!> @synthetic_turbulence_generator_mod.f90
!--------------------------------------------------------------------------------------------------!
! This file is part of the PALM model system.
!
! PALM is free software: you can redistribute it and/or modify it under the terms of the GNU General
! Public License as published by the Free Software Foundation, either version 3 of the License, or
! (at your option) any later version.
!
! PALM is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the
! implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General
! Public License for more details.
!
! You should have received a copy of the GNU General Public License along with PALM. If not, see
! .
!
! Copyright 1997-2021 Leibniz Universitaet Hannover
!--------------------------------------------------------------------------------------------------!
!
!
! Current revisions:
! -----------------
!
!
! Former revisions:
! -----------------
! $Id: synthetic_turbulence_generator_mod.f90 4843 2021-01-15 15:22:11Z raasch $
! local namelist parameter added to switch off the module although the respective module namelist
! appears in the namelist file
!
! 4842 2021-01-14 10:42:28Z raasch
! reading of namelist file and actions in case of namelist errors revised so that statement labels
! and goto statements are not required any more
!
! 4828 2021-01-05 11:21:41Z Giersch
! Implementation of downward facing USM and LSM surfaces
!
! 4647 2020-08-24 16:36:18Z suehring
! Change default value of synthetic turbulence adjustment as well as compute_velocity_seeds_local
! By default, the random-seed computation is now distributed among several cores. Especially for
! large length scales this is significantly faster.
!
! 4640 2020-08-11 16:28:32Z suehring
! - to avoid that the correction term in r11/r22 computation becomes unrealistically high, limit
! Obukhov length (term is not valid for near neutral conditions)
! - to avoid unrealistically large perturbations, change computation of r21 so that this resembles
! the vertical transport of horizontal momentum
!
! 4629 2020-07-29 09:37:56Z raasch
! support for MPI Fortran77 interface (mpif.h) removed
!
! 4603 2020-07-14 16:08:30Z suehring
! Bugfix in initialization from ASCII file - x-length scales at the bottom boundary were not
! initialized properly
!
! 4566 2020-06-16 10:11:51Z suehring
! - revise parametrization for reynolds-stress components, turbulent length- and time scales
! - revise computation of velocity disturbances to be consistent to Xie and Castro (2008)
! - change default value of time interval to adjust turbulence parametrization
! - bugfix in computation of amplitude-tensor (vertical flux of horizontal momentum)
!
! 4562 2020-06-12 08:38:47Z raasch
! Parts of r4559 re-formatted
!
! 4559 2020-06-11 08:51:48Z raasch
! File re-formatted to follow the PALM coding standard
!
! 4535 2020-05-15 12:07:23Z raasch
! Bugfix for restart data format query
!
! 4495 2020-04-13 20:11:20Z raasch
! Restart data handling with MPI-IO added
!
! 4481 2020-03-31 18:55:54Z maronga
! Bugfix: cpp-directives for serial mode added, dummy statements to prevent compile errors added
!
! 4442 2020-03-04 19:21:13Z suehring
! Set back turbulent length scale to 8 x grid spacing in the parametrized mode
! (was accidantly changed).
!
! 4441 2020-03-04 19:20:35Z suehring
! Correct misplaced preprocessor directive
!
! 4438 2020-03-03 20:49:28Z suehring
! Performance optimizations in velocity-seed calculation:
! - Random number array is only defined and computed locally (except for normalization to zero mean
! and unit variance)
! - Parallel random number generator is applied independent on the 2D random numbers in other
! routines
! - Option to decide wheter velocity seeds are computed locally without any further communication
! or are computed by all processes along the communicator
!
! 4346 2019-12-18 11:55:56Z motisi
! Introduction of wall_flags_total_0, which currently sets bits based on static topography
! information used in wall_flags_static_0
!
! 4335 2019-12-12 16:39:05Z suehring
! Commentation of last commit
!
! 4332 2019-12-10 19:44:12Z suehring
! Limit initial velocity seeds in restart runs, if not the seed calculation may become unstable.
! Further, minor bugfix in initial velocity seed calculation.
!
! 4329 2019-12-10 15:46:36Z motisi
! Renamed wall_flags_0 to wall_flags_static_0
!
! 4309 2019-11-26 18:49:59Z suehring
! Computation of velocity seeds optimized. This implies that random numbers are computed now using
! the parallel random number generator. Random numbers are now only computed and normalized locally,
! while distributed over all mpi ranks afterwards, instead of computing random numbers on a global
! array.
! Further, the number of calls for the time-consuming velocity-seed generation is reduced - now the
! left and right, as well as the north and south boundary share the same velocity-seed matrices.
!
! 4182 2019-08-22 15:20:23Z scharf
! Corrected "Former revisions" section
!
! 4148 2019-08-08 11:26:00Z suehring
! Remove unused variable
!
! 4144 2019-08-06 09:11:47Z raasch
! Relational operators .EQ., .NE., etc. replaced by ==, /=, etc.
!
! 4071 2019-07-03 20:02:00Z suehring
! Bugfix, initialize mean_inflow_profiles in case turbulence and inflow information is not read from
! file.
!
! 4022 2019-06-12 11:52:39Z suehring
! Several bugfixes and improvements
! - Revise bias correction of the imposed perturbations (correction via volume flow can create
! instabilities in case the mean volume flow is close to zero)
! - Introduce lower limits in calculation of coefficient matrix, else the calculation may become
! numerically unstable
! - Impose perturbations every timestep, even though no new set of perturbations is generated in
! case dt_stg_call /= dt_3d
! - Implement a gradual decrease of Reynolds stress and length scales above ABL height (within 1
! length scale above ABL depth to 1/10) rather than a discontinuous decrease
! - Bugfix in non-nested case: use ABL height for parametrized turbulence
!
! 3987 2019-05-22 09:52:13Z kanani
! Introduce alternative switch for debug output during timestepping
!
! 3938 2019-04-29 16:06:25Z suehring
! Remove unused variables
!
! 3937 2019-04-29 15:09:07Z suehring
! Minor bugfix in case of a very early restart where mc_factor is sill not present.
! Some modification and fixing of potential bugs in the calculation of scaling parameters used for
! synthetic turbulence parametrization.
!
! 3909 2019-04-17 09:13:25Z suehring
! Minor bugfix for last commit
!
! 3900 2019-04-16 15:17:43Z suehring
! Missing re-calculation of perturbation seeds in case of restarts
!
! 3891 2019-04-12 17:52:01Z suehring
! Bugfix in initialization in case of restart runs.
!
! 3885 2019-04-11 11:29:34Z kanani
! Changes related to global restructuring of location messages and introduction of additional debug
! messages
!
!
! Removed unused variables
!
! 3719 2019-02-06 13:10:18Z kanani
! Removed log_point measurement from stg_init, since this part is counted to log_point(2)
! 'initialization' already. Moved other log_points to calls of the subroutines in time_integration
! for better overview.
!
! 2259 2017-06-08 09:09:11Z gronemeier
! Initial revision
!
! Authors:
! --------
! @author Tobias Gronemeier, Matthias Suehring, Atsushi Inagaki, Micha Gryschka, Christoph Knigge
!
!
! Description:
! ------------
!> The module generates turbulence at the inflow boundary based on a method by Xie and Castro (2008)
!> utilizing a Lund rotation (Lund, 1998) and a mass-flux correction by Kim et al. (2013).
!> The turbulence is correlated based on length scales in y- and z-direction and a time scale for
!> each velocity component. The profiles of length and time scales, mean u, v, w, e and pt, and all
!> components of the Reynolds stress tensor can be either read from file STG_PROFILES, or will be
!> parametrized within the boundary layer.
!>
!> @todo Enable cyclic_fill
!> Implement turbulence generation for e and pt
!> @note
!> @bug Height information from input file is not used. Profiles from input must match with current
!> PALM grid.
!> In case of restart, velocity seeds differ from precursor run if a11, a22, or a33 are zero.
!--------------------------------------------------------------------------------------------------!
MODULE synthetic_turbulence_generator_mod
USE arrays_3d, &
ONLY: dzw, &
ddzw, &
drho_air, &
mean_inflow_profiles, &
pt, &
pt_init, &
q, &
q_init, &
u, &
u_init, &
v, &
v_init, &
w, &
zu, &
zw
USE basic_constants_and_equations_mod, &
ONLY: g, &
kappa, &
pi
USE control_parameters, &
ONLY: bc_lr, &
bc_ns, &
child_domain, &
coupling_char, &
debug_output_timestep, &
dt_3d, &
e_init, &
humidity, &
initializing_actions, &
intermediate_timestep_count, &
intermediate_timestep_count_max, &
length, &
message_string, &
nesting_offline, &
neutral, &
num_mean_inflow_profiles, &
random_generator, &
rans_mode, &
restart_data_format_output, &
restart_string, &
syn_turb_gen, &
time_since_reference_point, &
turbulent_inflow
USE cpulog, &
ONLY: cpu_log, &
log_point_s
USE grid_variables, &
ONLY: ddx, &
ddy, &
dx, &
dy
USE indices, &
ONLY: nbgp, &
nz, &
nzb, &
nzt, &
nx, &
nxl, &
nxlu, &
nxr, &
ny, &
nys, &
nysv, &
nyn, &
wall_flags_total_0
USE kinds
#if defined( __parallel )
USE MPI
#endif
USE nesting_offl_mod, &
ONLY: nesting_offl_calc_zi, &
zi_ribulk
USE pegrid, &
ONLY: comm1dx, &
comm1dy, &
comm2d, &
ierr, &
myidx, &
myidy, &
pdims
USE pmc_interface, &
ONLY : rans_mode_parent
USE random_generator_parallel, &
ONLY: init_parallel_random_generator, &
random_dummy, &
random_number_parallel, &
random_seed_parallel
USE restart_data_mpi_io_mod, &
ONLY: rrd_mpi_io, &
wrd_mpi_io
USE transpose_indices, &
ONLY: nzb_x, &
nzt_x
USE surface_mod, &
ONLY: surf_def_h, &
surf_lsm_h, &
surf_usm_h
IMPLICIT NONE
INTEGER(iwp) :: id_stg_left !< left lateral boundary core id in case of turbulence generator
INTEGER(iwp) :: id_stg_north !< north lateral boundary core id in case of turbulence generator
INTEGER(iwp) :: id_stg_right !< right lateral boundary core id in case of turbulence generator
INTEGER(iwp) :: id_stg_south !< south lateral boundary core id in case of turbulence generator
INTEGER(iwp) :: k_zi !< vertical index of boundary-layer top
INTEGER(iwp) :: mergp_limit = 1000 !< maximum length scale (in gp)
INTEGER(iwp) :: mergp_x !< maximum length scale in x (in gp)
INTEGER(iwp) :: mergp_xy !< maximum horizontal length scale (in gp)
INTEGER(iwp) :: mergp_y !< maximum length scale in y (in gp)
INTEGER(iwp) :: mergp_z !< maximum length scale in z (in gp)
INTEGER(iwp) :: nzb_x_stg !< lower bound of z coordinate (required for transposing z on PEs along x)
INTEGER(iwp) :: nzt_x_stg !< upper bound of z coordinate (required for transposing z on PEs along x)
INTEGER(iwp) :: nzb_y_stg !< lower bound of z coordinate (required for transposing z on PEs along y)
INTEGER(iwp) :: nzt_y_stg !< upper bound of z coordinate (required for transposing z on PEs along y)
#if defined( __parallel )
INTEGER(iwp) :: stg_type_xz !< MPI type for full z range
INTEGER(iwp) :: stg_type_xz_small !< MPI type for small z range
INTEGER(iwp) :: stg_type_yz !< MPI type for full z range
INTEGER(iwp) :: stg_type_yz_small !< MPI type for small z range
#endif
INTEGER(iwp), DIMENSION(3) :: nr_non_topo_xz = 0 !< number of non-topography grid points at xz cross-sections,
!< required for bias correction of imposed perturbations
INTEGER(iwp), DIMENSION(3) :: nr_non_topo_yz = 0 !< number of non-topography grid points at yz cross-sections,
!< required for bias correction of imposed perturbations
#if defined( __parallel )
INTEGER(iwp), DIMENSION(:), ALLOCATABLE :: displs_xz !< displacement for MPI_GATHERV
INTEGER(iwp), DIMENSION(:), ALLOCATABLE :: recv_count_xz !< receive count for MPI_GATHERV
INTEGER(iwp), DIMENSION(:), ALLOCATABLE :: displs_yz !< displacement for MPI_GATHERV
INTEGER(iwp), DIMENSION(:), ALLOCATABLE :: recv_count_yz !< receive count for MPI_GATHERV
#endif
INTEGER(iwp), DIMENSION(:), ALLOCATABLE :: nux !< length scale of u in x direction (in gp)
INTEGER(iwp), DIMENSION(:), ALLOCATABLE :: nuy !< length scale of u in y direction (in gp)
INTEGER(iwp), DIMENSION(:), ALLOCATABLE :: nuz !< length scale of u in z direction (in gp)
INTEGER(iwp), DIMENSION(:), ALLOCATABLE :: nvx !< length scale of v in x direction (in gp)
INTEGER(iwp), DIMENSION(:), ALLOCATABLE :: nvy !< length scale of v in y direction (in gp)
INTEGER(iwp), DIMENSION(:), ALLOCATABLE :: nvz !< length scale of v in z direction (in gp)
INTEGER(iwp), DIMENSION(:), ALLOCATABLE :: nwx !< length scale of w in x direction (in gp)
