(i,j,nzb+1) = surf_usm_h%z0q(m)
1338	ENDDO
1339	ELSE
1340	IF ( .NOT. ALLOCATED( z0q_av ) ) THEN
1341	ALLOCATE( z0q_av(nysg:nyng,nxlg:nxrg) )
1342	z0q_av = REAL( fill_value, KIND = wp )
1343	ENDIF
1344	DO i = nxl, nxr
1345	DO j = nys, nyn
1346	local_pf(i,j,nzb+1) = z0q_av(j,i)
1347	ENDDO
1348	ENDDO
1349	ENDIF
1350	resorted = .TRUE.
1351	two_d = .TRUE.
1352	level_z(nzb+1) = zu(nzb+1)
1353
1354	CASE DEFAULT
1355
1356	!
1357	!-- Quantities of other modules
1358	IF ( .NOT. found ) THEN
1359	CALL module_interface_data_output_2d( &
1360	av, do2d(av,ivar), found, grid, mode, &
1361	local_pf, two_d, nzb_do, nzt_do, &
1362	fill_value &
1363	)
1364	ENDIF
1365
1366	resorted = .TRUE.
1367
1368	IF ( grid == 'zu' ) THEN
1369	IF ( mode == 'xy' ) level_z = zu
1370	ELSEIF ( grid == 'zw' ) THEN
1371	IF ( mode == 'xy' ) level_z = zw
1372	ELSEIF ( grid == 'zu1' ) THEN
1373	IF ( mode == 'xy' ) level_z(nzb+1) = zu(nzb+1)
1374	ELSEIF ( grid == 'zs' ) THEN
1375	IF ( mode == 'xy' ) level_z = zs
1376	ENDIF
1377
1378	IF ( .NOT. found ) THEN
1379	message_string = 'no output provided for: ' // &
1380	TRIM( do2d(av,ivar) )
1381	CALL message( 'data_output_2d', 'PA0181', 0, 0, 0, 6, 0 )
1382	ENDIF
1383
1384	END SELECT
1385
1386	!
1387	!-- Resort the array to be output, if not done above. Flag topography
1388	!-- grid points with fill values, using the corresponding maksing flag.
1389	IF ( .NOT. resorted ) THEN
1390	DO i = nxl, nxr
1391	DO j = nys, nyn
1392	DO k = nzb_do, nzt_do
1393	local_pf(i,j,k) = MERGE( to_be_resorted(k,j,i), &
1394	REAL( fill_value, KIND = wp ), &
1395	BTEST( wall_flags_0(k,j,i), &
1396	flag_nr ) )
1397	ENDDO
1398	ENDDO
1399	ENDDO
1400	ENDIF
1401
1402	!
1403	!-- Output of the individual cross-sections, depending on the cross-
1404	!-- section mode chosen.
1405	is = 1
1406	loop1: DO WHILE ( section(is,s_ind) /= -9999 .OR. two_d )
1407
1408	SELECT CASE ( mode )
1409
1410	CASE ( 'xy' )
1411	!
1412	!-- Determine the cross section index
1413	IF ( two_d ) THEN
1414	layer_xy = nzb+1
1415	ELSE
1416	layer_xy = section(is,s_ind)
1417	ENDIF
1418
1419	!
1420	!-- Exit the loop for layers beyond the data output domain
1421	!-- (used for soil model)
1422	IF ( layer_xy > nzt_do ) THEN
1423	EXIT loop1
1424	ENDIF
1425
1426	!
1427	!-- Update the netCDF xy cross section time axis.
1428	!-- In case of parallel output, this is only done by PE0
1429	!-- to increase the performance.
1430	IF ( time_since_reference_point /= do2d_xy_last_time(av) ) THEN
1431	do2d_xy_time_count(av) = do2d_xy_time_count(av) + 1
1432	do2d_xy_last_time(av) = time_since_reference_point
1433	IF ( myid == 0 ) THEN
1434	IF ( .NOT. data_output_2d_on_each_pe &
1435	.OR. netcdf_data_format > 4 ) &
1436	THEN
1437	#if defined( __netcdf )
1438	nc_stat = NF90_PUT_VAR( id_set_xy(av), &
1439	id_var_time_xy(av), &
1440	(/ time_since_reference_point /), &
1441	start = (/ do2d_xy_time_count(av) /), &
1442	count = (/ 1 /) )
1443	CALL netcdf_handle_error( 'data_output_2d', 53 )
1444	#endif
1445	ENDIF
1446	ENDIF
1447	ENDIF
1448	!
1449	!-- If required, carry out averaging along z
1450	IF ( section(is,s_ind) == -1 .AND. .NOT. two_d ) THEN
1451
1452	local_2d = 0.0_wp
1453	!
1454	!-- Carry out the averaging (all data are on the PE)
1455	DO k = nzb_do, nzt_do
1456	DO j = nys, nyn
1457	DO i = nxl, nxr
1458	local_2d(i,j) = local_2d(i,j) + local_pf(i,j,k)
1459	ENDDO
1460	ENDDO
1461	ENDDO
1462
1463	local_2d = local_2d / ( nzt_do - nzb_do + 1.0_wp)
1464
1465	ELSE
1466	!
