!> @file vertical_nesting_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 1997-2019 Leibniz Universitaet Hannover ! Copyright 2017-2018 Karlsruhe Institute of Technology !------------------------------------------------------------------------------! ! ! Current revisions: ! ----------------- ! ! ! Former revisions: ! ----------------- ! $Id: vertical_nesting_mod.f90 4102 2019-07-17 16:00:03Z oliver.maas $ ! - Slightly revise setting of boundary conditions at horizontal walls, use ! data-structure offset index instead of pre-calculate it for each facing ! ! 4101 2019-07-17 15:14:26Z gronemeier ! remove old_dt ! ! 3802 2019-03-17 13:33:42Z raasch ! unused subroutines commented out ! ! 3655 2019-01-07 16:51:22Z knoop ! unused variables removed ! ! 3232 2018-09-07 12:21:44Z raasch ! references to mrun replaced by palmrun, and updated ! ! 3083 2018-06-19 14:03:12Z gronemeier ! Error messages revised ! ! 3065 2018-06-12 07:03:02Z Giersch ! dz was replaced by dz(1), error messages related to vertical grid stretching ! have been added ! ! 3049 2018-05-29 13:52:36Z Giersch ! Error messages revised ! ! 3045 2018-05-28 07:55:41Z Giersch ! Error message revised ! ! 2718 2018-01-02 08:49:38Z maronga ! Corrected "Former revisions" section ! ! 2712 2017-12-20 17:32:50Z kanani ! Formatting and clean-up (SadiqHuq) ! ! 2696 2017-12-14 17:12:51Z kanani ! Change in file header (GPL part) ! Renamed diffusivities to tcm_diffusivities (TG) ! ! 2516 2017-10-04 11:03:04Z suehring ! Remove tabs ! ! 2514 2017-10-04 09:52:37Z suehring ! Remove tabs ! ! 2514 2017-10-04 09:52:37Z suehring ! Added todo list ! ! 2365 2017-08-21 14:59:59Z kanani ! Initial revision (SadiqHuq) ! ! ! ! ! Description: ! ------------ !> Module for vertical nesting. !> !> Definition of parameters and variables for vertical nesting !> The horizontal extent of the parent (Coarse Grid) and the child (Fine Grid) !> have to be identical. The vertical extent of the FG should be smaller than CG. !> Only integer nesting ratio supported. Odd nesting ratio preferred !> The code follows MPI-1 standards. The available processors are split into !> two groups using MPI_COMM_SPLIT. Exchange of data from CG to FG is called !> interpolation. FG initialization by interpolation is done once at the start. !> FG boundary conditions are set by interpolated at every timestep. !> Exchange of data from CG to FG is called anterpolation, the two-way interaction !> occurs at every timestep. !> vnest_start_time set in PARIN allows delayed start of the coupling !> after spin-up of the CG !> !> @todo Replace dz(1) appropriatly to account for grid stretching !> @todo Ensure that code can be compiled for serial and parallel mode. Please !> check the placement of the directive "__parallel". !> @todo Add descriptions for all declared variables/parameters, one declaration !> statement per variable !> @todo Add a descriptive header above each subroutine (see land_surface_model) !> @todo FORTRAN language statements must not be used as names for variables !> (e.g. if). Please rename it. !> @todo Revise code according to PALM Coding Standard !------------------------------------------------------------------------------! MODULE vertical_nesting_mod USE kinds IMPLICIT NONE LOGICAL :: vnested = .FALSE. !> set to true if palmrun !> provides specific information via stdin LOGICAL :: vnest_init = .FALSE. !> set to true when FG is initialized REAL(wp) :: vnest_start_time = 9999999.9 !> simulated time when FG should be !> initialized. Should be !> identical in PARIN & PARIN_N INTEGER(iwp),DIMENSION(3,2) :: bdims = 0 !> sub-domain grid topology of current PE INTEGER(iwp),DIMENSION(3,2) :: bdims_rem = 0 !> sub-domain grid topology of partner PE INTEGER(iwp) :: cg_nprocs !> no. of PE in CG. Set by palmrun -Y INTEGER(iwp) :: fg_nprocs !> no. of PE in FG. Set by palmrun -Y INTEGER(iwp) :: TYPE_VNEST_BC !> derived contiguous data type for interpolation INTEGER(iwp) :: TYPE_VNEST_ANTER !> derived contiguous data type for anterpolation INTEGER(iwp),DIMENSION(:,:,:),ALLOCATABLE :: c2f_dims_cg !> One CG PE sends data to multiple FG PEs !