INTEGER(iwp), DIMENSION(:), ALLOCATABLE :: nwy !< length scale of w in y direction (in gp)
INTEGER(iwp), DIMENSION(:), ALLOCATABLE :: nwz !< length scale of w in z direction (in gp)
INTEGER(isp), DIMENSION(:), ALLOCATABLE :: id_rand_xz !< initial random IDs at xz inflow boundary
INTEGER(isp), DIMENSION(:), ALLOCATABLE :: id_rand_yz !< initial random IDs at yz inflow boundary
INTEGER(isp), DIMENSION(:,:), ALLOCATABLE :: seq_rand_xz !< initial random seeds at xz inflow boundary
INTEGER(isp), DIMENSION(:,:), ALLOCATABLE :: seq_rand_yz !< initial random seeds at yz inflow boundary
LOGICAL :: adjustment_step = .FALSE. !< control flag indicating that time and lenght scales have been updated and
!< no time correlation to the timestep before should be considered
LOGICAL :: compute_velocity_seeds_local = .FALSE. !< switch to decide whether velocity seeds are computed locally
!< or if computation is distributed over several processes
LOGICAL :: parametrize_inflow_turbulence = .FALSE. !< flag indicating that inflow turbulence is either read from file
!< (.FALSE.) or if it parametrized
LOGICAL :: use_syn_turb_gen = .FALSE. !< switch to use synthetic turbulence generator
LOGICAL :: velocity_seed_initialized = .FALSE. !< true after first call of stg_main
REAL(wp) :: blend !< value to create gradually and smooth blending of Reynolds stress and length
!< scales above the boundary layer
REAL(wp) :: blend_coeff = -9.3_wp !< coefficient used to ensure that blending functions decreases to 1/10 after
!< one length scale above ABL top
REAL(wp) :: d_l !< blend_coeff/length_scale
REAL(wp) :: d_nxy !< inverse of the total number of xy-grid points
REAL(wp) :: dt_stg_adjust = 1800.0_wp !< time interval for adjusting turbulence statistics
REAL(wp) :: dt_stg_call = 0.0_wp !< time interval for calling synthetic turbulence generator
REAL(wp) :: scale_l !< scaling parameter used for turbulence parametrization - Obukhov length
REAL(wp) :: scale_us !< scaling parameter used for turbulence parametrization - friction velocity
REAL(wp) :: scale_wm !< scaling parameter used for turbulence parametrization - momentum scale
REAL(wp) :: time_stg_adjust = 0.0_wp !< time counter for adjusting turbulence information
REAL(wp) :: time_stg_call = 0.0_wp !< time counter for calling generator
REAL(wp), DIMENSION(3) :: mc_factor = 1.0_wp !< correction factor for the u,v,w-components to maintain original mass flux
REAL(wp),DIMENSION(:), ALLOCATABLE :: r11 !< Reynolds parameter
REAL(wp),DIMENSION(:), ALLOCATABLE :: r21 !< Reynolds parameter
REAL(wp),DIMENSION(:), ALLOCATABLE :: r22 !< Reynolds parameter
REAL(wp),DIMENSION(:), ALLOCATABLE :: r31 !< Reynolds parameter
REAL(wp),DIMENSION(:), ALLOCATABLE :: r32 !< Reynolds parameter
REAL(wp),DIMENSION(:), ALLOCATABLE :: r33 !< Reynolds parameter
REAL(wp), DIMENSION(:), ALLOCATABLE :: a11 !< coefficient for Lund rotation
REAL(wp), DIMENSION(:), ALLOCATABLE :: a21 !< coefficient for Lund rotation
REAL(wp), DIMENSION(:), ALLOCATABLE :: a22 !< coefficient for Lund rotation
REAL(wp), DIMENSION(:), ALLOCATABLE :: a31 !< coefficient for Lund rotation
REAL(wp), DIMENSION(:), ALLOCATABLE :: a32 !< coefficient for Lund rotation
REAL(wp), DIMENSION(:), ALLOCATABLE :: a33 !< coefficient for Lund rotation
REAL(wp), DIMENSION(:), ALLOCATABLE :: tu !< Lagrangian time scale of u
REAL(wp), DIMENSION(:), ALLOCATABLE :: tv !< Lagrangian time scale of v
REAL(wp), DIMENSION(:), ALLOCATABLE :: tw !< Lagrangian time scale of w
REAL(wp), DIMENSION(:,:), ALLOCATABLE :: bux !< filter function for u in x direction
REAL(wp), DIMENSION(:,:), ALLOCATABLE :: buy !< filter function for u in y direction
REAL(wp), DIMENSION(:,:), ALLOCATABLE :: buz !< filter function for u in z direction
REAL(wp), DIMENSION(:,:), ALLOCATABLE :: bvx !< filter function for v in x direction
REAL(wp), DIMENSION(:,:), ALLOCATABLE :: bvy !< filter function for v in y direction
REAL(wp), DIMENSION(:,:), ALLOCATABLE :: bvz !< filter function for v in z direction
REAL(wp), DIMENSION(:,:), ALLOCATABLE :: bwx !< filter function for w in y direction
REAL(wp), DIMENSION(:,:), ALLOCATABLE :: bwy !< filter function for w in y direction
REAL(wp), DIMENSION(:,:), ALLOCATABLE :: bwz !< filter function for w in z direction
REAL(wp), DIMENSION(:,:), ALLOCATABLE :: fu_xz !< velocity seed for u at xz plane
REAL(wp), DIMENSION(:,:), ALLOCATABLE :: fuo_xz !< velocity seed for u at xz plane with new random number
REAL(wp), DIMENSION(:,:), ALLOCATABLE :: fu_yz !< velocity seed for u at yz plane
REAL(wp), DIMENSION(:,:), ALLOCATABLE :: fuo_yz !< velocity seed for u at yz plane with new random number
REAL(wp), DIMENSION(:,:), ALLOCATABLE :: fv_xz !< velocity seed for v at xz plane
REAL(wp), DIMENSION(:,:), ALLOCATABLE :: fvo_xz !< velocity seed for v at xz plane with new random number
REAL(wp), DIMENSION(:,:), ALLOCATABLE :: fv_yz !< velocity seed for v at yz plane
REAL(wp), DIMENSION(:,:), ALLOCATABLE :: fvo_yz !< velocity seed for v at yz plane with new random number
REAL(wp), DIMENSION(:,:), ALLOCATABLE :: fw_xz !< velocity seed for w at xz plane
REAL(wp), DIMENSION(:,:), ALLOCATABLE :: fwo_xz !< velocity seed for w at xz plane with new random number
REAL(wp), DIMENSION(:,:), ALLOCATABLE :: fw_yz !< velocity seed for w at yz plane
REAL(wp), DIMENSION(:,:), ALLOCATABLE :: fwo_yz !< velocity seed for w at yz plane with new random number
REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: dist_xz !< disturbances for parallel/crosswind/vertical component at north/south boundary
REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: dist_yz !< disturbances for parallel/crosswind/vertical component at north/south boundary
!
!-- PALM interfaces:
!-- Adjust time and lenght scales, Reynolds stress, and filter functions
INTERFACE stg_adjust
MODULE PROCEDURE stg_adjust
END INTERFACE stg_adjust
!
!-- Input parameter checks to be done in check_parameters
INTERFACE stg_check_parameters
MODULE PROCEDURE stg_check_parameters
END INTERFACE stg_check_parameters
!
!-- Calculate filter functions
INTERFACE stg_filter_func
MODULE PROCEDURE stg_filter_func
END INTERFACE stg_filter_func
!
!-- Generate velocity seeds at south and north domain boundary
INTERFACE stg_generate_seed_xz
MODULE PROCEDURE stg_generate_seed_xz
END INTERFACE stg_generate_seed_xz
!
!-- Generate velocity seeds at left and/or right domain boundary
INTERFACE stg_generate_seed_yz
MODULE PROCEDURE stg_generate_seed_yz
END INTERFACE stg_generate_seed_yz
!
!-- Output of information to the header file
INTERFACE stg_header
MODULE PROCEDURE stg_header
END INTERFACE stg_header
!
!-- Initialization actions
INTERFACE stg_init
MODULE PROCEDURE stg_init
END INTERFACE stg_init
!
!-- Main procedure of synth. turb. gen.
INTERFACE stg_main
MODULE PROCEDURE stg_main
END INTERFACE stg_main
!
!-- Reading of NAMELIST parameters
INTERFACE stg_parin
MODULE PROCEDURE stg_parin
END INTERFACE stg_parin
!
!-- Reading of parameters for restart runs
INTERFACE stg_rrd_global
MODULE PROCEDURE stg_rrd_global_ftn
MODULE PROCEDURE stg_rrd_global_mpi
END INTERFACE stg_rrd_global
!
!-- Writing of binary output for restart runs
INTERFACE stg_wrd_global
MODULE PROCEDURE stg_wrd_global
END INTERFACE stg_wrd_global
SAVE
PRIVATE
!
!-- Public interfaces
PUBLIC stg_adjust, &
stg_check_parameters, &
stg_header, &
stg_init, &
stg_main, &
stg_parin, &
stg_rrd_global, &
stg_wrd_global
!
!-- Public variables
PUBLIC dt_stg_call, &
dt_stg_adjust, &
id_stg_left, &
id_stg_north, &
id_stg_right, &
id_stg_south, &
parametrize_inflow_turbulence, &
time_stg_adjust, &
time_stg_call, &
use_syn_turb_gen
CONTAINS
!--------------------------------------------------------------------------------------------------!
! Description:
! ------------
!> Check parameters routine for synthetic turbulence generator
!--------------------------------------------------------------------------------------------------!
SUBROUTINE stg_check_parameters
IF ( .NOT. use_syn_turb_gen .AND. .NOT. rans_mode .AND. &
nesting_offline ) THEN
message_string = 'Synthetic turbulence generator is required ' // &
'if offline nesting is applied and PALM operates in LES mode.'
CALL message( 'stg_check_parameters', 'PA0520', 0, 0, 0, 6, 0 )
ENDIF
IF ( .NOT. use_syn_turb_gen .AND. child_domain &
.AND. rans_mode_parent .AND. .NOT. rans_mode ) THEN
message_string = 'Synthetic turbulence generator is required when nesting is applied ' // &
'and parent operates in RANS-mode but current child in LES mode.'
CALL message( 'stg_check_parameters', 'PA0524', 1, 2, 0, 6, 0 )
ENDIF
IF ( use_syn_turb_gen ) THEN
IF ( child_domain .AND. .NOT. rans_mode .AND. .NOT. rans_mode_parent ) THEN
message_string = 'Using synthetic turbulence generator is not allowed in LES-LES nesting.'
CALL message( 'stg_check_parameters', 'PA0620', 1, 2, 0, 6, 0 )
ENDIF
IF ( child_domain .AND. rans_mode .AND. rans_mode_parent ) THEN
message_string = 'Using synthetic turbulence generator is not allowed in RANS-RANS nesting.'
CALL message( 'stg_check_parameters', 'PA0621', 1, 2, 0, 6, 0 )
ENDIF
IF ( .NOT. nesting_offline .AND. .NOT. child_domain ) THEN
IF ( INDEX( initializing_actions, 'set_constant_profiles' ) == 0 &
.AND. INDEX( initializing_actions, 'read_restart_data' ) == 0 ) THEN
message_string = 'Using synthetic turbulence generator requires ' // &
'%initializing_actions = "set_constant_profiles" or ' // &
' "read_restart_data", if not offline nesting is applied.'
CALL message( 'stg_check_parameters', 'PA0015', 1, 2, 0, 6, 0 )
ENDIF
IF ( bc_lr /= 'dirichlet/radiation' ) THEN
message_string = 'Using synthetic turbulence generator requires &bc_lr = ' // &
' "dirichlet/radiation", if not offline nesting is applied.'
CALL message( 'stg_check_parameters', 'PA0035', 1, 2, 0, 6, 0 )
ENDIF
IF ( bc_ns /= 'cyclic' ) THEN
message_string = 'Using synthetic turbulence generator requires &bc_ns = ' // &
' "cyclic", if not offline nesting is applied.'
CALL message( 'stg_check_parameters', 'PA0037', 1, 2, 0, 6, 0 )
ENDIF
ENDIF
IF ( turbulent_inflow ) THEN
message_string = 'Using synthetic turbulence generator in combination ' // &
'&with turbulent_inflow = .T. is not allowed'
CALL message( 'stg_check_parameters', 'PA0039', 1, 2, 0, 6, 0 )
ENDIF
!
!-- Synthetic turbulence generator requires the parallel random generator
IF ( random_generator /= 'random-parallel' ) THEN
message_string = 'Using synthetic turbulence generator requires random_generator = ' // &
'random-parallel.'
CALL message( 'stg_check_parameters', 'PA0421', 1, 2, 0, 6, 0 )
ENDIF
ENDIF
END SUBROUTINE stg_check_parameters
!--------------------------------------------------------------------------------------------------!
! Description:
! ------------
!> Header output for synthetic turbulence generator
!--------------------------------------------------------------------------------------------------!
SUBROUTINE stg_header ( io )
INTEGER(iwp), INTENT(IN) :: io !< Unit of the output file
!
!-- Write synthetic turbulence generator Header
WRITE( io, 1 )
IF ( use_syn_turb_gen ) THEN
WRITE( io, 2 )
ELSE
WRITE( io, 3 )
ENDIF
IF ( parametrize_inflow_turbulence ) THEN
WRITE( io, 4 ) dt_stg_adjust
ELSE
WRITE( io, 5 )
ENDIF
1 FORMAT (//' Synthetic turbulence generator information:'/ &
' ------------------------------------------'/)
2 FORMAT (' synthetic turbulence generator is switched on')
3 FORMAT (' synthetic turbulence generator is switched off')
4 FORMAT (' imposed turbulence statistics are parametrized and ajdusted to boundary-layer development each ', F8.2, ' s' )
5 FORMAT (' imposed turbulence is read from file' )
END SUBROUTINE stg_header
!--------------------------------------------------------------------------------------------------!
! Description:
! ------------
!> Initialization of the synthetic turbulence generator
!--------------------------------------------------------------------------------------------------!
SUBROUTINE stg_init
#if defined( __parallel )
INTEGER(KIND=MPI_ADDRESS_KIND) :: extent !< extent of new MPI type
INTEGER(KIND=MPI_ADDRESS_KIND) :: tob = 0 !< dummy variable
#endif
INTEGER(iwp) :: i !> grid index in x-direction
INTEGER(iwp) :: j !> loop index
INTEGER(iwp) :: k !< index
#if defined( __parallel )
INTEGER(iwp) :: newtype !< dummy MPI type
INTEGER(iwp) :: realsize !< size of REAL variables
#endif
INTEGER(iwp), DIMENSION(3) :: nr_non_topo_xz_l = 0 !< number of non-topography grid points at xz-cross-section on subdomain
INTEGER(iwp), DIMENSION(3) :: nr_non_topo_yz_l = 0 !< number of non-topography grid points at yz-cross-section on subdomain
LOGICAL :: file_stg_exist = .FALSE. !< flag indicating whether parameter file for Reynolds stress and length scales exist
!