1467	!-- Just store the respective section on the local array
1468	local_2d = local_pf(:,:,layer_xy)
1469
1470	ENDIF
1471
1472	#if defined( __parallel )
1473	IF ( netcdf_data_format > 4 ) THEN
1474	!
1475	!-- Parallel output in netCDF4/HDF5 format.
1476	IF ( two_d ) THEN
1477	iis = 1
1478	ELSE
1479	iis = is
1480	ENDIF
1481
1482	#if defined( __netcdf )
1483	!
1484	!-- For parallel output, all cross sections are first stored
1485	!-- here on a local array and will be written to the output
1486	!-- file afterwards to increase the performance.
1487	DO i = nxl, nxr
1488	DO j = nys, nyn
1489	local_2d_sections(i,j,iis) = local_2d(i,j)
1490	ENDDO
1491	ENDDO
1492	#endif
1493	ELSE
1494
1495	IF ( data_output_2d_on_each_pe ) THEN
1496	!
1497	!-- Output of partial arrays on each PE
1498	#if defined( __netcdf )
1499	IF ( myid == 0 ) THEN
1500	WRITE ( 21 ) time_since_reference_point, &
1501	do2d_xy_time_count(av), av
1502	ENDIF
1503	#endif
1504	DO i = 0, io_blocks-1
1505	IF ( i == io_group ) THEN
1506	WRITE ( 21 ) nxl, nxr, nys, nyn, nys, nyn
1507	WRITE ( 21 ) local_2d
1508	ENDIF
1509	#if defined( __parallel )
1510	CALL MPI_BARRIER( comm2d, ierr )
1511	#endif
1512	ENDDO
1513
1514	ELSE
1515	!
1516	!-- PE0 receives partial arrays from all processors and
1517	!-- then outputs them. Here a barrier has to be set,
1518	!-- because otherwise "-MPI- FATAL: Remote protocol queue
1519	!-- full" may occur.
1520	CALL MPI_BARRIER( comm2d, ierr )
1521
1522	ngp = ( nxr-nxl+1 ) * ( nyn-nys+1 )
1523	IF ( myid == 0 ) THEN
1524	!
1525	!-- Local array can be relocated directly.
1526	total_2d(nxl:nxr,nys:nyn) = local_2d
1527	!
1528	!-- Receive data from all other PEs.
1529	DO n = 1, numprocs-1
1530	!
1531	!-- Receive index limits first, then array.
1532	!-- Index limits are received in arbitrary order from
1533	!-- the PEs.
1534	CALL MPI_RECV( ind(1), 4, MPI_INTEGER, &
1535	MPI_ANY_SOURCE, 0, comm2d, &
1536	status, ierr )
1537	sender = status(MPI_SOURCE)
1538	DEALLOCATE( local_2d )
1539	ALLOCATE( local_2d(ind(1):ind(2),ind(3):ind(4)) )
1540	CALL MPI_RECV( local_2d(ind(1),ind(3)), ngp, &
1541	MPI_REAL, sender, 1, comm2d, &
1542	status, ierr )
1543	total_2d(ind(1):ind(2),ind(3):ind(4)) = local_2d
1544	ENDDO
1545	!
1546	!-- Relocate the local array for the next loop increment
1547	DEALLOCATE( local_2d )
1548	ALLOCATE( local_2d(nxl:nxr,nys:nyn) )
1549
1550	#if defined( __netcdf )
1551	IF ( two_d ) THEN
1552	nc_stat = NF90_PUT_VAR( id_set_xy(av), &
1553	id_var_do2d(av,ivar), &
1554	total_2d(0:nx,0:ny), &
1555	start = (/ 1, 1, 1, do2d_xy_time_count(av) /), &
1556	count = (/ nx+1, ny+1, 1, 1 /) )
1557	ELSE
1558	nc_stat = NF90_PUT_VAR( id_set_xy(av), &
1559	id_var_do2d(av,ivar), &
1560	total_2d(0:nx,0:ny), &
1561	start = (/ 1, 1, is, do2d_xy_time_count(av) /), &
1562	count = (/ nx+1, ny+1, 1, 1 /) )
1563	ENDIF
1564	CALL netcdf_handle_error( 'data_output_2d', 54 )
1565	#endif
1566
1567	ELSE
1568	!
1569	!-- First send the local index limits to PE0
1570	ind(1) = nxl; ind(2) = nxr
1571	ind(3) = nys; ind(4) = nyn
1572	CALL MPI_SEND( ind(1), 4, MPI_INTEGER, 0, 0, &
1573	comm2d, ierr )
1574	!