> list of grid-topology of partners INTEGER(iwp),DIMENSION(:,:,:),ALLOCATABLE :: f2c_dims_cg !> One CG PE receives data from multiple FG PEs !> list of grid-topology of partners INTEGER(iwp),DIMENSION(:),ALLOCATABLE :: c2f_dims_fg !> One FG PE sends data to multiple CG PE !> list of grid-topology of partner INTEGER(iwp),DIMENSION(:),ALLOCATABLE :: f2c_dims_fg !> One FG PE sends data to only one CG PE !> list of grid-topology of partner INTEGER(iwp),DIMENSION(:,:),ALLOCATABLE :: f_rnk_lst !> list storing rank of FG PE denoted by pdims INTEGER(iwp),DIMENSION(:,:),ALLOCATABLE :: c_rnk_lst !> list storing rank of CG PE denoted by pdims INTEGER(iwp),DIMENSION(3) :: cfratio !> Nesting ratio in x,y and z-directions INTEGER(iwp) :: nxc !> no. of CG grid points in x-direction INTEGER(iwp) :: nxf !> no. of FG grid points in x-direction INTEGER(iwp) :: nyc !> no. of CG grid points in y-direction INTEGER(iwp) :: nyf !> no. of FG grid points in y-direction INTEGER(iwp) :: nzc !> no. of CG grid points in z-direction INTEGER(iwp) :: nzf !> no. of FG grid points in z-direction INTEGER(iwp) :: ngp_c !> no. of CG grid points in one vertical level INTEGER(iwp) :: ngp_f !> no. of FG grid points in one vertical level INTEGER(iwp) :: n_cell_c !> total no. of CG grid points in a PE INTEGER(iwp),DIMENSION(2) :: pdims_partner !> processor topology of partner PE INTEGER(iwp) :: target_idex !> temporary variable INTEGER(iwp),DIMENSION(2) :: offset !> temporary variable INTEGER(iwp),DIMENSION(2) :: map_coord !> temporary variable REAL(wp) :: dxc !> CG grid pacing in x-direction REAL(wp) :: dyc !> FG grid pacing in x-direction REAL(wp) :: dxf !> CG grid pacing in y-direction REAL(wp) :: dyf !> FG grid pacing in y-direction REAL(wp) :: dzc !> CG grid pacing in z-direction REAL(wp) :: dzf !> FG grid pacing in z-direction REAL(wp) :: dtc !> dt calculated for CG REAL(wp) :: dtf !> dt calculated for FG REAL(wp), DIMENSION(:), ALLOCATABLE :: zuc !> CG vertical u-levels REAL(wp), DIMENSION(:), ALLOCATABLE :: zuf !> FG vertical u-levels REAL(wp), DIMENSION(:), ALLOCATABLE :: zwc !> CG vertical w-levels REAL(wp), DIMENSION(:), ALLOCATABLE :: zwf !> FG vertical w-levels REAL(wp), DIMENSION(:,:,:), POINTER :: interpol3d !> pointers to simplify function calls REAL(wp), DIMENSION(:,:,:), POINTER :: anterpol3d !> pointers to simplify function calls REAL(wp),DIMENSION(:,:,:), ALLOCATABLE :: work3d !> temporary array for exchange of 3D data REAL(wp),DIMENSION(:,:), ALLOCATABLE :: work2dusws !> temporary array for exchange of 2D data REAL(wp),DIMENSION(:,:), ALLOCATABLE :: work2dvsws !> temporary array for exchange of 2D data REAL(wp),DIMENSION(:,:), ALLOCATABLE :: work2dts !> temporary array for exchange of 2D data REAL(wp),DIMENSION(:,:), ALLOCATABLE :: work2dus !> temporary array for exchange of 2D data SAVE !-- Public functions PUBLIC vnest_init_fine, vnest_boundary_conds, vnest_anterpolate, & vnest_boundary_conds_khkm, vnest_anterpolate_e, & vnest_init_pegrid_rank, vnest_init_pegrid_domain, vnest_init_grid, & vnest_timestep_sync, vnest_deallocate !-- Public constants and variables PUBLIC vnested, vnest_init, vnest_start_time PRIVATE bdims, bdims_rem, & work3d, work2dusws, work2dvsws, work2dts, work2dus, & dxc, dyc, dxf, dyf, dzc, dzf, dtc, dtf, & zuc, zuf, zwc, zwf, interpol3d, anterpol3d, & cg_nprocs, fg_nprocs, & c2f_dims_cg, c2f_dims_fg, f2c_dims_cg, f2c_dims_fg, & f_rnk_lst, c_rnk_lst, cfratio, pdims_partner, & nxc, nxf, nyc, nyf, nzc, nzf, & ngp_c, ngp_f, target_idex, n_cell_c, & offset, map_coord, TYPE_VNEST_BC, TYPE_VNEST_ANTER INTERFACE vnest_anterpolate MODULE PROCEDURE vnest_anterpolate END INTERFACE vnest_anterpolate INTERFACE vnest_anterpolate_e MODULE PROCEDURE vnest_anterpolate_e END INTERFACE vnest_anterpolate_e INTERFACE vnest_boundary_conds MODULE PROCEDURE vnest_boundary_conds END INTERFACE vnest_boundary_conds INTERFACE vnest_boundary_conds_khkm MODULE PROCEDURE vnest_boundary_conds_khkm END INTERFACE vnest_boundary_conds_khkm INTERFACE vnest_check_parameters MODULE PROCEDURE vnest_check_parameters END INTERFACE vnest_check_parameters INTERFACE vnest_deallocate MODULE