!-- Dummy variables used for reading profiles
REAL(wp) :: d1 !< u profile
REAL(wp) :: d2 !< v profile
REAL(wp) :: d3 !< w profile
REAL(wp) :: d5 !< e profile
REAL(wp) :: luy !< length scale for u in y direction
REAL(wp) :: luz !< length scale for u in z direction
REAL(wp) :: lvy !< length scale for v in y direction
REAL(wp) :: lvz !< length scale for v in z direction
REAL(wp) :: lwy !< length scale for w in y direction
REAL(wp) :: lwz !< length scale for w in z direction
#if defined( __parallel )
REAL(wp) :: nnz !< increment used to determine processor decomposition of z-axis along x and y direction
#endif
REAL(wp) :: zz !< height
#if defined( __parallel )
CALL MPI_BARRIER( comm2d, ierr )
#endif
!
!-- Create mpi-datatypes for exchange in case of non-local but distributed computation of the
!-- velocity seeds. This option is useful in case large turbulent length scales are present, where
!-- the computational effort becomes large and needs to be parallelized. For parameterized turbulence
!-- the length scales are small and computing the velocity seeds locally is faster (no overhead by
!-- communication).
IF ( .NOT. compute_velocity_seeds_local ) THEN
#if defined( __parallel )
!
!-- Determine processor decomposition of z-axis along x- and y-direction
nnz = nz / pdims(1)
nzb_x_stg = 1 + myidx * INT( nnz )
nzt_x_stg = ( myidx + 1 ) * INT( nnz )
IF ( MOD( nz , pdims(1) ) /= 0 .AND. myidx == id_stg_right ) &
nzt_x_stg = nzt_x_stg + myidx * ( nnz - INT( nnz ) )
IF ( nesting_offline .OR. ( child_domain .AND. rans_mode_parent &
.AND. .NOT. rans_mode ) ) THEN
nnz = nz / pdims(2)
nzb_y_stg = 1 + myidy * INT( nnz )
nzt_y_stg = ( myidy + 1 ) * INT( nnz )
IF ( MOD( nz , pdims(2) ) /= 0 .AND. myidy == id_stg_north ) &
nzt_y_stg = nzt_y_stg + myidy * ( nnz - INT( nnz ) )
ENDIF
!
!-- Define MPI type used in stg_generate_seed_yz to gather vertical splitted velocity seeds
CALL MPI_TYPE_SIZE( MPI_REAL, realsize, ierr )
extent = 1 * realsize
!
!-- Set-up MPI datatyp to involve all cores for turbulence generation at yz layer
!-- stg_type_yz: yz-slice with vertical bounds nzb:nzt+1
CALL MPI_TYPE_CREATE_SUBARRAY( 2, [nzt-nzb+2,nyn-nys+1], &
[1,nyn-nys+1], [0,0], MPI_ORDER_FORTRAN, MPI_REAL, newtype, ierr )
CALL MPI_TYPE_CREATE_RESIZED( newtype, tob, extent, stg_type_yz, ierr )
CALL MPI_TYPE_COMMIT( stg_type_yz, ierr )
CALL MPI_TYPE_FREE( newtype, ierr )
! stg_type_yz_small: yz-slice with vertical bounds nzb_x_stg:nzt_x_stg+1
CALL MPI_TYPE_CREATE_SUBARRAY( 2, [nzt_x_stg-nzb_x_stg+2,nyn-nys+1], &
[1,nyn-nys+1], [0,0], MPI_ORDER_FORTRAN, MPI_REAL, newtype, ierr )
CALL MPI_TYPE_CREATE_RESIZED( newtype, tob, extent, stg_type_yz_small, ierr )
CALL MPI_TYPE_COMMIT( stg_type_yz_small, ierr )
CALL MPI_TYPE_FREE( newtype, ierr )
! Receive count and displacement for MPI_GATHERV in stg_generate_seed_yz
ALLOCATE( recv_count_yz(pdims(1)), displs_yz(pdims(1)) )
recv_count_yz = nzt_x_stg-nzb_x_stg + 1
recv_count_yz(pdims(1)) = recv_count_yz(pdims(1)) + 1
DO j = 1, pdims(1)
displs_yz(j) = 0 + (nzt_x_stg-nzb_x_stg+1) * (j-1)
ENDDO
!
!-- Set-up MPI datatyp to involve all cores for turbulence generation at xz layer
!-- stg_type_xz: xz-slice with vertical bounds nzb:nzt+1
IF ( nesting_offline .OR. ( child_domain .AND. rans_mode_parent &
.AND. .NOT. rans_mode ) ) THEN
CALL MPI_TYPE_CREATE_SUBARRAY( 2, [nzt-nzb+2,nxr-nxl+1], &
[1,nxr-nxl+1], [0,0], MPI_ORDER_FORTRAN, MPI_REAL, newtype, ierr )
CALL MPI_TYPE_CREATE_RESIZED( newtype, tob, extent, stg_type_xz, ierr )
CALL MPI_TYPE_COMMIT( stg_type_xz, ierr )
CALL MPI_TYPE_FREE( newtype, ierr )
! stg_type_yz_small: xz-slice with vertical bounds nzb_x_stg:nzt_x_stg+1
CALL MPI_TYPE_CREATE_SUBARRAY( 2, [nzt_y_stg-nzb_y_stg+2,nxr-nxl+1], &
[1,nxr-nxl+1], [0,0], MPI_ORDER_FORTRAN, MPI_REAL, newtype, ierr )
CALL MPI_TYPE_CREATE_RESIZED( newtype, tob, extent, stg_type_xz_small, ierr )
CALL MPI_TYPE_COMMIT( stg_type_xz_small, ierr )
CALL MPI_TYPE_FREE( newtype, ierr )
! Receive count and displacement for MPI_GATHERV in stg_generate_seed_yz
ALLOCATE( recv_count_xz(pdims(2)), displs_xz(pdims(2)) )
recv_count_xz = nzt_y_stg-nzb_y_stg + 1
recv_count_xz(pdims(2)) = recv_count_xz(pdims(2)) + 1
DO j = 1, pdims(2)
displs_xz(j) = 0 + (nzt_y_stg-nzb_y_stg+1) * (j-1)
ENDDO
ENDIF
#endif
ENDIF
!
!-- Allocate required arrays.
!-- In case no offline nesting or self nesting is used, the arrary mean_inflow profiles is required.
!-- Check if it is already allocated, else allocate and initialize it appropriately. Note, in case
!-- turbulence and inflow information is read from file, mean_inflow_profiles is already allocated
!-- and initialized appropriately.
IF ( .NOT. nesting_offline .AND. .NOT. child_domain ) THEN
IF ( .NOT. ALLOCATED( mean_inflow_profiles ) ) THEN
ALLOCATE( mean_inflow_profiles(nzb:nzt+1,1:num_mean_inflow_profiles) )
mean_inflow_profiles = 0.0_wp
mean_inflow_profiles(:,1) = u_init
mean_inflow_profiles(:,2) = v_init
!
!-- Even though potential temperature and humidity are not perturbed, they need to be
!-- initialized appropriately.
IF ( .NOT. neutral ) &
mean_inflow_profiles(:,4) = pt_init
IF ( humidity ) &
mean_inflow_profiles(:,6) = q_init
ENDIF
ENDIF
!
!-- Assign initial profiles. Note, this is only required if turbulent inflow from the left is
!-- desired, not in case of any of the nesting (offline or self nesting) approaches.
IF ( .NOT. nesting_offline .AND. .NOT. child_domain ) THEN
u_init = mean_inflow_profiles(:,1)
v_init = mean_inflow_profiles(:,2)
e_init = MAXVAL( mean_inflow_profiles(:,5) )
ENDIF
ALLOCATE ( a11(nzb:nzt+1), a21(nzb:nzt+1), a22(nzb:nzt+1), &
a31(nzb:nzt+1), a32(nzb:nzt+1), a33(nzb:nzt+1), &
nux(nzb:nzt+1), nuy(nzb:nzt+1), nuz(nzb:nzt+1), &
nvx(nzb:nzt+1), nvy(nzb:nzt+1), nvz(nzb:nzt+1), &
nwx(nzb:nzt+1), nwy(nzb:nzt+1), nwz(nzb:nzt+1), &
r11(nzb:nzt+1), r21(nzb:nzt+1), r22(nzb:nzt+1), &
r31(nzb:nzt+1), r32(nzb:nzt+1), r33(nzb:nzt+1), &
tu(nzb:nzt+1), tv(nzb:nzt+1), tw(nzb:nzt+1) )
r11 = 0.0_wp
r21 = 0.0_wp
r22 = 0.0_wp
r31 = 0.0_wp
r32 = 0.0_wp
r33 = 0.0_wp
tu = 0.0_wp
tv = 0.0_wp
tw = 0.0_wp
ALLOCATE ( dist_xz(nzb:nzt+1,nxl:nxr,3) )
ALLOCATE ( dist_yz(nzb:nzt+1,nys:nyn,3) )
dist_xz = 0.0_wp
dist_yz = 0.0_wp
!
!-- Read inflow profile
!-- Height levels of profiles in input profile is as follows:
!-- zu: luy, luz, tu, lvy, lvz, tv, r11, r21, r22, d1, d2, d5 zw: lwy, lwz, tw, r31, r32, r33, d3
!-- WARNING: zz is not used at the moment
INQUIRE( FILE = 'STG_PROFILES' // TRIM( coupling_char ), EXIST = file_stg_exist )
IF ( file_stg_exist ) THEN
OPEN( 90, FILE = 'STG_PROFILES' // TRIM( coupling_char ), STATUS = 'OLD', FORM = 'FORMATTED' )
!
!-- Skip header
READ( 90, * )
DO k = nzb, nzt
READ( 90, * ) zz, luy, luz, tu(k), lvy, lvz, tv(k), lwy, lwz, tw(k), r11(k), r21(k), &
r22(k), r31(k), r32(k), r33(k), d1, d2, d3, d5
!
!-- Convert length scales from meter to number of grid points.
IF ( k /= nzb ) THEN
nuz(k) = INT( luz * ddzw(k) )
nvz(k) = INT( lvz * ddzw(k) )
nwz(k) = INT( lwz * ddzw(k) )
ELSE
nuz(k) = INT( luz * ddzw(k+1) )
nvz(k) = INT( lvz * ddzw(k+1) )
nwz(k) = INT( lwz * ddzw(k+1) )
ENDIF
nuy(k) = INT( luy * ddy )
nvy(k) = INT( lvy * ddy )
nwy(k) = INT( lwy * ddy )
!
!-- Set length scales in x-direction. As a workaround assume isotropic turbulence in x- and
!-- y-direction.
nwx(k) = nwy(k)
nvx(k) = nvy(k)
nux(k) = nuy(k)
!
!-- Save Mean inflow profiles
IF ( TRIM( initializing_actions ) /= 'read_restart_data' ) THEN
mean_inflow_profiles(k,1) = d1
mean_inflow_profiles(k,2) = d2
! mean_inflow_profiles(k,4) = d4
mean_inflow_profiles(k,5) = d5
ENDIF
ENDDO
!
!-- Set length scales at the surface and top boundary. At the surface the lengths scales are
!-- simply overwritten.
nwx(nzb) = nwy(nzb+1)
nvx(nzb) = nvy(nzb+1)
nux(nzb) = nuy(nzb+1)
nuy(nzb) = nuy(nzb+1)
nuz(nzb) = nuz(nzb+1)
nvy(nzb) = nvy(nzb+1)
nvz(nzb) = nvz(nzb+1)
nwy(nzb) = nwy(nzb+1)
nwz(nzb) = nwz(nzb+1)
nwx(nzt+1) = nwy(nzt)
nvx(nzt+1) = nvy(nzt)
nux(nzt+1) = nuy(nzt)
nuy(nzt+1) = nuy(nzt)
nuz(nzt+1) = nuz(nzt)
nvy(nzt+1) = nvy(nzt)
nvz(nzt+1) = nvz(nzt)
nwy(nzt+1) = nwy(nzt)
nwz(nzt+1) = nwz(nzt)
CLOSE( 90 )
!
!-- Calculate coefficient matrix from Reynolds stress tensor (Lund rotation)
CALL calc_coeff_matrix
!
!-- No information about turbulence and its length scales are available. Instead, parametrize
!-- turbulence which is imposed at the boundaries. Set flag which indicates that turbulence is
!-- parametrized, which is done later when energy-balance models are already initialized. This is
!-- because the STG needs information about surface properties such as roughness to build
!-- 'realistic' turbulence profiles.
ELSE
!
!-- Precalulate the inversion number of xy-grid points - later used for mass conservation
d_nxy = 1.0_wp / REAL( ( nx + 1 ) * ( ny + 1 ), KIND = wp )
!
!-- Set flag indicating that turbulence is parametrized
parametrize_inflow_turbulence = .TRUE.
!
!-- In case of dirichlet inflow boundary conditions only at one lateral boundary, i.e. in the
!-- case no offline or self nesting is applied but synthetic turbulence shall be parametrized
!-- nevertheless, the boundary-layer depth needs to be determined first.
IF ( .NOT. nesting_offline .AND. .NOT. child_domain ) &
CALL nesting_offl_calc_zi
!
!-- Determine boundary-layer depth, which is used to initialize lenght scales
CALL calc_scaling_variables
!
!-- Parametrize Reynolds-stress tensor, diagonal elements as well as r21 (v'u'), r31 (w'u'),
!-- r32 (w'v'). Parametrization follows Rotach et al. (1996) and is based on boundary-layer depth,
!-- friction velocity and velocity scale.
CALL parametrize_turbulence
!
!-- Calculate coefficient matrix from Reynolds stress tensor (Lund rotation)
CALL calc_coeff_matrix
ENDIF
!
!-- Initial filter functions
CALL stg_setup_filter_function
!
!-- Allocate velocity seeds for turbulence at xz-layer
ALLOCATE ( fu_xz(nzb:nzt+1,nxl:nxr), fuo_xz(nzb:nzt+1,nxl:nxr), &
fv_xz(nzb:nzt+1,nxl:nxr), fvo_xz(nzb:nzt+1,nxl:nxr), &
fw_xz(nzb:nzt+1,nxl:nxr), fwo_xz(nzb:nzt+1,nxl:nxr) )
!