1575	!-- Send data to PE0
1576	CALL MPI_SEND( local_2d(nxl,nys), ngp, &
1577	MPI_REAL, 0, 1, comm2d, ierr )
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	IF ( two_d ) THEN
1590	nc_stat = NF90_PUT_VAR( id_set_xy(av), &
1591	id_var_do2d(av,ivar), &
1592	local_2d(nxl:nxr,nys:nyn), &
1593	start = (/ 1, 1, 1, do2d_xy_time_count(av) /), &
1594	count = (/ nx+1, ny+1, 1, 1 /) )
1595	ELSE
1596	nc_stat = NF90_PUT_VAR( id_set_xy(av), &
1597	id_var_do2d(av,ivar), &
1598	local_2d(nxl:nxr,nys:nyn), &
1599	start = (/ 1, 1, is, do2d_xy_time_count(av) /), &
1600	count = (/ nx+1, ny+1, 1, 1 /) )
1601	ENDIF
1602	CALL netcdf_handle_error( 'data_output_2d', 447 )
1603	#endif
1604	#endif
1605
1606	!
1607	!-- For 2D-arrays (e.g. u*) only one cross-section is available.
1608	!-- Hence exit loop of output levels.
1609	IF ( two_d ) THEN
1610	IF ( netcdf_data_format < 5 ) two_d = .FALSE.
1611	EXIT loop1
1612	ENDIF
1613
1614	CASE ( 'xz' )
1615	!
1616	!-- Update the netCDF xz cross section time axis.
1617	!-- In case of parallel output, this is only done by PE0
1618	!-- to increase the performance.
1619	IF ( time_since_reference_point /= do2d_xz_last_time(av) ) THEN
1620	do2d_xz_time_count(av) = do2d_xz_time_count(av) + 1
1621	do2d_xz_last_time(av) = time_since_reference_point
1622	IF ( myid == 0 ) THEN
1623	IF ( .NOT. data_output_2d_on_each_pe &
1624	.OR. netcdf_data_format > 4 ) &
1625	THEN
1626	#if defined( __netcdf )
1627	nc_stat = NF90_PUT_VAR( id_set_xz(av), &
1628	id_var_time_xz(av), &
1629	(/ time_since_reference_point /), &
1630	start = (/ do2d_xz_time_count(av) /), &
1631	count = (/ 1 /) )
1632	CALL netcdf_handle_error( 'data_output_2d', 56 )
1633	#endif
1634	ENDIF
1635	ENDIF
1636	ENDIF
1637
1638	!
1639	!-- If required, carry out averaging along y
1640	IF ( section(is,s_ind) == -1 ) THEN
1641
1642	ALLOCATE( local_2d_l(nxl:nxr,nzb_do:nzt_do) )
1643	local_2d_l = 0.0_wp
1644	ngp = ( nxr-nxl + 1 ) * ( nzt_do-nzb_do + 1 )
1645	!
1646	!-- First local averaging on the PE
1647	DO k = nzb_do, nzt_do
1648	DO j = nys, nyn
1649	DO i = nxl, nxr
1650	local_2d_l(i,k) = local_2d_l(i,k) + &
1651	local_pf(i,j,k)
1652	ENDDO
1653	ENDDO
1654	ENDDO
1655	#if defined( __parallel )
1656	!
1657	!-- Now do the averaging over all PEs along y
1658	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
1659	CALL MPI_ALLREDUCE( local_2d_l(nxl,nzb_do), &
1660	local_2d(nxl,nzb_do), ngp, MPI_REAL, &
1661	MPI_SUM, comm1dy, ierr )
1662	#else
1663	local_2d = local_2d_l
1664	#endif
1665	local_2d = local_2d / ( ny + 1.0_wp )
1666
1667	DEALLOCATE( local_2d_l )
1668
1669	ELSE
1670	!
1671	!-- Just store the respective section on the local array
1672	!-- (but only if it is available on this PE!)
1673	IF ( section(is,s_ind) >= nys .AND. section(is,s_ind) <= nyn ) &
1674	THEN
1675	local_2d = local_pf(:,section(is,s_ind),nzb_do:nzt_do)
1676	ENDIF
1677
1678	ENDIF
1679
1680	#if defined( __parallel )
1681	IF ( netcdf_data_format > 4 ) THEN
1682	!
1683	!-- Output in netCDF4/HDF5 format.
1684	!-- Output only on those PEs where the respective cross
1685	!-- sections reside. Cross sections averaged along y are
1686	!-- output on the respective first PE along y (myidy=0).
1687	IF ( ( section(is,s_ind) >= nys .AND. &
1688	section(is,s_ind) <= nyn ) .OR. &
1689	( section(is,s_ind) == -1 .AND. myidy == 0 ) ) THEN
1690	#if defined( __netcdf )
1691	!
1692	!-- For parallel output, all cross sections are first
1693	!-- stored here on a local array and will be written to the
1694	!-- output file afterwards to increase the performance.
1695	DO i = nxl, nxr
1696	DO k = nzb_do, nzt_do
1697	local_2d_sections_l(i,is,k) = local_2d(i,k)
1698	ENDDO
1699	ENDDO
1700	#endif
1701	ENDIF
1702
1703	ELSE
1704
1705	IF ( data_output_2d_on_each_pe ) THEN
1706	!
1707	!-- Output of partial arrays on each PE. If the cross
1708	!-- section does not reside on the PE, output special
1709	!-- index values.