PROCEDURE vnest_deallocate END INTERFACE vnest_deallocate INTERFACE vnest_init_fine MODULE PROCEDURE vnest_init_fine END INTERFACE vnest_init_fine INTERFACE vnest_init_grid MODULE PROCEDURE vnest_init_grid END INTERFACE vnest_init_grid INTERFACE vnest_init_pegrid_domain MODULE PROCEDURE vnest_init_pegrid_domain END INTERFACE vnest_init_pegrid_domain INTERFACE vnest_init_pegrid_rank MODULE PROCEDURE vnest_init_pegrid_rank END INTERFACE vnest_init_pegrid_rank INTERFACE vnest_timestep_sync MODULE PROCEDURE vnest_timestep_sync END INTERFACE vnest_timestep_sync CONTAINS SUBROUTINE vnest_init_fine #if defined( __parallel ) !--------------------------------------------------------------------------------! ! Description: ! ------------ ! At the specified vnest_start_time initialize the Fine Grid based on the coarse ! grid values !------------------------------------------------------------------------------! USE arrays_3d USE control_parameters USE grid_variables USE indices USE interfaces USE pegrid USE turbulence_closure_mod, & ONLY : tcm_diffusivities IMPLICIT NONE REAL(wp) :: time_since_reference_point_rem INTEGER(iwp) :: i INTEGER(iwp) :: j INTEGER(iwp) :: iif INTEGER(iwp) :: jjf INTEGER(iwp) :: kkf if (myid ==0 ) print *, ' TIME TO INIT FINE from COARSE', simulated_time ! !-- In case of model termination initiated by the remote model !-- (terminate_coupled_remote > 0), initiate termination of the local model. !-- The rest of the coupler must then be skipped because it would cause an MPI !-- intercomminucation hang. !-- If necessary, the coupler will be called at the beginning of the next !-- restart run. IF ( myid == 0) THEN CALL MPI_SENDRECV( terminate_coupled, 1, MPI_INTEGER, & target_id, 0, & terminate_coupled_remote, 1, MPI_INTEGER, & target_id, 0, & comm_inter, status, ierr ) ENDIF CALL MPI_BCAST( terminate_coupled_remote, 1, MPI_INTEGER, 0, comm2d, & ierr ) IF ( terminate_coupled_remote > 0 ) THEN WRITE( message_string, * ) 'remote model "', & TRIM( coupling_mode_remote ), & '" terminated', & '&with terminate_coupled_remote = ', & terminate_coupled_remote, & '&local model "', TRIM( coupling_mode ), & '" has', & '&terminate_coupled = ', & terminate_coupled CALL message( 'vnest_init_fine', 'PA0310', 1, 2, 0, 6, 0 ) RETURN ENDIF ! !-- Exchange the current simulated time between the models, !-- currently just for total_2ding IF ( myid == 0 ) THEN CALL MPI_SEND( time_since_reference_point, 1, MPI_REAL, target_id, & 11, comm_inter, ierr ) CALL MPI_RECV( time_since_reference_point_rem, 1, MPI_REAL, & target_id, 11, comm_inter, status, ierr ) ENDIF CALL MPI_BCAST( time_since_reference_point_rem, 1, MPI_REAL, 0, comm2d, & ierr ) IF ( coupling_mode == 'vnested_crse' ) THEN !-- Send data to fine grid for initialization offset(1) = ( pdims_partner(1) / pdims(1) ) * pcoord(1) offset(2) = ( pdims_partner(2) / pdims(2) ) * pcoord(2) do j = 0, ( pdims_partner(2) / pdims(2) ) - 1 do i = 0, ( pdims_partner(1) / pdims(1) ) - 1 map_coord(1) = i+offset(1) map_coord(2) = j+offset(2) target_idex = f_rnk_lst(map_coord(1),map_coord(2)) + numprocs CALL MPI_RECV( bdims_rem, 6, MPI_INTEGER, target_idex, 10, & comm_inter,status, ierr ) bdims (1,1) = bdims_rem (1,1) / cfratio(1) bdims (1,2) = bdims_rem (1,2) / cfratio(1) bdims (2,1) = bdims_rem (2,1) / cfratio(2) bdims (2,2) = bdims_rem (2,2) / cfratio(2) bdims (3,1) = bdims_rem (3,1) bdims (3,2) = bdims_rem (3,2) / cfratio(3) CALL MPI_SEND( bdims, 6, MPI_INTEGER, target_idex, 9, & comm_inter, ierr ) n_cell_c = (bdims(1,2)-bdims(1,1)+3) * & (bdims(2,2)-bdims(2,1)+3) * & (bdims(3,2)-bdims(3,1)+3) CALL MPI_SEND( u( bdims(3,1):bdims(3,2)+2, & bdims(2,1)-1:bdims(2,2)+1, & bdims(1,1)-1:bdims(1,2)+1),& n_cell_c, MPI_REAL, target_idex, & 101, comm_inter, ierr) CALL MPI_SEND( v( bdims(3,1):bdims(3,2)+2, & bdims(2,1)-1:bdims(2,2)+1, & bdims(1,1)-1:bdims(1,2)+1),& n_cell_c, MPI_REAL, target_idex, & 102, comm_inter, ierr) CALL MPI_SEND( w( bdims(3,1):bdims(3,2)+2, & bdims(2,1)-1:bdims(2,2)+1, & bdims(1,1)-1:bdims(1,2)+1),& n_cell_c, MPI_REAL, target_idex, & 103, comm_inter, ierr) CALL