!-- Allocate velocity seeds for turbulence at yz-layer
ALLOCATE ( fu_yz(nzb:nzt+1,nys:nyn), fuo_yz(nzb:nzt+1,nys:nyn), &
fv_yz(nzb:nzt+1,nys:nyn), fvo_yz(nzb:nzt+1,nys:nyn), &
fw_yz(nzb:nzt+1,nys:nyn), fwo_yz(nzb:nzt+1,nys:nyn) )
fu_xz = 0.0_wp
fuo_xz = 0.0_wp
fv_xz = 0.0_wp
fvo_xz = 0.0_wp
fw_xz = 0.0_wp
fwo_xz = 0.0_wp
fu_yz = 0.0_wp
fuo_yz = 0.0_wp
fv_yz = 0.0_wp
fvo_yz = 0.0_wp
fw_yz = 0.0_wp
fwo_yz = 0.0_wp
#if defined( __parallel )
CALL MPI_BARRIER( comm2d, ierr )
#endif
!
!-- In case of restart, calculate velocity seeds fu, fv, fw from former time step.
! Bug: fu, fv, fw are different in those heights where a11, a22, a33 are 0 compared to the prerun.
!-- This is mostly for k=nzt+1.
IF ( TRIM( initializing_actions ) == 'read_restart_data' ) THEN
IF ( myidx == id_stg_left .OR. myidx == id_stg_right ) THEN
IF ( myidx == id_stg_left ) i = -1
IF ( myidx == id_stg_right ) i = nxr+1
DO j = nys, nyn
DO k = nzb, nzt+1
IF ( a11(k) > 10E-8_wp ) THEN
fu_yz(k,j) = ( u(k,j,i) - u_init(k) ) / a11(k)
ELSE
fu_yz(k,j) = 10E-8_wp
ENDIF
IF ( a22(k) > 10E-8_wp ) THEN
fv_yz(k,j) = ( v(k,j,i) - a21(k) * fu_yz(k,j) - v_init(k) ) / a22(k)
ELSE
fv_yz(k,j) = 10E-8_wp
ENDIF
IF ( a33(k) > 10E-8_wp ) THEN
fw_yz(k,j) = ( w(k,j,i) - a31(k) * fu_yz(k,j) - a32(k) * fv_yz(k,j) ) / a33(k)
ELSE
fw_yz(k,j) = 10E-8_wp
ENDIF
ENDDO
ENDDO
ENDIF
IF ( myidy == id_stg_south .OR. myidy == id_stg_north ) THEN
IF ( myidy == id_stg_south ) j = -1
IF ( myidy == id_stg_north ) j = nyn+1
DO i = nxl, nxr
DO k = nzb, nzt+1
!
!-- In case the correlation coefficients are very small, the velocity seeds may become
!-- very large finally creating numerical instabilities in the synthetic turbulence
!-- generator. Empirically, a value of 10E-8 seems to be sufficient.
IF ( a11(k) > 10E-8_wp ) THEN
fu_xz(k,i) = ( u(k,j,i) - u_init(k) ) / a11(k)
ELSE
fu_xz(k,i) = 10E-8_wp
ENDIF
IF ( a22(k) > 10E-8_wp ) THEN
fv_xz(k,i) = ( v(k,j,i) - a21(k) * fu_xz(k,i) - v_init(k) ) / a22(k)
ELSE
fv_xz(k,i) = 10E-8_wp
ENDIF
IF ( a33(k) > 10E-8_wp ) THEN
fw_xz(k,i) = ( w(k,j,i) - a31(k) * fu_xz(k,i) - a32(k) * fv_xz(k,i) ) / a33(k)
ELSE
fw_xz(k,i) = 10E-8_wp
ENDIF
ENDDO
ENDDO
ENDIF
ENDIF
!
!-- Count the number of non-topography grid points at the boundaries where perturbations are imposed.
!-- This number is later used for bias corrections of the perturbations, i.e. to force their mean to
!-- be zero. Please note, due to the asymetry of u and v along x and y direction, respectively,
!-- different cases must be distinguished.
IF ( myidx == id_stg_left .OR. myidx == id_stg_right ) THEN
!
!-- Number of grid points where perturbations are imposed on u
IF ( myidx == id_stg_left ) i = nxl
IF ( myidx == id_stg_right ) i = nxr+1
nr_non_topo_yz_l(1) = SUM( MERGE( 1, 0, BTEST( wall_flags_total_0(nzb:nzt,nys:nyn,i), 1 ) ) )
!
!-- Number of grid points where perturbations are imposed on v and w
IF ( myidx == id_stg_left ) i = nxl-1
IF ( myidx == id_stg_right ) i = nxr+1
nr_non_topo_yz_l(2) = SUM( MERGE( 1, 0, BTEST( wall_flags_total_0(nzb:nzt,nysv:nyn,i), 2 ) ) )
nr_non_topo_yz_l(3) = SUM( MERGE( 1, 0, BTEST( wall_flags_total_0(nzb:nzt,nys:nyn,i), 3 ) ) )
#if defined( __parallel )
CALL MPI_ALLREDUCE( nr_non_topo_yz_l, nr_non_topo_yz, 3, MPI_INTEGER, MPI_SUM, comm1dy, ierr )
#else
nr_non_topo_yz = nr_non_topo_yz_l
#endif
ENDIF
IF ( myidy == id_stg_south .OR. myidy == id_stg_north ) THEN
!
!-- Number of grid points where perturbations are imposed on v
IF ( myidy == id_stg_south ) j = nys
IF ( myidy == id_stg_north ) j = nyn+1
nr_non_topo_xz_l(2) = SUM( MERGE( 1, 0, BTEST( wall_flags_total_0(nzb:nzt,j,nxl:nxr), 2 ) ) )
!
!-- Number of grid points where perturbations are imposed on u and w
IF ( myidy == id_stg_south ) j = nys-1
IF ( myidy == id_stg_north ) j = nyn+1
nr_non_topo_xz_l(1) = SUM( MERGE( 1, 0, BTEST( wall_flags_total_0(nzb:nzt,j,nxlu:nxr), 1 ) ) )
nr_non_topo_xz_l(3) = SUM( MERGE( 1, 0, BTEST( wall_flags_total_0(nzb:nzt,j,nxl:nxr), 3 ) ) )
#if defined( __parallel )
CALL MPI_ALLREDUCE( nr_non_topo_xz_l, nr_non_topo_xz, 3, MPI_INTEGER, MPI_SUM, comm1dx, ierr )
#else
nr_non_topo_xz = nr_non_topo_xz_l
#endif
ENDIF
!
!-- Initialize random number generator at xz- and yz-layers. Random numbers are initialized at each
!-- core. In case there is only inflow from the left, it is sufficient to generate only random
!-- numbers for the yz-layer, else random numbers for the xz-layer are also required.
ALLOCATE ( id_rand_yz(-mergp_limit+nys:nyn+mergp_limit) )
ALLOCATE ( seq_rand_yz(5,-mergp_limit+nys:nyn+mergp_limit) )
id_rand_yz = 0
seq_rand_yz = 0
CALL init_parallel_random_generator( ny, -mergp_limit+nys, nyn+mergp_limit, id_rand_yz, seq_rand_yz )
IF ( nesting_offline .OR. ( child_domain .AND. rans_mode_parent &
.AND. .NOT. rans_mode ) ) THEN
ALLOCATE ( id_rand_xz(-mergp_limit+nxl:nxr+mergp_limit) )
ALLOCATE ( seq_rand_xz(5,-mergp_limit+nxl:nxr+mergp_limit) )
id_rand_xz = 0
seq_rand_xz = 0
CALL init_parallel_random_generator( nx, -mergp_limit+nxl, nxr+mergp_limit, id_rand_xz, seq_rand_xz )
ENDIF
END SUBROUTINE stg_init
!--------------------------------------------------------------------------------------------------!
! Description:
! ------------
!> Calculate filter function bxx from length scale nxx following Eg.9 and 10 (Xie and Castro, 2008)
!--------------------------------------------------------------------------------------------------!
SUBROUTINE stg_filter_func( nxx, bxx, mergp )
INTEGER(iwp) :: k !< loop index
INTEGER(iwp) :: mergp !< passed length scale in grid points
INTEGER(iwp) :: n_k !< length scale nXX in height k
INTEGER(iwp) :: nf !< index for length scales
INTEGER(iwp), DIMENSION(nzb:nzt+1) :: nxx !< length scale (in gp)
REAL(wp) :: bdenom !< denominator for filter functions bXX
REAL(wp) :: qsi = 1.0_wp !< minimization factor
REAL(wp), DIMENSION(-mergp:mergp,nzb:nzt+1) :: bxx !< filter function
bxx = 0.0_wp
DO k = nzb, nzt+1
bdenom = 0.0_wp
n_k = nxx(k)
IF ( n_k /= 0 ) THEN
!
!-- ( Eq.10 )^2
DO nf = -n_k, n_k
bdenom = bdenom + EXP( -qsi * pi * ABS( nf ) / n_k )**2
ENDDO
!
!-- ( Eq.9 )
bdenom = SQRT( bdenom )
DO nf = -n_k, n_k
bxx(nf,k) = EXP( -qsi * pi * ABS( nf ) / n_k ) / bdenom
ENDDO
ENDIF
ENDDO
END SUBROUTINE stg_filter_func
!------------------------------------------------------------------------------!
! Description:
! ------------
!> Calculate filter function bxx from length scale nxx following Eg.9 and 10
!> (Xie and Castro, 2008)
!------------------------------------------------------------------------------!
SUBROUTINE stg_setup_filter_function
INTEGER(iwp) :: j !< dummy value to calculate maximum length scale index
INTEGER(iwp) :: k !< loop index along vertical direction
!
!-- Define the size of the filter functions and allocate them.
mergp_x = 0
mergp_y = 0
mergp_z = 0
DO k = nzb, nzt+1
j = MAX( ABS(nux(k)), ABS(nvx(k)), ABS(nwx(k)) )
IF ( j > mergp_x ) mergp_x = j
ENDDO
DO k = nzb, nzt+1
j = MAX( ABS(nuy(k)), ABS(nvy(k)), ABS(nwy(k)) )
IF ( j > mergp_y ) mergp_y = j
ENDDO
DO k = nzb, nzt+1
j = MAX( ABS(nuz(k)), ABS(nvz(k)), ABS(nwz(k)) )
IF ( j > mergp_z ) mergp_z = j
ENDDO
mergp_xy = MAX( mergp_x, mergp_y )
IF ( ALLOCATED( bux ) ) DEALLOCATE( bux )
IF ( ALLOCATED( bvx ) ) DEALLOCATE( bvx )
IF ( ALLOCATED( bwx ) ) DEALLOCATE( bwx )
IF ( ALLOCATED( buy ) ) DEALLOCATE( buy )
IF ( ALLOCATED( bvy ) ) DEALLOCATE( bvy )
IF ( ALLOCATED( bwy ) ) DEALLOCATE( bwy )
IF ( ALLOCATED( buz ) ) DEALLOCATE( buz )
IF ( ALLOCATED( bvz ) ) DEALLOCATE( bvz )
IF ( ALLOCATED( bwz ) ) DEALLOCATE( bwz )
ALLOCATE ( bux(-mergp_x:mergp_x,nzb:nzt+1), &
buy(-mergp_y:mergp_y,nzb:nzt+1), &
buz(-mergp_z:mergp_z,nzb:nzt+1), &
bvx(-mergp_x:mergp_x,nzb:nzt+1), &
bvy(-mergp_y:mergp_y,nzb:nzt+1), &
bvz(-mergp_z:mergp_z,nzb:nzt+1), &
bwx(-mergp_x:mergp_x,nzb:nzt+1), &
bwy(-mergp_y:mergp_y,nzb:nzt+1), &
bwz(-mergp_z:mergp_z,nzb:nzt+1) )
!
!-- Create filter functions
CALL stg_filter_func( nux, bux, mergp_x ) !filter ux
CALL stg_filter_func( nuy, buy, mergp_y ) !filter uy
CALL stg_filter_func( nuz, buz, mergp_z ) !filter uz
CALL stg_filter_func( nvx, bvx, mergp_x ) !filter vx
CALL stg_filter_func( nvy, bvy, mergp_y ) !filter vy
CALL stg_filter_func( nvz, bvz, mergp_z ) !filter vz
CALL stg_filter_func( nwx, bwx, mergp_x ) !filter wx
CALL stg_filter_func( nwy, bwy, mergp_y ) !filter wy
CALL stg_filter_func( nwz, bwz, mergp_z ) !filter wz
END SUBROUTINE stg_setup_filter_function
!--------------------------------------------------------------------------------------------------!
! Description:
! ------------
!> Parin for &stg_par for synthetic turbulence generator
!--------------------------------------------------------------------------------------------------!
SUBROUTINE stg_parin
CHARACTER(LEN=100) :: line !< dummy string that contains the current line of the parameter file
INTEGER(iwp) :: io_status !< status after reading the namelist file
LOGICAL :: switch_off_module = .FALSE. !< local namelist parameter to switch off the module
!< although the respective module namelist appears in
!< the namelist file
NAMELIST /stg_par/ compute_velocity_seeds_local, &
dt_stg_adjust, &
dt_stg_call, &
switch_off_module, &
use_syn_turb_gen
!
!-- Move to the beginning of the namelist file and try to find and read the namelist.
REWIND( 11 )
READ( 11, stg_par, IOSTAT=io_status )
!
!-- Action depending on the READ status
IF ( io_status == 0 ) THEN
!
!-- stg_par namelist was found and read correctly. Set flag that indicates that the synthetic
!-- turbulence generator is switched on.
IF ( .NOT. switch_off_module ) syn_turb_gen = .TRUE.
ELSEIF ( io_status > 0 ) THEN
!
!-- stg_par namelist was found but contained errors. Print an error message including the line
!-- that caused the problem.
BACKSPACE( 11 )
READ( 11 , '(A)') line
CALL parin_fail_message( 'stg_par', line )
ENDIF
END SUBROUTINE stg_parin
!--------------------------------------------------------------------------------------------------!
! Description:
! ------------
!> Read module-specific global restart data (Fortran binary format).
!--------------------------------------------------------------------------------------------------!
SUBROUTINE stg_rrd_global_ftn( found )
LOGICAL, INTENT(OUT) :: found !< flag indicating if variable was found
found = .TRUE.