1710	#if defined( __netcdf )
1711	IF ( myid == 0 ) THEN
1712	WRITE ( 22 ) time_since_reference_point, &
1713	do2d_xz_time_count(av), av
1714	ENDIF
1715	#endif
1716	DO i = 0, io_blocks-1
1717	IF ( i == io_group ) THEN
1718	IF ( ( section(is,s_ind) >= nys .AND. &
1719	section(is,s_ind) <= nyn ) .OR. &
1720	( section(is,s_ind) == -1 .AND. &
1721	nys-1 == -1 ) ) &
1722	THEN
1723	WRITE (22) nxl, nxr, nzb_do, nzt_do, nzb, nzt+1
1724	WRITE (22) local_2d
1725	ELSE
1726	WRITE (22) -1, -1, -1, -1, -1, -1
1727	ENDIF
1728	ENDIF
1729	#if defined( __parallel )
1730	CALL MPI_BARRIER( comm2d, ierr )
1731	#endif
1732	ENDDO
1733
1734	ELSE
1735	!
1736	!-- PE0 receives partial arrays from all processors of the
1737	!-- respective cross section and outputs them. Here a
1738	!-- barrier has to be set, because otherwise
1739	!-- "-MPI- FATAL: Remote protocol queue full" may occur.
1740	CALL MPI_BARRIER( comm2d, ierr )
1741
1742	ngp = ( nxr-nxl + 1 ) * ( nzt_do-nzb_do + 1 )
1743	IF ( myid == 0 ) THEN
1744	!
1745	!-- Local array can be relocated directly.
1746	IF ( ( section(is,s_ind) >= nys .AND. &
1747	section(is,s_ind) <= nyn ) .OR. &
1748	( section(is,s_ind) == -1 .AND. &
1749	nys-1 == -1 ) ) THEN
1750	total_2d(nxl:nxr,nzb_do:nzt_do) = local_2d
1751	ENDIF
1752	!
1753	!-- Receive data from all other PEs.
1754	DO n = 1, numprocs-1
1755	!
1756	!-- Receive index limits first, then array.
1757	!-- Index limits are received in arbitrary order from
1758	!-- the PEs.
1759	CALL MPI_RECV( ind(1), 4, MPI_INTEGER, &
1760	MPI_ANY_SOURCE, 0, comm2d, &
1761	status, ierr )
1762	!
1763	!-- Not all PEs have data for XZ-cross-section.
1764	IF ( ind(1) /= -9999 ) THEN
1765	sender = status(MPI_SOURCE)
1766	DEALLOCATE( local_2d )
1767	ALLOCATE( local_2d(ind(1):ind(2), &
1768	ind(3):ind(4)) )
1769	CALL MPI_RECV( local_2d(ind(1),ind(3)), ngp, &
1770	MPI_REAL, sender, 1, comm2d, &
1771	status, ierr )
1772	total_2d(ind(1):ind(2),ind(3):ind(4)) = &
1773	local_2d
1774	ENDIF
1775	ENDDO
1776	!
1777	!-- Relocate the local array for the next loop increment
1778	DEALLOCATE( local_2d )
1779	ALLOCATE( local_2d(nxl:nxr,nzb_do:nzt_do) )
1780
1781	#if defined( __netcdf )
1782	nc_stat = NF90_PUT_VAR( id_set_xz(av), &
1783	id_var_do2d(av,ivar), &
1784	total_2d(0:nx,nzb_do:nzt_do), &
1785	start = (/ 1, is, 1, do2d_xz_time_count(av) /), &
1786	count = (/ nx+1, 1, nzt_do-nzb_do+1, 1 /) )
1787	CALL netcdf_handle_error( 'data_output_2d', 58 )
1788	#endif
1789
1790	ELSE
1791	!
1792	!-- If the cross section resides on the PE, send the
1793	!-- local index limits, otherwise send -9999 to PE0.
1794	IF ( ( section(is,s_ind) >= nys .AND. &
1795	section(is,s_ind) <= nyn ) .OR. &
1796	( section(is,s_ind) == -1 .AND. nys-1 == -1 ) ) &
1797	THEN
1798	ind(1) = nxl; ind(2) = nxr
1799	ind(3) = nzb_do; ind(4) = nzt_do
1800	ELSE
1801	ind(1) = -9999; ind(2) = -9999
1802	ind(3) = -9999; ind(4) = -9999
1803	ENDIF
1804	CALL MPI_SEND( ind(1), 4, MPI_INTEGER, 0, 0, &
1805	comm2d, ierr )
1806	!
1807	!-- If applicable, send data to PE0.
1808	IF ( ind(1) /= -9999 ) THEN
1809	CALL MPI_SEND( local_2d(nxl,nzb_do), ngp, &
1810	MPI_REAL, 0, 1, comm2d, ierr )
1811	ENDIF
1812	ENDIF
1813	!