MPI_SEND( pt(bdims(3,1):bdims(3,2)+2, & bdims(2,1)-1:bdims(2,2)+1, & bdims(1,1)-1:bdims(1,2)+1),& n_cell_c, MPI_REAL, target_idex, & 105, comm_inter, ierr) IF ( humidity ) THEN CALL MPI_SEND( q(bdims(3,1):bdims(3,2)+2, & bdims(2,1)-1:bdims(2,2)+1, & bdims(1,1)-1:bdims(1,2)+1),& n_cell_c, MPI_REAL, target_idex, & 116, comm_inter, ierr) ENDIF CALL MPI_SEND( e( bdims(3,1):bdims(3,2)+2, & bdims(2,1)-1:bdims(2,2)+1, & bdims(1,1)-1:bdims(1,2)+1),& n_cell_c, MPI_REAL, target_idex, & 104, comm_inter, ierr) CALL MPI_SEND(kh( bdims(3,1):bdims(3,2)+2, & bdims(2,1)-1:bdims(2,2)+1, & bdims(1,1)-1:bdims(1,2)+1),& n_cell_c, MPI_REAL, target_idex, & 106, comm_inter, ierr) CALL MPI_SEND(km( bdims(3,1):bdims(3,2)+2, & bdims(2,1)-1:bdims(2,2)+1, & bdims(1,1)-1:bdims(1,2)+1),& n_cell_c, MPI_REAL, target_idex, & 107, comm_inter, ierr) !-- Send Surface fluxes IF ( use_surface_fluxes ) THEN n_cell_c = (bdims(1,2)-bdims(1,1)+3) * & (bdims(2,2)-bdims(2,1)+3) ! !-- shf and z0 for CG / FG need to initialized in input file or user_code !-- TODO !-- initialization of usws, vsws, ts and us not vital to vnest FG !-- variables are not compatible with the new surface layer module ! ! CALL MPI_SEND(surf_def_h(0)%usws( bdims(2,1)-1:bdims(2,2)+1, & ! bdims(1,1)-1:bdims(1,2)+1),& ! n_cell_c, MPI_REAL, target_idex, & ! 110, comm_inter, ierr ) ! ! CALL MPI_SEND(surf_def_h(0)%vsws( bdims(2,1)-1:bdims(2,2)+1, & ! bdims(1,1)-1:bdims(1,2)+1),& ! n_cell_c, MPI_REAL, target_idex, & ! 111, comm_inter, ierr ) ! ! CALL MPI_SEND(ts ( bdims(2,1)-1:bdims(2,2)+1, & ! bdims(1,1)-1:bdims(1,2)+1),& ! n_cell_c, MPI_REAL, target_idex, & ! 112, comm_inter, ierr ) ! ! CALL MPI_SEND(us ( bdims(2,1)-1:bdims(2,2)+1, & ! bdims(1,1)-1:bdims(1,2)+1),& ! n_cell_c, MPI_REAL, target_idex, & ! 113, comm_inter, ierr ) ! ENDIF end do end do ELSEIF ( coupling_mode == 'vnested_fine' ) THEN !-- Receive data from coarse grid for initialization offset(1) = pcoord(1) / ( pdims(1)/pdims_partner(1) ) offset(2) = pcoord(2) / ( pdims(2)/pdims_partner(2) ) map_coord(1) = offset(1) map_coord(2) = offset(2) target_idex = c_rnk_lst(map_coord(1),map_coord(2)) bdims (1,1) = nxl bdims (1,2) = nxr bdims (2,1) = nys bdims (2,2) = nyn bdims (3,1) = nzb bdims (3,2) = nzt CALL MPI_SEND( bdims, 6, MPI_INTEGER, target_idex, 10, & comm_inter, ierr ) CALL MPI_RECV( bdims_rem, 6, MPI_INTEGER, target_idex, 9, & comm_inter,status, ierr ) n_cell_c = (bdims_rem(1,2)-bdims_rem(1,1)+3) * & (bdims_rem(2,2)-bdims_rem(2,1)+3) * & (bdims_rem(3,2)-bdims_rem(3,1)+3) ALLOCATE( work3d ( bdims_rem(3,1) :bdims_rem(3,2)+2, & bdims_rem(2,1)-1:bdims_rem(2,2)+1, & bdims_rem(1,1)-1:bdims_rem(1,2)+1)) CALL MPI_RECV( work3d,n_cell_c, MPI_REAL, target_idex, 101, & comm_inter,status, ierr ) interpol3d => u call interpolate_to_fine_u CALL MPI_RECV( work3d,n_cell_c, MPI_REAL, target_idex, 102, & comm_inter,status, ierr ) interpol3d => v call interpolate_to_fine_v CALL MPI_RECV( work3d,n_cell_c, MPI_REAL, target_idex, 103, & comm_inter,status, ierr ) interpol3d => w call interpolate_to_fine_w CALL MPI_RECV( work3d,n_cell_c, MPI_REAL, target_idex, 105, & comm_inter,status, ierr ) interpol3d => pt call interpolate_to_fine_s IF ( humidity ) THEN CALL MPI_RECV( work3d,n_cell_c, MPI_REAL, target_idex, 116, & comm_inter,status, ierr ) interpol3d => q call interpolate_to_fine_s ENDIF CALL MPI_RECV( work3d,n_cell_c, MPI_REAL, target_idex, 104, & comm_inter,status, ierr ) interpol3d => e call interpolate_to_fine_s !-- kh,km no target attribute, use of pointer not possible CALL MPI_RECV( work3d,n_cell_c, MPI_REAL, target_idex, 106, & comm_inter,status, ierr ) call interpolate_to_fine_kh CALL MPI_RECV( work3d,n_cell_c, MPI_REAL, target_idex, 107, & comm_inter,status, ierr ) call interpolate_to_fine_km DEALLOCATE( work3d ) NULLIFY ( interpol3d ) !-- Recv Surface Fluxes IF ( use_surface_fluxes ) THEN n_cell_c = (bdims_rem(1,2)-bdims_rem(1,1)+3) * & (bdims_rem(2,2)-bdims_rem(2,1)+3) ALLOCATE( work2dusws ( bdims_rem(2,1)-1:bdims_rem(2,2)+1, & bdims_rem(1,1)-1:bdims_rem(1,2)+1) ) ALLOCATE( work2dvsws ( bdims_rem(2,1)-1:bdims_rem(2,2)+1, & bdims_rem(1,1)-1:bdims_rem(1,2)+1) ) ALLOCATE( work2dts ( bdims_rem(2,1)-1:bdims_rem(2,2)+1, & bdims_rem(1,1)-1:bdims_rem(1,2)+1) ) ALLOCATE( work2dus ( bdims_rem(2,1)-1:bdims_rem(2,2)+1, & bdims_rem(1,1)-1:bdims_rem(1,2)+1) ) ! !