SELECT CASE ( restart_string(1:length) )
CASE ( 'time_stg_adjust' )
READ ( 13 ) time_stg_adjust
CASE ( 'time_stg_call' )
READ ( 13 ) time_stg_call
CASE ( 'use_syn_turb_gen' )
READ ( 13 ) use_syn_turb_gen
CASE DEFAULT
found = .FALSE.
END SELECT
END SUBROUTINE stg_rrd_global_ftn
!--------------------------------------------------------------------------------------------------!
! Description:
! ------------
!> Read module-specific global restart data (MPI-IO).
!--------------------------------------------------------------------------------------------------!
SUBROUTINE stg_rrd_global_mpi
CALL rrd_mpi_io( 'time_stg_adjust', time_stg_adjust )
CALL rrd_mpi_io( 'time_stg_call', time_stg_call )
CALL rrd_mpi_io( 'use_syn_turb_gen', use_syn_turb_gen )
END SUBROUTINE stg_rrd_global_mpi
!--------------------------------------------------------------------------------------------------!
! Description:
! ------------
!> This routine writes the respective restart data.
!--------------------------------------------------------------------------------------------------!
SUBROUTINE stg_wrd_global
IF ( TRIM( restart_data_format_output ) == 'fortran_binary' ) THEN
CALL wrd_write_string( 'time_stg_adjust' )
WRITE ( 14 ) time_stg_adjust
CALL wrd_write_string( 'time_stg_call' )
WRITE ( 14 ) time_stg_call
CALL wrd_write_string( 'use_syn_turb_gen' )
WRITE ( 14 ) use_syn_turb_gen
ELSEIF ( restart_data_format_output(1:3) == 'mpi' ) THEN
CALL wrd_mpi_io( 'time_stg_adjust', time_stg_adjust )
CALL wrd_mpi_io( 'time_stg_call', time_stg_call )
CALL wrd_mpi_io( 'use_syn_turb_gen', use_syn_turb_gen )
ENDIF
END SUBROUTINE stg_wrd_global
!--------------------------------------------------------------------------------------------------!
! Description:
! ------------
!> Create turbulent inflow fields for u, v, w with prescribed length scales and Reynolds stress
!> tensor after a method of Xie and Castro (2008), modified following suggestions of Kim et al.
!> (2013), and using a Lund rotation (Lund, 1998).
!--------------------------------------------------------------------------------------------------!
SUBROUTINE stg_main
USE exchange_horiz_mod, &
ONLY: exchange_horiz
INTEGER(iwp) :: i !< grid index in x-direction
INTEGER(iwp) :: j !< loop index in y-direction
INTEGER(iwp) :: k !< loop index in z-direction
LOGICAL :: stg_call = .FALSE. !< control flag indicating whether turbulence was updated or only restored from last call
REAL(wp) :: dt_stg !< time interval the STG is called
REAL(wp), DIMENSION(3) :: mc_factor_l !< local mass flux correction factor
IF ( debug_output_timestep ) CALL debug_message( 'stg_main', 'start' )
!
!-- Calculate time step which is needed for filter functions
dt_stg = MAX( dt_3d, dt_stg_call )
!
!-- Check if synthetic turbulence generator needs to be called and new perturbations need to be
!-- created or if old disturbances can be imposed again.
IF ( time_stg_call >= dt_stg_call .AND. &
intermediate_timestep_count == intermediate_timestep_count_max ) THEN
stg_call = .TRUE.
ELSE
stg_call = .FALSE.
ENDIF
!
!-- Initial value of fu, fv, fw
IF ( time_since_reference_point == 0.0_wp .AND. .NOT. velocity_seed_initialized ) THEN
!
!-- Generate turbulence at the left and right boundary. Random numbers for the yz-planes at the
!-- left/right boundary are generated by the left-sided mpi ranks only. After random numbers are
!-- calculated, they are distributed to all other mpi ranks in the model, so that the velocity
!-- seed matrices are available on all mpi ranks (i.e. also on the right-sided boundary mpi ranks).
!-- In case of offline nesting, this implies, that the left- and the right-sided lateral boundary
!-- have the same initial seeds.
!-- Note, in case of inflow from the right only, only turbulence at the left boundary is required.
IF ( .NOT. ( nesting_offline .OR. ( child_domain .AND. rans_mode_parent &
.AND. .NOT. rans_mode ) ) ) THEN
CALL stg_generate_seed_yz( nuy, nuz, buy, buz, fu_yz, id_stg_left )
CALL stg_generate_seed_yz( nvy, nvz, bvy, bvz, fv_yz, id_stg_left )
CALL stg_generate_seed_yz( nwy, nwz, bwy, bwz, fw_yz, id_stg_left )
ELSE
CALL stg_generate_seed_yz( nuy, nuz, buy, buz, fu_yz, id_stg_left, id_stg_right )
CALL stg_generate_seed_yz( nvy, nvz, bvy, bvz, fv_yz, id_stg_left, id_stg_right )
CALL stg_generate_seed_yz( nwy, nwz, bwy, bwz, fw_yz, id_stg_left, id_stg_right )
!
!-- Generate turbulence at the south and north boundary. Random numbers for the xz-planes at
!-- the south/north boundary are generated by the south-sided mpi ranks only. Please see also
!-- comment above.
CALL stg_generate_seed_xz( nux, nuz, bux, buz, fu_xz, id_stg_south, id_stg_north )
CALL stg_generate_seed_xz( nvx, nvz, bvx, bvz, fv_xz, id_stg_south, id_stg_north )
CALL stg_generate_seed_xz( nwx, nwz, bwx, bwz, fw_xz, id_stg_south, id_stg_north )
ENDIF
velocity_seed_initialized = .TRUE.
ENDIF
!
!-- New set of fu, fv, fw. Note, for inflow from the left side only, velocity seeds are only
!-- required at the left boundary, while in case of offline nesting or RANS-LES nesting velocity
!-- seeds are required also at the right, south and north boundaries.
IF ( stg_call ) THEN
IF ( .NOT. ( nesting_offline .OR. &
( child_domain .AND. rans_mode_parent &
.AND. .NOT. rans_mode ) ) ) THEN
CALL stg_generate_seed_yz( nuy, nuz, buy, buz, fuo_yz, id_stg_left )
CALL stg_generate_seed_yz( nvy, nvz, bvy, bvz, fvo_yz, id_stg_left )
CALL stg_generate_seed_yz( nwy, nwz, bwy, bwz, fwo_yz, id_stg_left )
ELSE
CALL stg_generate_seed_yz( nuy, nuz, buy, buz, fuo_yz, id_stg_left, id_stg_right )
CALL stg_generate_seed_yz( nvy, nvz, bvy, bvz, fvo_yz, id_stg_left, id_stg_right )
CALL stg_generate_seed_yz( nwy, nwz, bwy, bwz, fwo_yz, id_stg_left, id_stg_right )
CALL stg_generate_seed_xz( nux, nuz, bux, buz, fuo_xz, id_stg_south, id_stg_north )
CALL stg_generate_seed_xz( nvx, nvz, bvx, bvz, fvo_xz, id_stg_south, id_stg_north )
CALL stg_generate_seed_xz( nwx, nwz, bwx, bwz, fwo_xz, id_stg_south, id_stg_north )
ENDIF
ENDIF
!
!-- Turbulence generation at left and/or right boundary
IF ( myidx == id_stg_left .OR. myidx == id_stg_right ) THEN
!
!-- Calculate new set of perturbations. Do this only at last RK3-substep and when dt_stg_call is
!-- exceeded. Else the old set of perturbations is imposed
IF ( stg_call ) THEN
DO j = nys, nyn
DO k = nzb, nzt + 1
!
!-- Update fu, fv, fw following Eq. 14 of Xie and Castro (2008)
IF ( tu(k) == 0.0_wp .OR. adjustment_step ) THEN
fu_yz(k,j) = fuo_yz(k,j)
ELSE
fu_yz(k,j) = fu_yz(k,j) * EXP( -pi * dt_stg * 0.5_wp / tu(k) ) + &
fuo_yz(k,j) * SQRT( 1.0_wp - EXP( -pi * dt_stg / tu(k) ) )
ENDIF
IF ( tv(k) == 0.0_wp .OR. adjustment_step ) THEN
fv_yz(k,j) = fvo_yz(k,j)
ELSE
fv_yz(k,j) = fv_yz(k,j) * EXP( -pi * dt_stg * 0.5_wp / tv(k) ) + &
fvo_yz(k,j) * SQRT( 1.0_wp - EXP( -pi * dt_stg / tv(k) ) )
ENDIF
IF ( tw(k) == 0.0_wp .OR. adjustment_step ) THEN
fw_yz(k,j) = fwo_yz(k,j)
ELSE
fw_yz(k,j) = fw_yz(k,j) * EXP( -pi * dt_stg * 0.5_wp / tw(k) ) + &
fwo_yz(k,j) * SQRT( 1.0_wp - EXP( -pi * dt_stg / tw(k) ) )
ENDIF
ENDDO
ENDDO
dist_yz(nzb,:,1) = 0.0_wp
dist_yz(nzb,:,2) = 0.0_wp
dist_yz(nzb,:,3) = 0.0_wp
IF ( myidx == id_stg_left ) i = nxl
IF ( myidx == id_stg_right ) i = nxr+1
DO j = nys, nyn
DO k = nzb+1, nzt + 1
!
!-- Lund rotation following Eq. 17 in Xie and Castro (2008).
dist_yz(k,j,1) = a11(k) * fu_yz(k,j) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(k,j,i), 1 ) )
ENDDO
ENDDO
IF ( myidx == id_stg_left ) i = nxl-1
IF ( myidx == id_stg_right ) i = nxr+1
DO j = nys, nyn
DO k = nzb+1, nzt + 1
!
!-- Lund rotation following Eq. 17 in Xie and Castro (2008).
!-- Additional factors are added to improve the variance of v and w experimental test
!-- of 1.2
dist_yz(k,j,2) = ( a21(k) * fu_yz(k,j) + a22(k) * fv_yz(k,j) ) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(k,j,i), 2 ) )
dist_yz(k,j,3) = ( a31(k) * fu_yz(k,j) + a32(k) * fv_yz(k,j) + &
a33(k) * fw_yz(k,j) ) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(k,j,i), 3 ) )
ENDDO
ENDDO
ENDIF
!
!-- Mass flux correction following Kim et al. (2013)
!-- This correction factor ensures that the mass flux is preserved at the inflow boundary. First,
!-- calculate mean value of the imposed perturbations at yz boundary. Note, this needs to be done
!-- only after the last RK3-substep, else the boundary values won't be accessed.
IF ( intermediate_timestep_count == intermediate_timestep_count_max ) THEN
mc_factor_l = 0.0_wp
mc_factor = 0.0_wp
!
!-- Sum up the original volume flows (with and without perturbations).
!-- Note, for non-normal components (here v and w) it is actually no volume flow.
IF ( myidx == id_stg_left ) i = nxl
IF ( myidx == id_stg_right ) i = nxr+1
mc_factor_l(1) = SUM( dist_yz(nzb:nzt,nys:nyn,1) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(nzb:nzt,nys:nyn,i), 1 ) ) )
IF ( myidx == id_stg_left ) i = nxl-1
IF ( myidx == id_stg_right ) i = nxr+1
mc_factor_l(2) = SUM( dist_yz(nzb:nzt,nysv:nyn,2) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(nzb:nzt,nysv:nyn,i), 2 ) ) )
mc_factor_l(3) = SUM( dist_yz(nzb:nzt,nys:nyn,3) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(nzb:nzt,nys:nyn,i), 3 ) ) )
#if defined( __parallel )
CALL MPI_ALLREDUCE( mc_factor_l, mc_factor, 3, MPI_REAL, MPI_SUM, comm1dy, ierr )
#else
mc_factor = mc_factor_l
#endif
!
!-- Calculate correction factor and force zero mean perturbations.
mc_factor = mc_factor / REAL( nr_non_topo_yz, KIND = wp )
IF ( myidx == id_stg_left ) i = nxl
IF ( myidx == id_stg_right ) i = nxr+1
dist_yz(:,nys:nyn,1) = ( dist_yz(:,nys:nyn,1) - mc_factor(1) ) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(:,nys:nyn,i), 1 ) )
IF ( myidx == id_stg_left ) i = nxl-1
IF ( myidx == id_stg_right ) i = nxr+1
dist_yz(:,nys:nyn,2) = ( dist_yz(:,nys:nyn,2) - mc_factor(2) ) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(:,nys:nyn,i), 2 ) )
dist_yz(:,nys:nyn,3) = ( dist_yz(:,nys:nyn,3) - mc_factor(3) ) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(:,nys:nyn,i), 3 ) )
!
!-- Add disturbances
IF ( myidx == id_stg_left ) THEN
!
!-- For the left boundary distinguish between mesoscale offline / self nesting and
!-- turbulent inflow at the left boundary only. In the latter case turbulence is imposed on
!-- the mean inflow profiles.
IF ( .NOT. nesting_offline .AND. .NOT. child_domain ) THEN
!
!-- Add disturbance at the inflow
DO j = nys, nyn
DO k = nzb, nzt+1
u(k,j,-nbgp+1:0) = ( mean_inflow_profiles(k,1) + dist_yz(k,j,1) ) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(k,j,0), 1 ) )
v(k,j,-nbgp:-1) = ( mean_inflow_profiles(k,2) + dist_yz(k,j,2) ) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(k,j,-1), 2 ) )
w(k,j,-nbgp:-1) = dist_yz(k,j,3) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(k,j,-1), 3 ) )
ENDDO
ENDDO
ELSE
DO j = nys, nyn
DO k = nzb+1, nzt
u(k,j,0) = ( u(k,j,0) + dist_yz(k,j,1) ) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(k,j,0), 1 ) )
u(k,j,-1) = u(k,j,0)
v(k,j,-1) = ( v(k,j,-1) + dist_yz(k,j,2) ) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(k,j,-1), 2 ) )
w(k,j,-1) = ( w(k,j,-1) + dist_yz(k,j,3) ) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(k,j,-1), 3 ) )
ENDDO
ENDDO
ENDIF
ENDIF
IF ( myidx == id_stg_right ) THEN
DO j = nys, nyn
DO k = nzb+1, nzt
u(k,j,nxr+1) = ( u(k,j,nxr+1) + dist_yz(k,j,1) ) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(k,j,nxr+1), 1 ) )
v(k,j,nxr+1) = ( v(k,j,nxr+1) + dist_yz(k,j,2) ) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(k,j,nxr+1), 2 ) )
w(k,j,nxr+1) = ( w(k,j,nxr+1) + dist_yz(k,j,3) ) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(k,j,nxr+1), 3 ) )
ENDDO
ENDDO
ENDIF
ENDIF
ENDIF
!