1814	!-- A barrier has to be set, because otherwise some PEs may
1815	!-- proceed too fast so that PE0 may receive wrong data on
1816	!-- tag 0
1817	CALL MPI_BARRIER( comm2d, ierr )
1818	ENDIF
1819
1820	ENDIF
1821	#else
1822	#if defined( __netcdf )
1823	nc_stat = NF90_PUT_VAR( id_set_xz(av), &
1824	id_var_do2d(av,ivar), &
1825	local_2d(nxl:nxr,nzb_do:nzt_do), &
1826	start = (/ 1, is, 1, do2d_xz_time_count(av) /), &
1827	count = (/ nx+1, 1, nzt_do-nzb_do+1, 1 /) )
1828	CALL netcdf_handle_error( 'data_output_2d', 451 )
1829	#endif
1830	#endif
1831
1832	CASE ( 'yz' )
1833	!
1834	!-- Update the netCDF yz cross section time axis.
1835	!-- In case of parallel output, this is only done by PE0
1836	!-- to increase the performance.
1837	IF ( time_since_reference_point /= do2d_yz_last_time(av) ) THEN
1838	do2d_yz_time_count(av) = do2d_yz_time_count(av) + 1
1839	do2d_yz_last_time(av) = time_since_reference_point
1840	IF ( myid == 0 ) THEN
1841	IF ( .NOT. data_output_2d_on_each_pe &
1842	.OR. netcdf_data_format > 4 ) &
1843	THEN
1844	#if defined( __netcdf )
1845	nc_stat = NF90_PUT_VAR( id_set_yz(av), &
1846	id_var_time_yz(av), &
1847	(/ time_since_reference_point /), &
1848	start = (/ do2d_yz_time_count(av) /), &
1849	count = (/ 1 /) )
1850	CALL netcdf_handle_error( 'data_output_2d', 59 )
1851	#endif
1852	ENDIF
1853	ENDIF
1854	ENDIF
1855
1856	!
1857	!-- If required, carry out averaging along x
1858	IF ( section(is,s_ind) == -1 ) THEN
1859
1860	ALLOCATE( local_2d_l(nys:nyn,nzb_do:nzt_do) )
1861	local_2d_l = 0.0_wp
1862	ngp = ( nyn-nys+1 ) * ( nzt_do-nzb_do+1 )
1863	!
1864	!-- First local averaging on the PE
1865	DO k = nzb_do, nzt_do
1866	DO j = nys, nyn
1867	DO i = nxl, nxr
1868	local_2d_l(j,k) = local_2d_l(j,k) + &
1869	local_pf(i,j,k)
1870	ENDDO
1871	ENDDO
1872	ENDDO
1873	#if defined( __parallel )
1874	!
1875	!-- Now do the averaging over all PEs along x
1876	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
1877	CALL MPI_ALLREDUCE( local_2d_l(nys,nzb_do), &
1878	local_2d(nys,nzb_do), ngp, MPI_REAL, &
1879	MPI_SUM, comm1dx, ierr )
1880	#else
1881	local_2d = local_2d_l
1882	#endif
1883	local_2d = local_2d / ( nx + 1.0_wp )
1884
1885	DEALLOCATE( local_2d_l )
1886
1887	ELSE
1888	!
1889	!-- Just store the respective section on the local array
1890	!-- (but only if it is available on this PE!)
1891	IF ( section(is,s_ind) >= nxl .AND. section(is,s_ind) <= nxr ) &
1892	THEN
1893	local_2d = local_pf(section(is,s_ind),:,nzb_do:nzt_do)
1894	ENDIF
1895
1896	ENDIF
1897
1898	#if defined( __parallel )
1899	IF ( netcdf_data_format > 4 ) THEN
1900	!
1901	!-- Output in netCDF4/HDF5 format.
1902	!-- Output only on those PEs where the respective cross
1903	!-- sections reside. Cross sections averaged along x are
1904	!-- output on the respective first PE along x (myidx=0).
1905	IF ( ( section(is,s_ind) >= nxl .AND. &
1906	section(is,s_ind) <= nxr ) .OR. &
1907	( section(is,s_ind) == -1 .AND. myidx == 0 ) ) THEN
1908	#if defined( __netcdf )
1909	!
1910	!-- For parallel output, all cross sections are first
1911	!-- stored here on a local array and will be written to the
1912	!-- output file afterwards to increase the performance.
1913	DO j = nys, nyn
1914	DO k = nzb_do, nzt_do
1915	local_2d_sections_l(is,j,k) = local_2d(j,k)
1916	ENDDO
1917	ENDDO
1918	#endif
1919	ENDIF
1920
1921	ELSE
1922
1923	IF ( data_output_2d_on_each_pe ) THEN
1924	!
1925	!-- Output of partial arrays on each PE. If the cross
1926	!-- section does not reside on the PE, output special
1927	!-- index values.