-- shf and z0 for CG / FG need to initialized in input file or user_code !-- TODO !-- initialization of usws, vsws, ts and us not vital to vnest FG !-- variables are not compatible with the new surface layer module ! ! CALL MPI_RECV( work2dusws,n_cell_c, MPI_REAL, target_idex, 110, & ! comm_inter,status, ierr ) ! ! CALL MPI_RECV( work2dvsws,n_cell_c, MPI_REAL, target_idex, 111, & ! comm_inter,status, ierr ) ! ! CALL MPI_RECV( work2dts ,n_cell_c, MPI_REAL, target_idex, 112, & ! comm_inter,status, ierr ) ! ! CALL MPI_RECV( work2dus ,n_cell_c, MPI_REAL, target_idex, 113, & ! comm_inter,status, ierr ) ! ! CALL interpolate_to_fine_flux ( 108 ) DEALLOCATE( work2dusws ) DEALLOCATE( work2dvsws ) DEALLOCATE( work2dts ) DEALLOCATE( work2dus ) ENDIF IF ( .NOT. constant_diffusion ) THEN DO kkf = bdims(3,1)+1,bdims(3,2)+1 DO jjf = bdims(2,1),bdims(2,2) DO iif = bdims(1,1),bdims(1,2) IF ( e(kkf,jjf,iif) < 0.0 ) THEN e(kkf,jjf,iif) = 1E-15_wp END IF END DO END DO END DO ENDIF w(nzt+1,:,:) = w(nzt,:,:) CALL exchange_horiz( u, nbgp ) CALL exchange_horiz( v, nbgp ) CALL exchange_horiz( w, nbgp ) CALL exchange_horiz( pt, nbgp ) IF ( .NOT. constant_diffusion ) CALL exchange_horiz( e, nbgp ) IF ( humidity ) CALL exchange_horiz( q, nbgp ) ! !-- Velocity boundary conditions at the bottom boundary IF ( ibc_uv_b == 0 ) THEN u(nzb,:,:) = 0.0_wp v(nzb,:,:) = 0.0_wp ELSE u(nzb,:,:) = u(nzb+1,:,:) v(nzb,:,:) = v(nzb+1,:,:) END IF w(nzb,:,:) = 0.0_wp ! !-- Temperature boundary conditions at the bottom boundary IF ( ibc_pt_b /= 0 ) THEN pt(nzb,:,:) = pt(nzb+1,:,:) END IF ! !-- Bottom boundary condition for the turbulent kinetic energy !-- Generally a Neumann condition with de/dz=0 is assumed IF ( .NOT. constant_diffusion ) THEN e(nzb,:,:) = e(nzb+1,:,:) END IF ! !-- Bottom boundary condition for turbulent diffusion coefficients km(nzb,:,:) = km(nzb+1,:,:) kh(nzb,:,:) = kh(nzb+1,:,:) !diffusivities required IF ( .NOT. humidity ) THEN CALL tcm_diffusivities( pt, pt_reference ) ELSE CALL tcm_diffusivities( vpt, pt_reference ) ENDIF ! !-- Reset Fine Grid top Boundary Condition !-- At the top of the FG, the scalars always follow Dirichlet condition ibc_pt_t = 0 !-- Initialize old time levels pt_p = pt; u_p = u; v_p = v; w_p = w IF ( .NOT. constant_diffusion ) e_p = e IF ( humidity ) THEN ibc_q_t = 0 q_p = q ENDIF ENDIF if (myid==0) print *, '** Fine Initalized ** simulated_time:', simulated_time CONTAINS SUBROUTINE interpolate_to_fine_w USE arrays_3d USE control_parameters USE grid_variables USE indices USE pegrid IMPLICIT NONE INTEGER(iwp) :: i INTEGER(iwp) :: j INTEGER(iwp) :: k INTEGER(iwp) :: iif INTEGER(iwp) :: jjf INTEGER(iwp) :: kkf INTEGER(iwp) :: nzbottom INTEGER(iwp) :: nztop INTEGER(iwp) :: bottomx INTEGER(iwp) :: bottomy INTEGER(iwp) :: bottomz INTEGER(iwp) :: topx INTEGER(iwp) :: topy INTEGER(iwp) :: topz REAL(wp) :: eps REAL(wp) :: alpha REAL(wp) :: eminus REAL(wp) :: edot REAL(wp) :: eplus REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: wprs REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: wprf nzbottom = bdims_rem (3,1) nztop = bdims_rem (3,2) ALLOCATE( wprf(nzbottom:nztop, bdims_rem(2,1)-1: bdims_rem(2,2)+1,nxl:nxr) ) ALLOCATE( wprs(nzbottom:nztop,nys:nyn,nxl:nxr) ) ! !-- Initialisation of the velocity component w ! !-- Interpolation in x-direction DO k = nzbottom, nztop DO j = bdims_rem(2,1)-1, bdims_rem(2,2)+1 DO i = bdims_rem(1,1),bdims_rem(1,2) bottomx = (nxf+1)/(nxc+1) * i topx = (nxf+1)/(nxc+1) * (i+1) - 1 DO iif = bottomx, topx eps = ( iif * dxf + 0.5 * dxf - i * dxc - 0.5 * dxc ) / dxc alpha = ( ( dxf / dxc )**2.0 - 1.0 ) / 24.0 eminus = eps * ( eps - 1.0 ) / 2.0 + alpha edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha eplus = eps * ( eps + 1.0 ) / 2.0 + alpha wprf(k,j,iif) = eminus * work3d(k,j,i-1) & + edot * work3d(k,j,i) & + eplus * work3d(k,j,i+1) END DO END DO END DO END DO ! !