!-- Turbulence generation at north and south boundary
IF ( myidy == id_stg_north .OR. myidy == id_stg_south ) THEN
!
!-- Calculate new set of perturbations. Do this only at last RK3-substep and when dt_stg_call is
!-- exceeded. Else the old set of perturbations is imposed
IF ( stg_call ) THEN
DO i = nxl, nxr
DO k = nzb, nzt + 1
!
!-- Update fu, fv, fw following Eq. 14 of Xie and Castro (2008)
IF ( tu(k) == 0.0_wp .OR. adjustment_step ) THEN
fu_xz(k,i) = fuo_xz(k,i)
ELSE
fu_xz(k,i) = fu_xz(k,i) * EXP( -pi * dt_stg * 0.5_wp / tu(k) ) + &
fuo_xz(k,i) * SQRT( 1.0_wp - EXP( -pi * dt_stg / tu(k) ) )
ENDIF
IF ( tv(k) == 0.0_wp .OR. adjustment_step ) THEN
fv_xz(k,i) = fvo_xz(k,i)
ELSE
fv_xz(k,i) = fv_xz(k,i) * EXP( -pi * dt_stg * 0.5_wp / tv(k) ) + &
fvo_xz(k,i) * SQRT( 1.0_wp - EXP( -pi * dt_stg / tv(k) ) )
ENDIF
IF ( tw(k) == 0.0_wp .OR. adjustment_step ) THEN
fw_xz(k,i) = fwo_xz(k,i)
ELSE
fw_xz(k,i) = fw_xz(k,i) * EXP( -pi * dt_stg * 0.5_wp / tw(k) ) + &
fwo_xz(k,i) * SQRT( 1.0_wp - EXP( -pi * dt_stg / tw(k) ) )
ENDIF
ENDDO
ENDDO
dist_xz(nzb,:,1) = 0.0_wp
dist_xz(nzb,:,2) = 0.0_wp
dist_xz(nzb,:,3) = 0.0_wp
IF ( myidy == id_stg_south ) j = nys
IF ( myidy == id_stg_north ) j = nyn+1
DO i = nxl, nxr
DO k = nzb+1, nzt + 1
!
!-- Lund rotation following Eq. 17 in Xie and Castro (2008).
!-- Additional factors are added to improve the variance of v and w
!experimental test of 1.2
dist_xz(k,i,2) = ( a21(k) * fu_xz(k,i) + a22(k) * fv_xz(k,i) ) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(k,j,i), 2 ) )
ENDDO
ENDDO
IF ( myidy == id_stg_south ) j = nys-1
IF ( myidy == id_stg_north ) j = nyn+1
DO i = nxl, nxr
DO k = nzb+1, nzt + 1
!
!-- Lund rotation following Eq. 17 in Xie and Castro (2008).
!-- Additional factors are added to improve the variance of v and w
dist_xz(k,i,1) = a11(k) * fu_xz(k,i) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(k,j,i), 1 ) )
dist_xz(k,i,3) = ( a31(k) * fu_xz(k,i) + a32(k) * fv_xz(k,i) + &
a33(k) * fw_xz(k,i) ) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(k,j,i), 3 ) )
ENDDO
ENDDO
ENDIF
!
!-- Mass flux correction following Kim et al. (2013). This correction factor ensures that the
!-- mass flux is preserved at the inflow boundary. First, calculate mean value of the imposed
!-- perturbations at yz boundary. Note, this needs to be done only after the last RK3-substep,
!-- else the boundary values won't be accessed.
IF ( intermediate_timestep_count == intermediate_timestep_count_max ) THEN
mc_factor_l = 0.0_wp
mc_factor = 0.0_wp
IF ( myidy == id_stg_south ) j = nys
IF ( myidy == id_stg_north ) j = nyn+1
mc_factor_l(2) = SUM( dist_xz(nzb:nzt,nxl:nxr,2) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(nzb:nzt,j,nxl:nxr), 2 ) ) )
IF ( myidy == id_stg_south ) j = nys-1
IF ( myidy == id_stg_north ) j = nyn+1
mc_factor_l(1) = SUM( dist_xz(nzb:nzt,nxlu:nxr,1) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(nzb:nzt,j,nxlu:nxr), 1 ) ) )
mc_factor_l(3) = SUM( dist_xz(nzb:nzt,nxl:nxr,3) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(nzb:nzt,j,nxl:nxr), 3 ) ) )
#if defined( __parallel )
CALL MPI_ALLREDUCE( mc_factor_l, mc_factor, 3, MPI_REAL, MPI_SUM, comm1dx, ierr )
#else
mc_factor = mc_factor_l
#endif
mc_factor = mc_factor / REAL( nr_non_topo_xz, KIND = wp )
IF ( myidy == id_stg_south ) j = nys
IF ( myidy == id_stg_north ) j = nyn+1
dist_xz(:,nxl:nxr,2) = ( dist_xz(:,nxl:nxr,2) - mc_factor(2) ) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(:,j,nxl:nxr), 2 ) )
IF ( myidy == id_stg_south ) j = nys-1
IF ( myidy == id_stg_north ) j = nyn+1
dist_xz(:,nxl:nxr,1) = ( dist_xz(:,nxl:nxr,1) - mc_factor(1) ) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(:,j,nxl:nxr), 1 ) )
dist_xz(:,nxl:nxr,3) = ( dist_xz(:,nxl:nxr,3) - mc_factor(3) ) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(:,j,nxl:nxr), 3 ) )
!
!-- Add disturbances
IF ( myidy == id_stg_south ) THEN
DO i = nxl, nxr
DO k = nzb+1, nzt
u(k,-1,i) = ( u(k,-1,i) + dist_xz(k,i,1) ) &
* MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(k,-1,i), 1 ) )
v(k,0,i) = ( v(k,0,i) + dist_xz(k,i,2) ) &
* MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(k,0,i), 2 ) )
v(k,-1,i) = v(k,0,i)
w(k,-1,i) = ( w(k,-1,i) + dist_xz(k,i,3) ) &
* MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(k,-1,i), 3 ) )
ENDDO
ENDDO
ENDIF
IF ( myidy == id_stg_north ) THEN
DO i = nxl, nxr
DO k = nzb+1, nzt
u(k,nyn+1,i) = ( u(k,nyn+1,i) + dist_xz(k,i,1) ) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(k,nyn+1,i), 1 ) )
v(k,nyn+1,i) = ( v(k,nyn+1,i) + dist_xz(k,i,2) ) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(k,nyn+1,i), 2 ) )
w(k,nyn+1,i) = ( w(k,nyn+1,i) + dist_xz(k,i,3) ) * &
MERGE( 1.0_wp, 0.0_wp, BTEST( wall_flags_total_0(k,nyn+1,i), 3 ) )
ENDDO
ENDDO
ENDIF
ENDIF
ENDIF
!
!-- Exchange ghost points.
CALL exchange_horiz( u, nbgp )
CALL exchange_horiz( v, nbgp )
CALL exchange_horiz( w, nbgp )
!
!-- Finally, set time counter for calling STG to zero
IF ( stg_call ) time_stg_call = 0.0_wp
!
!-- Set adjustment step to False to indicate that time correlation can be
!-- switched-on again.
adjustment_step = .FALSE.
IF ( debug_output_timestep ) CALL debug_message( 'stg_main', 'end' )
END SUBROUTINE stg_main
!--------------------------------------------------------------------------------------------------!
! Description:
! ------------
!> Generate a set of random number rand_it wich is equal on each PE and calculate the velocity seed
!> f_n. f_n is splitted in vertical direction by the number of PEs in x-direction and each PE
!> calculates a vertical subsection of f_n. At the the end, all parts are collected to form the full
!> array.
!--------------------------------------------------------------------------------------------------!
SUBROUTINE stg_generate_seed_yz( n_y, n_z, b_y, b_z, f_n, id_left, id_right )
INTEGER(iwp) :: id_left !< core ids at respective boundaries
INTEGER(iwp), OPTIONAL :: id_right !< core ids at respective boundaries
INTEGER(iwp) :: j !< loop index in y-direction
INTEGER(iwp) :: jj !< loop index in y-direction
INTEGER(iwp) :: k !< loop index in z-direction
INTEGER(iwp) :: kk !< loop index in z-direction
INTEGER(iwp) :: send_count !< send count for MPI_GATHERV
INTEGER(iwp), DIMENSION(nzb:nzt+1) :: n_y !< length scale in y-direction
INTEGER(iwp), DIMENSION(nzb:nzt+1) :: n_z !< length scale in z-direction
REAL(wp) :: nyz_inv !< inverse of number of grid points in yz-slice
REAL(wp) :: rand_av !< average of random number
REAL(wp) :: rand_sigma_inv !< inverse of stdev of random number
REAL(wp), DIMENSION(-mergp_y:mergp_y,nzb:nzt+1) :: b_y !< filter function in y-direction
REAL(wp), DIMENSION(-mergp_z:mergp_z,nzb:nzt+1) :: b_z !< filter function in z-direction
REAL(wp), DIMENSION(nzb_x_stg:nzt_x_stg+1,nys:nyn) :: f_n_l !< local velocity seed
REAL(wp), DIMENSION(nzb:nzt+1,nys:nyn) :: f_n !< velocity seed
REAL(wp), DIMENSION(:,:), ALLOCATABLE :: rand_it !< global array of random numbers
!
!-- Generate random numbers using the parallel random generator. The set of random numbers are
!-- modified to have an average of 0 and unit variance. Note, at the end the random number array
!-- must be defined globally in order to compute the correlation matrices. However, the calculation
!-- and normalization of random numbers is done locally, while the result is later distributed to
!-- all processes. Further, please note, a set of random numbers is only calculated for the west
!-- boundary, while the east boundary uses the same random numbers and thus also computes the same
!-- correlation matrix.
ALLOCATE( rand_it(nzb-mergp_z:nzt+1+mergp_z,-mergp_y+nys:nyn+mergp_y) )
rand_it = 0.0_wp
rand_av = 0.0_wp
rand_sigma_inv = 0.0_wp
nyz_inv = 1.0_wp / REAL( ( nzt + 1 + mergp_z - ( nzb - mergp_z ) + 1 ) &
* ( ny + mergp_y - ( 0 - mergp_y ) + 1 ), KIND = wp )
!
!-- Compute and normalize random numbers.
DO j = nys - mergp_y, nyn + mergp_y
!
!-- Put the random seeds at grid point j
CALL random_seed_parallel( put=seq_rand_yz(:,j) )
DO k = nzb - mergp_z, nzt + 1 + mergp_z
CALL random_number_parallel( random_dummy )
rand_it(k,j) = random_dummy
ENDDO
!
!-- Get the new random seeds from last call at grid point j
CALL random_seed_parallel( get=seq_rand_yz(:,j) )
ENDDO
!
!-- For normalization to zero mean, sum-up the global random numers. To normalize the global set of
!-- random numbers, the inner ghost layers mergp must not be summed-up, else the random numbers on
!-- the ghost layers will be stronger weighted as they also occur on the inner subdomains.
DO j = MERGE( nys, nys - mergp_y, nys /= 0 ), MERGE( nyn, nyn + mergp_y, nyn /= ny )
DO k = nzb - mergp_z, nzt + 1 + mergp_z
rand_av = rand_av + rand_it(k,j)
ENDDO
ENDDO
#if defined( __parallel )
!
!-- Sum-up the local averages of the random numbers
CALL MPI_ALLREDUCE( MPI_IN_PLACE, rand_av, 1, MPI_REAL, MPI_SUM, comm1dy, ierr )
#endif
rand_av = rand_av * nyz_inv
!
!-- Obtain zero mean
rand_it= rand_it - rand_av
!
!-- Now, compute the variance
DO j = MERGE( nys, nys - mergp_y, nys /= 0 ), MERGE( nyn, nyn + mergp_y, nyn /= ny )
DO k = nzb - mergp_z, nzt + 1 + mergp_z
rand_sigma_inv = rand_sigma_inv + rand_it(k,j)**2
ENDDO
ENDDO
#if defined( __parallel )
!
!-- Sum-up the local quadratic averages of the random numbers
CALL MPI_ALLREDUCE( MPI_IN_PLACE, rand_sigma_inv, 1, MPI_REAL, MPI_SUM, comm1dy, ierr )
#endif
!
!-- Compute standard deviation
IF ( rand_sigma_inv /= 0.0_wp ) THEN
rand_sigma_inv = 1.0_wp / SQRT( rand_sigma_inv * nyz_inv )
ELSE
rand_sigma_inv = 1.0_wp
ENDIF
!
!-- Normalize with standard deviation to obtain unit variance
rand_it = rand_it * rand_sigma_inv
CALL cpu_log( log_point_s(31), 'STG f_n factors', 'start' )
!
!-- Generate velocity seed following Eq.6 of Xie and Castro (2008). There are two options. In the
!-- first one, the computation of the seeds is distributed to all processes along the communicator
!-- comm1dy and gathered on the leftmost and, if necessary, on the rightmost process. For huge
!-- length scales the computational effort can become quite huge (it scales with the turbulent
!-- length scales), so that gain by parallelization exceeds the costs by the subsequent
!-- communication. In the second option, which performs better when the turbulent length scales
!-- are parametrized and thus the loops are smaller, the seeds are computed locally and no
!-- communication is necessary.
IF ( compute_velocity_seeds_local ) THEN
f_n = 0.0_wp
DO j = nys, nyn
DO k = nzb, nzt+1
DO jj = -n_y(k), n_y(k)
DO kk = -n_z(k), n_z(k)
f_n(k,j) = f_n(k,j) + b_y(jj,k) * b_z(kk,k) * rand_it(k+kk,j+jj)
ENDDO
ENDDO
ENDDO
ENDDO
ELSE
f_n_l = 0.0_wp
DO j = nys, nyn
DO k = nzb_x_stg, nzt_x_stg+1
DO jj = -n_y(k), n_y(k)
DO kk = -n_z(k), n_z(k)
f_n_l(k,j) = f_n_l(k,j) + b_y(jj,k) * b_z(kk,k) * rand_it(k+kk,j+jj)
ENDDO
ENDDO
ENDDO
ENDDO
!