1928	#if defined( __netcdf )
1929	IF ( myid == 0 ) THEN
1930	WRITE ( 23 ) time_since_reference_point, &
1931	do2d_yz_time_count(av), av
1932	ENDIF
1933	#endif
1934	DO i = 0, io_blocks-1
1935	IF ( i == io_group ) THEN
1936	IF ( ( section(is,s_ind) >= nxl .AND. &
1937	section(is,s_ind) <= nxr ) .OR. &
1938	( section(is,s_ind) == -1 .AND. &
1939	nxl-1 == -1 ) ) &
1940	THEN
1941	WRITE (23) nys, nyn, nzb_do, nzt_do, nzb, nzt+1
1942	WRITE (23) local_2d
1943	ELSE
1944	WRITE (23) -1, -1, -1, -1, -1, -1
1945	ENDIF
1946	ENDIF
1947	#if defined( __parallel )
1948	CALL MPI_BARRIER( comm2d, ierr )
1949	#endif
1950	ENDDO
1951
1952	ELSE
1953	!
1954	!-- PE0 receives partial arrays from all processors of the
1955	!-- respective cross section and outputs them. Here a
1956	!-- barrier has to be set, because otherwise
1957	!-- "-MPI- FATAL: Remote protocol queue full" may occur.
1958	CALL MPI_BARRIER( comm2d, ierr )
1959
1960	ngp = ( nyn-nys+1 ) * ( nzt_do-nzb_do+1 )
1961	IF ( myid == 0 ) THEN
1962	!
1963	!-- Local array can be relocated directly.
1964	IF ( ( section(is,s_ind) >= nxl .AND. &
1965	section(is,s_ind) <= nxr ) .OR. &
1966	( section(is,s_ind) == -1 .AND. nxl-1 == -1 ) ) &
1967	THEN
1968	total_2d(nys:nyn,nzb_do:nzt_do) = local_2d
1969	ENDIF
1970	!
1971	!-- Receive data from all other PEs.
1972	DO n = 1, numprocs-1
1973	!
1974	!-- Receive index limits first, then array.
1975	!-- Index limits are received in arbitrary order from
1976	!-- the PEs.
1977	CALL MPI_RECV( ind(1), 4, MPI_INTEGER, &
1978	MPI_ANY_SOURCE, 0, comm2d, &
1979	status, ierr )
1980	!
1981	!-- Not all PEs have data for YZ-cross-section.
1982	IF ( ind(1) /= -9999 ) THEN
1983	sender = status(MPI_SOURCE)
1984	DEALLOCATE( local_2d )
1985	ALLOCATE( local_2d(ind(1):ind(2), &
1986	ind(3):ind(4)) )
1987	CALL MPI_RECV( local_2d(ind(1),ind(3)), ngp, &
1988	MPI_REAL, sender, 1, comm2d, &
1989	status, ierr )
1990	total_2d(ind(1):ind(2),ind(3):ind(4)) = &
1991	local_2d
1992	ENDIF
1993	ENDDO
1994	!
1995	!-- Relocate the local array for the next loop increment
1996	DEALLOCATE( local_2d )
1997	ALLOCATE( local_2d(nys:nyn,nzb_do:nzt_do) )
1998
1999	#if defined( __netcdf )
2000	nc_stat = NF90_PUT_VAR( id_set_yz(av), &
2001	id_var_do2d(av,ivar), &
2002	total_2d(0:ny,nzb_do:nzt_do), &
2003	start = (/ is, 1, 1, do2d_yz_time_count(av) /), &
2004	count = (/ 1, ny+1, nzt_do-nzb_do+1, 1 /) )
2005	CALL netcdf_handle_error( 'data_output_2d', 61 )
2006	#endif
2007
2008	ELSE
2009	!
2010	!-- If the cross section resides on the PE, send the
2011	!-- local index limits, otherwise send -9999 to PE0.
2012	IF ( ( section(is,s_ind) >= nxl .AND. &
2013	section(is,s_ind) <= nxr ) .OR. &
2014	( section(is,s_ind) == -1 .AND. nxl-1 == -1 ) ) &
2015	THEN
2016	ind(1) = nys; ind(2) = nyn
2017	ind(3) = nzb_do; ind(4) = nzt_do
2018	ELSE
2019	ind(1) = -9999; ind(2) = -9999
2020	ind(3) = -9999; ind(4) = -9999
2021	ENDIF
2022	CALL MPI_SEND( ind(1), 4, MPI_INTEGER, 0, 0, &
2023	comm2d, ierr )
2024	!
2025	!-- If applicable, send data to PE0.
2026	IF ( ind(1) /= -9999 ) THEN
2027	CALL MPI_SEND( local_2d(nys,nzb_do), ngp, &
2028	MPI_REAL, 0, 1, comm2d, ierr )
2029	ENDIF
2030	ENDIF
2031	!
2032	!-- A barrier has to be set, because otherwise some PEs may
2033	!-- proceed too fast so that PE0 may receive wrong data on
2034	!-- tag 0
2035	CALL MPI_BARRIER( comm2d, ierr )
2036	ENDIF
2037
2038	ENDIF
2039	#else
2040	#if defined( __netcdf )
2041	nc_stat = NF90_PUT_VAR( id_set_yz(av), &
2042	id_var_do2d(av,ivar), &
2043	local_2d(nys:nyn,nzb_do:nzt_do), &
2044	start = (/ is, 1, 1, do2d_xz_time_count(av) /), &
2045	count = (/ 1, ny+1, nzt_do-nzb_do+1, 1 /) )
2046	CALL netcdf_handle_error( 'data_output_2d', 452 )
2047	#endif
2048	#endif
2049
2050	END SELECT
2051
2052	is = is + 1
2053	ENDDO loop1
2054
2055	!