-- Interpolation in y-direction (quadratic, Clark and Farley) DO k = nzbottom, nztop DO j = bdims_rem(2,1), bdims_rem(2,2) bottomy = (nyf+1)/(nyc+1) * j topy = (nyf+1)/(nyc+1) * (j+1) - 1 DO iif = nxl, nxr DO jjf = bottomy, topy eps = ( jjf * dyf + 0.5 * dyf - j * dyc - 0.5 * dyc ) / dyc alpha = ( ( dyf / dyc )**2.0 - 1.0 ) / 24.0 eminus = eps * ( eps - 1.0 ) / 2.0 + alpha edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha eplus = eps * ( eps + 1.0 ) / 2.0 + alpha wprs(k,jjf,iif) = eminus * wprf(k,j-1,iif) & + edot * wprf(k,j,iif) & + eplus * wprf(k,j+1,iif) END DO END DO END DO END DO ! !-- Interpolation in z-direction (linear) DO k = nzbottom, nztop-1 bottomz = (dzc/dzf) * k topz = (dzc/dzf) * (k+1) - 1 DO jjf = nys, nyn DO iif = nxl, nxr DO kkf = bottomz, topz w(kkf,jjf,iif) = wprs(k,jjf,iif) + ( zwf(kkf) - zwc(k) ) & * ( wprs(k+1,jjf,iif) - wprs(k,jjf,iif) ) / dzc END DO END DO END DO END DO DO jjf = nys, nyn DO iif = nxl, nxr w(nzt,jjf,iif) = wprs(nztop,jjf,iif) END DO END DO ! ! w(nzb:nzt+1,nys:nyn,nxl:nxr) = 0 DEALLOCATE( wprf, wprs ) END SUBROUTINE interpolate_to_fine_w SUBROUTINE interpolate_to_fine_u USE arrays_3d USE control_parameters USE grid_variables USE indices USE pegrid IMPLICIT NONE INTEGER(iwp) :: i INTEGER(iwp) :: j INTEGER(iwp) :: k INTEGER(iwp) :: iif INTEGER(iwp) :: jjf INTEGER(iwp) :: kkf INTEGER(iwp) :: nzbottom INTEGER(iwp) :: nztop INTEGER(iwp) :: bottomx INTEGER(iwp) :: bottomy INTEGER(iwp) :: bottomz INTEGER(iwp) :: topx INTEGER(iwp) :: topy INTEGER(iwp) :: topz REAL(wp) :: eps REAL(wp) :: alpha REAL(wp) :: eminus REAL(wp) :: edot REAL(wp) :: eplus REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: uprf REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: uprs nzbottom = bdims_rem (3,1) nztop = bdims_rem (3,2) ALLOCATE( uprf(nzbottom:nztop+2,nys:nyn,bdims_rem(1,1)-1:bdims_rem(1,2)+1) ) ALLOCATE( uprs(nzb+1:nzt+1,nys:nyn,bdims_rem(1,1)-1:bdims_rem(1,2)+1) ) ! !-- Initialisation of the velocity component uf ! !-- Interpolation in y-direction (quadratic, Clark and Farley) DO k = nzbottom, nztop+2 DO j = bdims_rem(2,1), bdims_rem(2,2) bottomy = (nyf+1)/(nyc+1) * j topy = (nyf+1)/(nyc+1) * (j+1) - 1 DO i = bdims_rem(1,1)-1, bdims_rem(1,2)+1 DO jjf = bottomy, topy eps = ( jjf * dyf + 0.5 * dyf - j * dyc - 0.5 * dyc ) / dyc alpha = ( ( dyf / dyc )**2.0 - 1.0 ) / 24.0 eminus = eps * ( eps - 1.0 ) / 2.0 + alpha edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha eplus = eps * ( eps + 1.0 ) / 2.0 + alpha uprf(k,jjf,i) = eminus * work3d(k,j-1,i) & + edot * work3d(k,j,i) & + eplus * work3d(k,j+1,i) END DO END DO END DO END DO ! !-- Interpolation in z-direction (quadratic, Clark and Farley) DO k = nzbottom+1, nztop bottomz = (dzc/dzf) * (k-1) + 1 topz = (dzc/dzf) * k DO jjf = nys, nyn DO i = bdims_rem(1,1)-1, bdims_rem(1,2)+1 DO kkf = bottomz, topz eps = ( zuf(kkf) - zuc(k) ) / dzc alpha = ( ( dzf / dzc )**2.0 - 1.0 ) / 24.0 eminus = eps * ( eps - 1.0 ) / 2.0 + alpha edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha eplus = eps * ( eps + 1.0 ) / 2.0 + alpha uprs(kkf,jjf,i) = eminus * uprf(k-1,jjf,i) & + edot * uprf(k,jjf,i) & + eplus * uprf(k+1,jjf,i) END DO END DO END DO END DO DO jjf = nys, nyn DO i = bdims_rem(1,1)-1, bdims_rem(1,2)+1 eps = ( zuf(nzt+1) - zuc(nztop+1) ) / dzc alpha = ( ( dzf / dzc )**2.0 - 1.0 ) / 24.0 eminus = eps * ( eps - 1.0 ) / 2.0 + alpha edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha eplus = eps * ( eps + 1.0 ) / 2.0 + alpha uprs(nzt+1,jjf,i) = eminus * uprf(nztop,jjf,i) & + edot * uprf(nztop+1,jjf,i) & + eplus * uprf(nztop+2,jjf,i) END DO END DO ! !-- Interpolation in x-direction (linear) DO kkf = nzb+1, nzt+1 DO jjf = nys, nyn DO i = bdims_rem(1,1), bdims_rem(1,2) bottomx = (nxf+1)/(nxc+1) * i topx = (nxf+1)/(nxc+1) * (i+1) - 1 DO iif = bottomx, topx u(kkf,jjf,iif) = uprs(kkf,jjf,i) + ( iif * dxf - i * dxc ) & * ( uprs(kkf,jjf,i+1) - uprs(kkf,jjf,i) ) / dxc END DO END DO END DO END DO ! !