!-- Gather velocity seeds of full subdomain
send_count = nzt_x_stg - nzb_x_stg + 1
IF ( nzt_x_stg == nzt ) send_count = send_count + 1
#if defined( __parallel )
!
!-- Gather the velocity seed matrix on left boundary mpi ranks.
CALL MPI_GATHERV( f_n_l(nzb_x_stg,nys), send_count, stg_type_yz_small, f_n(nzb+1,nys), &
recv_count_yz, displs_yz, stg_type_yz, id_left, comm1dx, ierr )
!
!-- If required, gather the same velocity seed matrix on right boundary mpi ranks (in offline
!-- nesting for example).
IF ( PRESENT( id_right ) ) THEN
CALL MPI_GATHERV( f_n_l(nzb_x_stg,nys), send_count, stg_type_yz_small, f_n(nzb+1,nys), &
recv_count_yz, displs_yz, stg_type_yz, id_right, comm1dx, ierr )
ENDIF
#else
f_n(nzb+1:nzt+1,nys:nyn) = f_n_l(nzb_x_stg:nzt_x_stg+1,nys:nyn)
!
!-- Next line required to avoid compile errors because of unused dummy arguments
IF ( id_left == 0 ) id_right = 0
#endif
ENDIF
DEALLOCATE( rand_it )
CALL cpu_log( log_point_s(31), 'STG f_n factors', 'stop' )
END SUBROUTINE stg_generate_seed_yz
!--------------------------------------------------------------------------------------------------!
! Description:
! ------------
!> Generate a set of random number rand_it wich is equal on each PE and calculate the velocity seed
!> f_n.
!> f_n is splitted in vertical direction by the number of PEs in y-direction and and each PE
!> calculates a vertical subsection of f_n. At the the end, all parts are collected to form the
!> full array.
!--------------------------------------------------------------------------------------------------!
SUBROUTINE stg_generate_seed_xz( n_x, n_z, b_x, b_z, f_n, id_south, id_north )
INTEGER(iwp) :: i !< loop index in x-direction
INTEGER(iwp) :: id_north !< core ids at respective boundaries
INTEGER(iwp) :: id_south !< core ids at respective boundaries
INTEGER(iwp) :: ii !< loop index in x-direction
INTEGER(iwp) :: k !< loop index in z-direction
INTEGER(iwp) :: kk !< loop index in z-direction
INTEGER(iwp) :: send_count !< send count for MPI_GATHERV
INTEGER(iwp), DIMENSION(nzb:nzt+1) :: n_x !< length scale in x-direction
INTEGER(iwp), DIMENSION(nzb:nzt+1) :: n_z !< length scale in z-direction
REAL(wp) :: nxz_inv !< inverse of number of grid points in xz-slice
REAL(wp) :: rand_av !< average of random number
REAL(wp) :: rand_sigma_inv !< inverse of stdev of random number
REAL(wp), DIMENSION(-mergp_x:mergp_x,nzb:nzt+1) :: b_x !< filter function in x-direction
REAL(wp), DIMENSION(-mergp_z:mergp_z,nzb:nzt+1) :: b_z !< filter function in z-direction
REAL(wp), DIMENSION(nzb_y_stg:nzt_y_stg+1,nxl:nxr) :: f_n_l !< local velocity seed
REAL(wp), DIMENSION(nzb:nzt+1,nxl:nxr) :: f_n !< velocity seed
REAL(wp), DIMENSION(:,:), ALLOCATABLE :: rand_it !< global array of random numbers
!
!-- Generate random numbers using the parallel random generator. The set of random numbers are
!-- modified to have an average of 0 and unit variance. Note, at the end the random number array
!-- must be defined globally in order to compute the correlation matrices. However, the calculation
!-- and normalization of random numbers is done locally, while the result is later distributed to
!-- all processes. Further, please note, a set of random numbers is only calculated for the south
!-- boundary, while the north boundary uses the same random numbers and thus also computes the same
!-- correlation matrix.
ALLOCATE( rand_it(nzb-mergp_z:nzt+1+mergp_z,-mergp_x+nxl:nxr+mergp_x) )
rand_it = 0.0_wp
rand_av = 0.0_wp
rand_sigma_inv = 0.0_wp
nxz_inv = 1.0_wp / REAL( ( nzt + 1 + mergp_z - ( nzb - mergp_z ) + 1 ) &
* ( nx + mergp_x - ( 0 - mergp_x ) +1 ), KIND = wp )
!
!-- Compute and normalize random numbers.
DO i = nxl - mergp_x, nxr + mergp_x
!
!-- Put the random seeds at grid point ii
CALL random_seed_parallel( put=seq_rand_xz(:,i) )
DO k = nzb - mergp_z, nzt + 1 + mergp_z
CALL random_number_parallel( random_dummy )
rand_it(k,i) = random_dummy
ENDDO
!
!-- Get the new random seeds from last call at grid point ii
CALL random_seed_parallel( get=seq_rand_xz(:,i) )
ENDDO
!
!-- For normalization to zero mean, sum-up the global random numers.
!-- To normalize the global set of random numbers, the inner ghost layers mergp must not be
!-- summed-up, else the random numbers on the ghost layers will be stronger weighted as they
!-- also occur on the inner subdomains.
DO i = MERGE( nxl, nxl - mergp_x, nxl /= 0 ), MERGE( nxr, nxr + mergp_x, nxr /= nx )
DO k = nzb - mergp_z, nzt + 1 + mergp_z
rand_av = rand_av + rand_it(k,i)
ENDDO
ENDDO
#if defined( __parallel )
!
!-- Sum-up the local averages of the random numbers
CALL MPI_ALLREDUCE( MPI_IN_PLACE, rand_av, 1, MPI_REAL, MPI_SUM, comm1dx, ierr )
#endif
rand_av = rand_av * nxz_inv
!
!-- Obtain zero mean
rand_it= rand_it - rand_av
!
!-- Now, compute the variance
DO i = MERGE( nxl, nxl - mergp_x, nxl /= 0 ), MERGE( nxr, nxr + mergp_x, nxr /= nx )
DO k = nzb - mergp_z, nzt + 1 + mergp_z
rand_sigma_inv = rand_sigma_inv + rand_it(k,i)**2
ENDDO
ENDDO
#if defined( __parallel )
!
!-- Sum-up the local quadratic averages of the random numbers
CALL MPI_ALLREDUCE( MPI_IN_PLACE, rand_sigma_inv, 1, MPI_REAL, MPI_SUM, comm1dx, ierr )
#endif
!
!-- Compute standard deviation
IF ( rand_sigma_inv /= 0.0_wp ) THEN
rand_sigma_inv = 1.0_wp / SQRT( rand_sigma_inv * nxz_inv )
ELSE
rand_sigma_inv = 1.0_wp
ENDIF
!
!-- Normalize with standard deviation to obtain unit variance
rand_it = rand_it * rand_sigma_inv
CALL cpu_log( log_point_s(31), 'STG f_n factors', 'start' )
!
!-- Generate velocity seed following Eq.6 of Xie and Castro (2008). There are two options. In the
!-- first one, the computation of the seeds is distributed to all processes along the communicator
!-- comm1dx and gathered on the southmost and, if necessary, on the northmost process. For huge
!-- length scales the computational effort can become quite huge (it scales with the turbulent
!-- length scales), so that gain by parallelization exceeds the costs by the subsequent communication.
!-- In the second option, which performs better when the turbulent length scales are parametrized
!-- and thus the loops are smaller, the seeds are computed locally and no communication is necessary.
IF ( compute_velocity_seeds_local ) THEN
f_n = 0.0_wp
DO i = nxl, nxr
DO k = nzb, nzt+1
DO ii = -n_x(k), n_x(k)
DO kk = -n_z(k), n_z(k)
f_n(k,i) = f_n(k,i) + b_x(ii,k) * b_z(kk,k) * rand_it(k+kk,i+ii)
ENDDO
ENDDO
ENDDO
ENDDO
ELSE
f_n_l = 0.0_wp
DO i = nxl, nxr
DO k = nzb_y_stg, nzt_y_stg+1
DO ii = -n_x(k), n_x(k)
DO kk = -n_z(k), n_z(k)
f_n_l(k,i) = f_n_l(k,i) + b_x(ii,k) * b_z(kk,k) * rand_it(k+kk,i+ii)
ENDDO
ENDDO
ENDDO
ENDDO
!
!-- Gather velocity seeds of full subdomain
send_count = nzt_y_stg - nzb_y_stg + 1
IF ( nzt_y_stg == nzt ) send_count = send_count + 1
#if defined( __parallel )
!
!-- Gather the processed velocity seed on south boundary mpi ranks.
CALL MPI_GATHERV( f_n_l(nzb_y_stg,nxl), send_count, stg_type_xz_small, f_n(nzb+1,nxl), &
recv_count_xz, displs_xz, stg_type_xz, id_south, comm1dy, ierr )
!
!-- Gather the processed velocity seed on north boundary mpi ranks.
CALL MPI_GATHERV( f_n_l(nzb_y_stg,nxl), send_count, stg_type_xz_small, f_n(nzb+1,nxl), &
recv_count_xz, displs_xz, stg_type_xz, id_north, comm1dy, ierr )
#else
f_n(nzb+1:nzt+1,nxl:nxr) = f_n_l(nzb_y_stg:nzt_y_stg+1,nxl:nxr)
!
!-- Next line required to avoid compile errors because of unused dummy arguments
IF ( id_north == 0 ) id_south = 0
#endif
ENDIF
DEALLOCATE( rand_it )
CALL cpu_log( log_point_s(31), 'STG f_n factors', 'stop' )
END SUBROUTINE stg_generate_seed_xz
!--------------------------------------------------------------------------------------------------!
! Description:
! ------------
!> Parametrization of the Reynolds-stress componentes, turbulent length- and time scales. The
!> parametrization follows Brost et al. (1982) with modifications described in Rotach et al. (1996),
!> which is applied in state-of-the-art dispserion modelling.
!--------------------------------------------------------------------------------------------------!
SUBROUTINE parametrize_turbulence
INTEGER(iwp) :: k !< loop index in z-direction
REAL(wp) :: corr_term_uh !< correction term in parametrization of horizontal variances for unstable stratification
REAL(wp) :: d_zi !< inverse boundary-layer depth
REAL(wp) :: length_scale_lon !< length scale in flow direction
REAL(wp) :: length_scale_lon_zi !< length scale in flow direction at boundary-layer top
REAL(wp) :: length_scale_lat !< length scale in crosswind direction
REAL(wp) :: length_scale_lat_zi !< length scale in crosswind direction at boundary-layer top
REAL(wp) :: length_scale_vert !< length scale in vertical direction
REAL(wp) :: length_scale_vert_zi !< length scale in vertical direction at boundary-layer top
REAL(wp) :: time_scale_zi !< time scale at boundary-layer top
REAL(wp) :: r11_zi !< longitudinal variance at boundary-layer top
REAL(wp) :: r22_zi !< crosswind variance at boundary-layer top
REAL(wp) :: r33_zi !< vertical variance at boundary-layer top
REAL(wp) :: r31_zi !< r31 at boundary-layer top
REAL(wp) :: r32_zi !< r32 at boundary-layer top
REAL(wp) :: rlat1 !< first dummy argument for crosswind compoment of reynolds stress
REAL(wp) :: rlat2 !< second dummy argument for crosswind compoment of reynolds stress
REAL(wp) :: rlon1 !< first dummy argument for longitudinal compoment of reynolds stress
REAL(wp) :: rlon2 !< second dummy argument for longitudinal compoment of reynolds stress
REAL(wp) :: zzi !< ratio of z/zi
REAL(wp), DIMENSION(nzb+1:nzt+1) :: cos_phi !< cosine of angle between mean flow direction and x-axis
REAL(wp), DIMENSION(nzb+1:nzt+1) :: phi !< angle between mean flow direction and x-axis
REAL(wp), DIMENSION(nzb+1:nzt+1) :: sin_phi !< sine of angle between mean flow direction and x-axis
!
!-- Calculate the boundary-layer top index. Boundary-layer top is calculated by Richardson-bulk
!-- criterion according to Heinze et al. (2017).
k_zi = MAX( 1, MINLOC( ABS( zu - zi_ribulk ), DIM = 1 ) )
IF ( zu(k_zi) > zi_ribulk ) k_zi = k_zi - 1
k_zi = MIN( k_zi, nzt )
!
!-- Calculate angle between flow direction and x-axis.
DO k = nzb+1, nzt + 1
IF ( u_init(k) /= 0.0_wp ) THEN
phi(k) = ATAN( v_init(k) / u_init(k) )
ELSE
phi(k) = 0.5 * pi
ENDIF
!
!-- Pre-calculate sine and cosine.
cos_phi(k) = COS( phi(k) )
sin_phi(k) = SIN( phi(k) )
ENDDO
!
!-- Parametrize Reynolds-stress components. Please note, parametrization is formulated for the
!-- stream- and spanwise components, while stream- and spanwise direction do not necessarily
!-- coincide with the grid axis. Hence, components are rotated after computation.
d_zi = 1.0_wp / zi_ribulk
DO k = nzb+1, k_zi
corr_term_uh = MERGE( 0.35_wp * ABS( &
- zi_ribulk / ( kappa * scale_l - 10E-4_wp ) &
)**( 2.0_wp / 3.0_wp ), &
0.0_wp, &
scale_l < -5.0_wp )
!
!-- Determine normalized height coordinate
zzi = zu(k) * d_zi
!
!-- Compute longitudinal and crosswind component of reynolds stress. Rotate these components by
!-- the wind direction to map onto the u- and v-component. Note, since reynolds stress for
!-- variances cannot become negative, take the absolute values.
rlon1 = scale_us**2 * ( corr_term_uh + 5.0_wp - 4.0_wp * zzi )
rlat1 = scale_us**2 * ( corr_term_uh + 2.0_wp - zzi )
r11(k) = ABS( cos_phi(k) * rlon1 + sin_phi(k) * rlat1 )
r22(k) = ABS( - sin_phi(k) * rlon1 + cos_phi(k) * rlat1 )
!
!-- w'w'
r33(k) = scale_wm**2 * ( &
1.5_wp * zzi**( 2.0_wp / 3.0_wp ) * EXP( -2.0_wp * zzi ) &
+ ( 1.7_wp - zzi ) * ( scale_us / scale_wm )**2 &
)
!