2056	!-- For parallel output, all data were collected before on a local array
2057	!-- and are written now to the netcdf file. This must be done to increase
2058	!-- the performance of the parallel output.
2059	#if defined( __netcdf )
2060	IF ( netcdf_data_format > 4 ) THEN
2061
2062	SELECT CASE ( mode )
2063
2064	CASE ( 'xy' )
2065	IF ( two_d ) THEN
2066	nis = 1
2067	two_d = .FALSE.
2068	ELSE
2069	nis = ns
2070	ENDIF
2071	!
2072	!-- Do not output redundant ghost point data except for the
2073	!-- boundaries of the total domain.
2074	! IF ( nxr == nx .AND. nyn /= ny ) THEN
2075	! nc_stat = NF90_PUT_VAR( id_set_xy(av), &
2076	! id_var_do2d(av,ivar), &
2077	! local_2d_sections(nxl:nxr+1, &
2078	! nys:nyn,1:nis), &
2079	! start = (/ nxl+1, nys+1, 1, &
2080	! do2d_xy_time_count(av) /), &
2081	! count = (/ nxr-nxl+2, &
2082	! nyn-nys+1, nis, 1 &
2083	! /) )
2084	! ELSEIF ( nxr /= nx .AND. nyn == ny ) THEN
2085	! nc_stat = NF90_PUT_VAR( id_set_xy(av), &
2086	! id_var_do2d(av,ivar), &
2087	! local_2d_sections(nxl:nxr, &
2088	! nys:nyn+1,1:nis), &
2089	! start = (/ nxl+1, nys+1, 1, &
2090	! do2d_xy_time_count(av) /), &
2091	! count = (/ nxr-nxl+1, &
2092	! nyn-nys+2, nis, 1 &
2093	! /) )
2094	! ELSEIF ( nxr == nx .AND. nyn == ny ) THEN
2095	! nc_stat = NF90_PUT_VAR( id_set_xy(av), &
2096	! id_var_do2d(av,ivar), &
2097	! local_2d_sections(nxl:nxr+1, &
2098	! nys:nyn+1,1:nis), &
2099	! start = (/ nxl+1, nys+1, 1, &
2100	! do2d_xy_time_count(av) /), &
2101	! count = (/ nxr-nxl+2, &
2102	! nyn-nys+2, nis, 1 &
2103	! /) )
2104	! ELSE
2105	nc_stat = NF90_PUT_VAR( id_set_xy(av), &
2106	id_var_do2d(av,ivar), &
2107	local_2d_sections(nxl:nxr, &
2108	nys:nyn,1:nis), &
2109	start = (/ nxl+1, nys+1, 1, &
2110	do2d_xy_time_count(av) /), &
2111	count = (/ nxr-nxl+1, &
2112	nyn-nys+1, nis, 1 &
2113	/) )
2114	! ENDIF
2115
2116	CALL netcdf_handle_error( 'data_output_2d', 55 )
2117
2118	CASE ( 'xz' )
2119	!
2120	!-- First, all PEs get the information of all cross-sections.
2121	!-- Then the data are written to the output file by all PEs
2122	!-- while NF90_COLLECTIVE is set in subroutine
2123	!-- define_netcdf_header. Although redundant information are
2124	!-- written to the output file in that case, the performance
2125	!-- is significantly better compared to the case where only
2126	!-- the first row of PEs in x-direction (myidx = 0) is given
2127	!-- the output while NF90_INDEPENDENT is set.
2128	IF ( npey /= 1 ) THEN
2129
2130	#if defined( __parallel )
2131	!
2132	!-- Distribute data over all PEs along y
2133	ngp = ( nxr-nxl+1 ) * ( nzt_do-nzb_do+1 ) * ns
2134	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
2135	CALL MPI_ALLREDUCE( local_2d_sections_l(nxl,1,nzb_do), &
2136	local_2d_sections(nxl,1,nzb_do), &
2137	ngp, MPI_REAL, MPI_SUM, comm1dy, &
2138	ierr )
2139	#else
2140	local_2d_sections = local_2d_sections_l
2141	#endif
2142	ENDIF
2143	!
2144	!-- Do not output redundant ghost point data except for the
2145	!-- boundaries of the total domain.