-- Determination of uf at the bottom boundary DEALLOCATE( uprf, uprs ) END SUBROUTINE interpolate_to_fine_u SUBROUTINE interpolate_to_fine_v USE arrays_3d USE control_parameters USE grid_variables USE indices USE pegrid IMPLICIT NONE INTEGER(iwp) :: i INTEGER(iwp) :: j INTEGER(iwp) :: k INTEGER(iwp) :: iif INTEGER(iwp) :: jjf INTEGER(iwp) :: kkf INTEGER(iwp) :: nzbottom INTEGER(iwp) :: nztop INTEGER(iwp) :: bottomx INTEGER(iwp) :: bottomy INTEGER(iwp) :: bottomz INTEGER(iwp) :: topx INTEGER(iwp) :: topy INTEGER(iwp) :: topz REAL(wp) :: eps REAL(wp) :: alpha REAL(wp) :: eminus REAL(wp) :: edot REAL(wp) :: eplus REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: vprs REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: vprf nzbottom = bdims_rem (3,1) nztop = bdims_rem (3,2) ALLOCATE( vprf(nzbottom:nztop+2,bdims_rem(2,1)-1:bdims_rem(2,2)+1,nxl:nxr) ) ALLOCATE( vprs(nzb+1:nzt+1, bdims_rem(2,1)-1:bdims_rem(2,2)+1,nxl:nxr) ) ! !-- Initialisation of the velocity component vf ! !-- Interpolation in x-direction (quadratic, Clark and Farley) DO k = nzbottom, nztop+2 DO j = bdims_rem(2,1)-1, bdims_rem(2,2)+1 DO i = bdims_rem(1,1), bdims_rem(1,2) bottomx = (nxf+1)/(nxc+1) * i topx = (nxf+1)/(nxc+1) * (i+1) - 1 DO iif = bottomx, topx eps = ( iif * dxf + 0.5 * dxf - i * dxc - 0.5 * dxc ) / dxc alpha = ( ( dxf / dxc )**2.0 - 1.0 ) / 24.0 eminus = eps * ( eps - 1.0 ) / 2.0 + alpha edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha eplus = eps * ( eps + 1.0 ) / 2.0 + alpha vprf(k,j,iif) = eminus * work3d(k,j,i-1) & + edot * work3d(k,j,i) & + eplus * work3d(k,j,i+1) END DO END DO END DO END DO ! !-- Interpolation in z-direction (quadratic, Clark and Farley) DO k = nzbottom+1, nztop bottomz = (dzc/dzf) * (k-1) + 1 topz = (dzc/dzf) * k DO j = bdims_rem(2,1)-1, bdims_rem(2,2)+1 DO iif = nxl, nxr DO kkf = bottomz, topz eps = ( zuf(kkf) - zuc(k) ) / dzc alpha = ( ( dzf / dzc )**2.0 - 1.0 ) / 24.0 eminus = eps * ( eps - 1.0 ) / 2.0 + alpha edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha eplus = eps * ( eps + 1.0 ) / 2.0 + alpha vprs(kkf,j,iif) = eminus * vprf(k-1,j,iif) & + edot * vprf(k,j,iif) & + eplus * vprf(k+1,j,iif) END DO END DO END DO END DO DO j = bdims_rem(2,1)-1, bdims_rem(2,2)+1 DO iif = nxl, nxr eps = ( zuf(nzt+1) - zuc(nztop+1) ) / dzc alpha = ( ( dzf / dzc )**2.0 - 1.0 ) / 24.0 eminus = eps * ( eps - 1.0 ) / 2.0 + alpha edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha eplus = eps * ( eps + 1.0 ) / 2.0 + alpha vprs(nzt+1,j,iif) = eminus * vprf(nztop,j,iif) & + edot * vprf(nztop+1,j,iif) & + eplus * vprf(nztop+2,j,iif) END DO END DO ! !-- Interpolation in y-direction (linear) DO kkf = nzb+1, nzt+1 DO j = bdims_rem(2,1), bdims_rem(2,2) bottomy = (nyf+1)/(nyc+1) * j topy = (nyf+1)/(nyc+1) * (j+1) - 1 DO iif = nxl, nxr DO jjf = bottomy, topy v (kkf,jjf,iif) = vprs(kkf,j,iif) + ( jjf * dyf - j * dyc ) & * ( vprs(kkf,j+1,iif) - vprs(kkf,j,iif) ) / dyc END DO END DO END DO END DO ! !-- Determination of vf at the bottom boundary DEALLOCATE( vprf, vprs ) END SUBROUTINE interpolate_to_fine_v SUBROUTINE interpolate_to_fine_s USE arrays_3d USE control_parameters USE grid_variables USE indices USE pegrid IMPLICIT NONE INTEGER(iwp) :: i INTEGER(iwp) :: j INTEGER(iwp) :: k INTEGER(iwp) :: iif INTEGER(iwp) :: jjf INTEGER(iwp) :: kkf INTEGER(iwp) :: nzbottom INTEGER(iwp) :: nztop INTEGER(iwp) :: bottomx INTEGER(iwp) :: bottomy INTEGER(iwp) :: bottomz INTEGER(iwp) :: topx INTEGER(iwp) :: topy INTEGER(iwp) :: topz REAL(wp) :: eps REAL(wp) :: alpha REAL(wp) :: eminus REAL(wp) :: edot REAL(wp) :: eplus REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: ptprs REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: ptprf nzbottom = bdims_rem (3,1) nztop = bdims_rem (3,2) ALLOCATE( ptprf(nzbottom:nztop+2,bdims_rem(2,1)-1:bdims_rem(2,2)+1,nxl:nxr) ) ALLOCATE( ptprs(nzbottom:nztop+2,nys:nyn,nxl:nxr) ) ! !-- Initialisation of scalar variables ! !-- Interpolation in x-direction (quadratic, Clark and Farley) DO k = nzbottom, nztop+2 DO j = bdims_rem(2,1)-1, bdims_rem(2,2)+1 DO i = bdims_rem(1,1), bdims_rem(1,2) bottomx = (nxf+1)/(nxc+1) * i topx = (nxf+1)/(nxc+1) * (i+1) - 1 DO iif = bottomx, topx eps = ( iif * dxf + 0.5 * dxf - i * dxc - 0.5 * dxc ) / dxc alpha = ( ( dxf / dxc )**2.0 - 1.0 ) / 24.0 eminus = eps * ( eps - 1.0 ) / 2.0 + alpha edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha eplus = eps * ( eps + 1.0 ) / 2.0 + alpha ptprf(k,j,iif) = eminus * work3d(k,j,i-1) & + edot * work3d(k,j,i) & + eplus * work3d(k,j,i+1) END DO END DO END DO END DO ! !