!-- u'w' and v'w'. After calculation of the longitudinal and crosswind component
!-- these are projected along the x- and y-direction. Note, it is assumed that
!-- the flux within the boundary points opposite to the vertical gradient.
rlon2 = scale_us**2 * ( zzi - 1.0_wp )
rlat2 = scale_us**2 * ( 0.4 * zzi * ( 1.0_wp - zzi ) )
r31(k) = SIGN( ABS( cos_phi(k) * rlon2 + sin_phi(k) * rlat2 ), &
- ( u_init(k+1) - u_init(k-1) ) )
r32(k) = SIGN( ABS( - sin_phi(k) * rlon2 + cos_phi(k) * rlat2 ), &
- ( v_init(k+1) - v_init(k-1) ) )
!
!-- For u'v' no parametrization exist so far. For simplicity assume a similar profile as
!-- for the vertical transport.
r21(k) = 0.5_wp * ( r31(k) + r32(k) )
!
!-- Compute turbulent time scales according to Brost et al. (1982). Note, time scales are
!-- limited to the adjustment time scales.
tu(k) = MIN( dt_stg_adjust, &
3.33_wp * zzi * ( 1.0 - 0.67_wp * zzi ) / scale_wm * zi_ribulk )
tv(k) = tu(k)
tw(k) = tu(k)
length_scale_lon = MIN( SQRT( r11(k) ) * tu(k), zi_ribulk )
length_scale_lat = MIN( SQRT( r22(k) ) * tv(k), zi_ribulk )
length_scale_vert = MIN( SQRT( r33(k) ) * tw(k), zi_ribulk )
!
!-- Assume isotropic turbulence length scales
nux(k) = MAX( INT( length_scale_lon * ddx ), 1 )
nuy(k) = MAX( INT( length_scale_lat * ddy ), 1 )
nvx(k) = MAX( INT( length_scale_lon * ddx ), 1 )
nvy(k) = MAX( INT( length_scale_lat * ddy ), 1 )
nwx(k) = MAX( INT( length_scale_lon * ddx ), 1 )
nwy(k) = MAX( INT( length_scale_lat * ddy ), 1 )
nuz(k) = MAX( INT( length_scale_vert * ddzw(k) ), 1 )
nvz(k) = MAX( INT( length_scale_vert * ddzw(k) ), 1 )
nwz(k) = MAX( INT( length_scale_vert * ddzw(k) ), 1 )
ENDDO
!
!-- Above boundary-layer top length- and timescales as well as reynolds-stress components are
!-- reduced to zero by a blending function.
length_scale_lon_zi = SQRT( r11(k_zi) ) * tu(k_zi)
length_scale_lat_zi = SQRT( r22(k_zi) ) * tv(k_zi)
length_scale_vert_zi = SQRT( r33(k_zi) ) * tu(k_zi)
time_scale_zi = tu(k_zi)
r11_zi = r11(k_zi)
r22_zi = r22(k_zi)
r33_zi = r33(k_zi)
r31_zi = r31(k_zi)
r32_zi = r32(k_zi)
d_l = blend_coeff / MAX( length_scale_vert_zi, dx, dy, MINVAL( dzw ) )
DO k = k_zi+1, nzt+1
!
!-- Calculate function to gradually decrease Reynolds stress above ABL top.
blend = MIN( 1.0_wp, EXP( d_l * zu(k) - d_l * zi_ribulk ) )
!
!-- u'u' and v'v'. Assume isotropy. Note, add a small negative number to the denominator, else
!-- the mergpe-function can crash if scale_l is zero.
r11(k) = r11_zi * blend
r22(k) = r22_zi * blend
r33(k) = r33_zi * blend
r31(k) = r31_zi * blend
r32(k) = r32_zi * blend
r21(k) = 0.5_wp * ( r31(k) + r32(k) )
!
!-- Compute turbulent time scales according to Brost et al. (1982).
!-- Note, time scales are limited to the adjustment time scales.
tu(k) = time_scale_zi * blend
tv(k) = time_scale_zi * blend
tw(k) = time_scale_zi * blend
length_scale_lon = length_scale_lon_zi * blend
length_scale_lat = length_scale_lat_zi * blend
length_scale_vert = length_scale_vert_zi * blend
!
!-- Assume isotropic turbulence length scales
nux(k) = MAX( INT( length_scale_lon * ddx ), 1 )
nuy(k) = MAX( INT( length_scale_lat * ddy ), 1 )
nvx(k) = MAX( INT( length_scale_lon * ddx ), 1 )
nvy(k) = MAX( INT( length_scale_lat * ddy ), 1 )
nwx(k) = MAX( INT( length_scale_lon * ddx ), 1 )
nwy(k) = MAX( INT( length_scale_lat * ddy ), 1 )
nuz(k) = MAX( INT( length_scale_vert * ddzw(k) ), 1 )
nvz(k) = MAX( INT( length_scale_vert * ddzw(k) ), 1 )
nwz(k) = MAX( INT( length_scale_vert * ddzw(k) ), 1 )
ENDDO
!
!-- Set bottom boundary condition for reynolds stresses
r11(nzb) = r11(nzb+1)
r22(nzb) = r22(nzb+1)
r33(nzb) = r33(nzb+1)
r21(nzb) = r11(nzb+1)
r31(nzb) = r31(nzb+1)
r32(nzb) = r32(nzb+1)
!
!-- Set bottom boundary condition for the length and time scales
nux(nzb) = nux(nzb+1)
nuy(nzb) = nuy(nzb+1)
nuz(nzb) = nuz(nzb+1)
nvx(nzb) = nvx(nzb+1)
nvy(nzb) = nvy(nzb+1)
nvz(nzb) = nvz(nzb+1)
nwx(nzb) = nwx(nzb+1)
nwy(nzb) = nwy(nzb+1)
nwz(nzb) = nwz(nzb+1)
tu(nzb) = tu(nzb+1)
tv(nzb) = tv(nzb+1)
tw(nzb) = tw(nzb+1)
adjustment_step = .TRUE.
END SUBROUTINE parametrize_turbulence
!--------------------------------------------------------------------------------------------------!
! Description:
! ------------
!> Calculate the coefficient matrix from the Lund rotation.
!--------------------------------------------------------------------------------------------------!
SUBROUTINE calc_coeff_matrix
INTEGER(iwp) :: k !< loop index in z-direction
!
!-- Calculate coefficient matrix. Split loops to allow for loop vectorization.
DO k = nzb+1, nzt+1
IF ( r11(k) > 10E-6_wp ) THEN
a11(k) = SQRT( r11(k) )
a21(k) = r21(k) / a11(k)
a31(k) = r31(k) / a11(k)
ELSE
a11(k) = 10E-8_wp
a21(k) = 10E-8_wp
a31(k) = 10E-8_wp
ENDIF
ENDDO
DO k = nzb+1, nzt+1
a22(k) = r22(k) - a21(k)**2
IF ( a22(k) > 10E-6_wp ) THEN
a22(k) = SQRT( a22(k) )
a32(k) = ( r32(k) - a21(k) * a31(k) ) / a22(k)
ELSE
a22(k) = 10E-8_wp
a32(k) = 10E-8_wp
ENDIF
ENDDO
DO k = nzb+1, nzt+1
a33(k) = r33(k) - a31(k)**2 - a32(k)**2
IF ( a33(k) > 10E-6_wp ) THEN
a33(k) = SQRT( a33(k) )
ELSE
a33(k) = 10E-8_wp
ENDIF
ENDDO
!
!-- Set bottom boundary condition
a11(nzb) = a11(nzb+1)
a22(nzb) = a22(nzb+1)
a21(nzb) = a21(nzb+1)
a33(nzb) = a33(nzb+1)
a31(nzb) = a31(nzb+1)
a32(nzb) = a32(nzb+1)
END SUBROUTINE calc_coeff_matrix
!--------------------------------------------------------------------------------------------------!
! Description:
! ------------
!> This routine controls the re-adjustment of the turbulence statistics used for generating
!> turbulence at the lateral boundaries.
!--------------------------------------------------------------------------------------------------!
SUBROUTINE stg_adjust
IF ( debug_output_timestep ) CALL debug_message( 'stg_adjust', 'start' )
!
!-- In case of dirichlet inflow boundary conditions only at one lateral boundary, i.e. in the case
!-- no offline or self nesting is applied but synthetic turbulence shall be parametrized
!-- nevertheless, the boundary-layer depth need to determined first.
IF ( .NOT. nesting_offline .AND. .NOT. child_domain ) &
CALL nesting_offl_calc_zi
!
!-- Compute scaling parameters (domain-averaged), such as friction velocity are calculated.
CALL calc_scaling_variables
!
!-- Parametrize Reynolds-stress tensor. Parametrization follows Brost et al. (1982) with
!-- modifications described in Rotach et al. (1996) and is based on boundary-layer depth, friction
!-- velocity and velocity scale.
CALL parametrize_turbulence
!
!-- Calculate coefficient matrix from Reynolds stress tensor
!-- (Lund rotation)
CALL calc_coeff_matrix
!
!-- Setup filter functions according to updated length scales.
CALL stg_setup_filter_function
!
!-- Reset time counter for controlling next adjustment to zero
time_stg_adjust = 0.0_wp
IF ( debug_output_timestep ) CALL debug_message( 'stg_adjust', 'end' )
END SUBROUTINE stg_adjust
!--------------------------------------------------------------------------------------------------!
! Description:
! ------------
!> Calculate scaling variables which are used for turbulence parametrization according to Rotach
!> et al. (1996). Scaling variables are: friction velocity, boundary-layer depth, momentum velocity
!> scale, and Obukhov length.
!--------------------------------------------------------------------------------------------------!
SUBROUTINE calc_scaling_variables
INTEGER(iwp) :: i !< loop index in x-direction
INTEGER(iwp) :: j !< loop index in y-direction
INTEGER(iwp) :: k !< loop index in z-direction
INTEGER(iwp) :: m !< surface element index
REAL(wp) :: friction_vel_l !< mean friction veloctiy on subdomain
REAL(wp) :: pt_surf_mean !< mean near surface temperature (at 1st grid point)
REAL(wp) :: pt_surf_mean_l !< mean near surface temperature (at 1st grid point) on subdomain
REAL(wp) :: scale_l_l !< mean Obukhov lenght on subdomain
REAL(wp) :: shf_mean !< mean surface sensible heat flux
REAL(wp) :: shf_mean_l !< mean surface sensible heat flux on subdomain
REAL(wp) :: w_convective !< convective velocity scale
!
!-- Calculate mean friction velocity, velocity scale, heat flux and near-surface temperature in the
!-- model domain.
pt_surf_mean_l = 0.0_wp
shf_mean_l = 0.0_wp
scale_l_l = 0.0_wp
friction_vel_l = 0.0_wp
DO m = 1, surf_def_h(0)%ns
i = surf_def_h(0)%i(m)
j = surf_def_h(0)%j(m)
k = surf_def_h(0)%k(m)
friction_vel_l = friction_vel_l + surf_def_h(0)%us(m)
shf_mean_l = shf_mean_l + surf_def_h(0)%shf(m) * drho_air(k)
scale_l_l = scale_l_l + surf_def_h(0)%ol(m)
pt_surf_mean_l = pt_surf_mean_l + pt(k,j,i)
ENDDO
DO m = 1, surf_lsm_h(0)%ns
i = surf_lsm_h(0)%i(m)
j = surf_lsm_h(0)%j(m)
k = surf_lsm_h(0)%k(m)
friction_vel_l = friction_vel_l + surf_lsm_h(0)%us(m)
shf_mean_l = shf_mean_l + surf_lsm_h(0)%shf(m) * drho_air(k)
scale_l_l = scale_l_l + surf_lsm_h(0)%ol(m)
pt_surf_mean_l = pt_surf_mean_l + pt(k,j,i)
ENDDO
DO m = 1, surf_usm_h(0)%ns
i = surf_usm_h(0)%i(m)
j = surf_usm_h(0)%j(m)
k = surf_usm_h(0)%k(m)
friction_vel_l = friction_vel_l + surf_usm_h(0)%us(m)
shf_mean_l = shf_mean_l + surf_usm_h(0)%shf(m) * drho_air(k)
scale_l_l = scale_l_l + surf_usm_h(0)%ol(m)
pt_surf_mean_l = pt_surf_mean_l + pt(k,j,i)
ENDDO
#if defined( __parallel )
CALL MPI_ALLREDUCE( friction_vel_l, scale_us, 1, MPI_REAL, MPI_SUM, comm2d, ierr )
CALL MPI_ALLREDUCE( shf_mean_l, shf_mean, 1, MPI_REAL, MPI_SUM, comm2d, ierr )
CALL MPI_ALLREDUCE( scale_l_l, scale_l, 1, MPI_REAL, MPI_SUM, comm2d, ierr )
CALL MPI_ALLREDUCE( pt_surf_mean_l, pt_surf_mean, 1, MPI_REAL, MPI_SUM, comm2d, ierr )
#else
scale_us = friction_vel_l
shf_mean = shf_mean_l
scale_l = scale_l_l
pt_surf_mean = pt_surf_mean_l
#endif
scale_us = scale_us * d_nxy
shf_mean = shf_mean * d_nxy
scale_l = scale_l * d_nxy
pt_surf_mean = pt_surf_mean * d_nxy
!
!-- Compute mean convective velocity scale. Note, in case the mean heat flux is negative, set
!-- convective velocity scale to zero.
IF ( shf_mean > 0.0_wp ) THEN
w_convective = ( g * shf_mean * zi_ribulk / pt_surf_mean )**( 1.0_wp / 3.0_wp )
ELSE
w_convective = 0.0_wp
ENDIF
!
!-- At the initial run the friction velocity is initialized close to zero, leading to almost zero
!-- disturbances at the boundaries. In order to obtain disturbances nevertheless, set a minium
!-- value of friction velocity for the reynolds-stress parametrization.
IF ( time_since_reference_point <= 0.0_wp ) scale_us = MAX( scale_us, 0.05_wp )
!
!-- Finally, in order to consider also neutral or stable stratification, compute momentum velocity
!-- scale from u* and convective velocity scale, according to Rotach et al. (1996).
scale_wm = ( scale_us**3 + 0.6_wp * w_convective**3 )**( 1.0_wp / 3.0_wp )
END SUBROUTINE calc_scaling_variables
END MODULE synthetic_turbulence_generator_mod