2146	! IF ( nxr == nx ) THEN
2147	! nc_stat = NF90_PUT_VAR( id_set_xz(av), &
2148	! id_var_do2d(av,ivar), &
2149	! local_2d_sections(nxl:nxr+1,1:ns, &
2150	! nzb_do:nzt_do), &
2151	! start = (/ nxl+1, 1, 1, &
2152	! do2d_xz_time_count(av) /), &
2153	! count = (/ nxr-nxl+2, ns, nzt_do-nzb_do+1, &
2154	! 1 /) )
2155	! ELSE
2156	nc_stat = NF90_PUT_VAR( id_set_xz(av), &
2157	id_var_do2d(av,ivar), &
2158	local_2d_sections(nxl:nxr,1:ns, &
2159	nzb_do:nzt_do), &
2160	start = (/ nxl+1, 1, 1, &
2161	do2d_xz_time_count(av) /), &
2162	count = (/ nxr-nxl+1, ns, nzt_do-nzb_do+1, &
2163	1 /) )
2164	! ENDIF
2165
2166	CALL netcdf_handle_error( 'data_output_2d', 57 )
2167
2168	CASE ( 'yz' )
2169	!
2170	!-- First, all PEs get the information of all cross-sections.
2171	!-- Then the data are written to the output file by all PEs
2172	!-- while NF90_COLLECTIVE is set in subroutine
2173	!-- define_netcdf_header. Although redundant information are
2174	!-- written to the output file in that case, the performance
2175	!-- is significantly better compared to the case where only
2176	!-- the first row of PEs in y-direction (myidy = 0) is given
2177	!-- the output while NF90_INDEPENDENT is set.
2178	IF ( npex /= 1 ) THEN
2179
2180	#if defined( __parallel )
2181	!
2182	!-- Distribute data over all PEs along x
2183	ngp = ( nyn-nys+1 ) * ( nzt-nzb + 2 ) * ns
2184	IF ( collective_wait ) CALL MPI_BARRIER( comm2d, ierr )
2185	CALL MPI_ALLREDUCE( local_2d_sections_l(1,nys,nzb_do), &
2186	local_2d_sections(1,nys,nzb_do), &
2187	ngp, MPI_REAL, MPI_SUM, comm1dx, &
2188	ierr )
2189	#else
2190	local_2d_sections = local_2d_sections_l
2191	#endif
2192	ENDIF
2193	!
2194	!-- Do not output redundant ghost point data except for the
2195	!-- boundaries of the total domain.
2196	! IF ( nyn == ny ) THEN
2197	! nc_stat = NF90_PUT_VAR( id_set_yz(av), &
2198	! id_var_do2d(av,ivar), &
2199	! local_2d_sections(1:ns, &
2200	! nys:nyn+1,nzb_do:nzt_do), &
2201	! start = (/ 1, nys+1, 1, &
2202	! do2d_yz_time_count(av) /), &
2203	! count = (/ ns, nyn-nys+2, &
2204	! nzt_do-nzb_do+1, 1 /) )
2205	! ELSE
2206	nc_stat = NF90_PUT_VAR( id_set_yz(av), &
2207	id_var_do2d(av,ivar), &
2208	local_2d_sections(1:ns,nys:nyn, &
2209	nzb_do:nzt_do), &
2210	start = (/ 1, nys+1, 1, &
2211	do2d_yz_time_count(av) /), &
2212	count = (/ ns, nyn-nys+1, &
2213	nzt_do-nzb_do+1, 1 /) )
2214	! ENDIF
2215
2216	CALL netcdf_handle_error( 'data_output_2d', 60 )
2217
2218	CASE DEFAULT
2219	message_string = 'unknown cross-section: ' // TRIM( mode )
2220	CALL message( 'data_output_2d', 'PA0180', 1, 2, 0, 6, 0 )
2221
2222	END SELECT
2223
2224	ENDIF
2225	#endif
2226	ENDIF
2227
2228	ivar = ivar + 1
2229	l = MAX( 2, LEN_TRIM( do2d(av,ivar) ) )
2230	do2d_mode = do2d(av,ivar)(l-1:l)
2231
2232	ENDDO
2233
2234	!
2235	!-- Deallocate temporary arrays.
2236	IF ( ALLOCATED( level_z ) ) DEALLOCATE( level_z )
2237	IF ( netcdf_data_format > 4 ) THEN
2238	DEALLOCATE( local_pf, local_2d, local_2d_sections )
2239	IF( mode == 'xz' .OR. mode == 'yz' ) DEALLOCATE( local_2d_sections_l )
2240	ENDIF
2241	#if defined( __parallel )
2242	IF ( .NOT. data_output_2d_on_each_pe .AND. myid == 0 ) THEN
2243	DEALLOCATE( total_2d )
2244	ENDIF
2245	#endif
2246
2247	!
2248	!-- Close plot output file.
2249	file_id = 20 + s_ind
2250
2251	IF ( data_output_2d_on_each_pe ) THEN
2252	DO i = 0, io_blocks-1
2253	IF ( i == io_group ) THEN
2254	CALL close_file( file_id )
2255	ENDIF
2256	#if defined( __parallel )
2257	CALL MPI_BARRIER( comm2d, ierr )
2258	#endif
2259	ENDDO
2260	ELSE
2261	IF ( myid == 0 ) CALL close_file( file_id )
2262	ENDIF
2263
2264	CALL cpu_log( log_point(3), 'data_output_2d', 'stop' )
2265
2266	END SUBROUTINE data_output_2d

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

Download in other formats:

| Impressum | ©Leibniz Universität Hannover |