-- Interpolation in y-direction (quadratic, Clark and Farley) DO k = nzbottom, nztop+2 DO j = bdims_rem(2,1), bdims_rem(2,2) bottomy = (nyf+1)/(nyc+1) * j topy = (nyf+1)/(nyc+1) * (j+1) - 1 DO iif = nxl, nxr DO jjf = bottomy, topy eps = ( jjf * dyf + 0.5 * dyf - j * dyc - 0.5 * dyc ) / dyc alpha = ( ( dyf / dyc )**2.0 - 1.0 ) / 24.0 eminus = eps * ( eps - 1.0 ) / 2.0 + alpha edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha eplus = eps * ( eps + 1.0 ) / 2.0 + alpha ptprs(k,jjf,iif) = eminus * ptprf(k,j-1,iif) & + edot * ptprf(k,j,iif) & + eplus * ptprf(k,j+1,iif) END DO END DO END DO END DO ! !-- Interpolation in z-direction (quadratic, Clark and Farley) DO k = nzbottom+1, nztop bottomz = (dzc/dzf) * (k-1) + 1 topz = (dzc/dzf) * k DO jjf = nys, nyn DO iif = nxl, nxr DO kkf = bottomz, topz eps = ( zuf(kkf) - zuc(k) ) / dzc alpha = ( ( dzf / dzc )**2.0 - 1.0 ) / 24.0 eminus = eps * ( eps - 1.0 ) / 2.0 + alpha edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha eplus = eps * ( eps + 1.0 ) / 2.0 + alpha interpol3d(kkf,jjf,iif) = eminus * ptprs(k-1,jjf,iif) & + edot * ptprs(k,jjf,iif) & + eplus * ptprs(k+1,jjf,iif) END DO END DO END DO END DO DO jjf = nys, nyn DO iif = nxl, nxr eps = ( zuf(nzt+1) - zuc(nztop+1) ) / dzc alpha = ( ( dzf / dzc )**2.0 - 1.0 ) / 24.0 eminus = eps * ( eps - 1.0 ) / 2.0 + alpha edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha eplus = eps * ( eps + 1.0 ) / 2.0 + alpha interpol3d(nzt+1,jjf,iif) = eminus * ptprs(nztop,jjf,iif) & + edot * ptprs(nztop+1,jjf,iif) & + eplus * ptprs(nztop+2,jjf,iif) END DO END DO DEALLOCATE( ptprf, ptprs ) END SUBROUTINE interpolate_to_fine_s SUBROUTINE interpolate_to_fine_kh USE arrays_3d USE control_parameters USE grid_variables USE indices USE pegrid IMPLICIT NONE INTEGER(iwp) :: i INTEGER(iwp) :: j INTEGER(iwp) :: k INTEGER(iwp) :: iif INTEGER(iwp) :: jjf INTEGER(iwp) :: kkf INTEGER(iwp) :: nzbottom INTEGER(iwp) :: nztop INTEGER(iwp) :: bottomx INTEGER(iwp) :: bottomy INTEGER(iwp) :: bottomz INTEGER(iwp) :: topx INTEGER(iwp) :: topy INTEGER(iwp) :: topz REAL(wp) :: eps REAL(wp) :: alpha REAL(wp) :: eminus REAL(wp) :: edot REAL(wp) :: eplus REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: ptprs REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: ptprf nzbottom = bdims_rem (3,1) nztop = bdims_rem (3,2) ! nztop = blk_dim_rem (3,2)+1 ALLOCATE( ptprf(nzbottom:nztop+2,bdims_rem(2,1)-1:bdims_rem(2,2)+1,nxl:nxr) ) ALLOCATE( ptprs(nzbottom:nztop+2,nys:nyn,nxl:nxr) ) ! !-- Initialisation of scalar variables ! !-- Interpolation in x-direction (quadratic, Clark and Farley) DO k = nzbottom, nztop+2 DO j = bdims_rem(2,1)-1, bdims_rem(2,2)+1 DO i = bdims_rem(1,1), bdims_rem(1,2) bottomx = (nxf+1)/(nxc+1) * i topx = (nxf+1)/(nxc+1) * (i+1) - 1 DO iif = bottomx, topx eps = ( iif * dxf + 0.5 * dxf - i * dxc - 0.5 * dxc ) / dxc alpha = ( ( dxf / dxc )**2.0 - 1.0 ) / 24.0 eminus = eps * ( eps - 1.0 ) / 2.0 + alpha edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha eplus = eps * ( eps + 1.0 ) / 2.0 + alpha ptprf(k,j,iif) = eminus * work3d(k,j,i-1) & + edot * work3d(k,j,i) & + eplus * work3d(k,j,i+1) END DO END DO END DO END DO ! !-- Interpolation in y-direction (quadratic, Clark and Farley) DO k = nzbottom, nztop+2 DO j = bdims_rem(2,1), bdims_rem(2,2) bottomy = (nyf+1)/(nyc+1) * j topy = (nyf+1)/(nyc+1) * (j+1) - 1 DO iif = nxl, nxr DO jjf = bottomy, topy eps = ( jjf * dyf + 0.5 * dyf - j * dyc - 0.5 * dyc ) / dyc alpha = ( ( dyf / dyc )**2.0 - 1.0 ) / 24.0 eminus = eps * ( eps - 1.0 ) / 2.0 + alpha edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha eplus = eps * ( eps + 1.0 ) / 2.0 + alpha ptprs(k,jjf,iif) = eminus * ptprf(k,j-1,iif) & + edot * ptprf(k,j,iif) & + eplus * ptprf(k,j+1,iif) END DO END DO END DO END DO ! !-- Interpolation in z-direction (quadratic, Clark and Farley) DO k = nzbottom+1, nztop bottomz = (dzc/dzf) * (k-1) + 1 topz = (dzc/dzf) * k DO jjf = nys, nyn DO iif = nxl, nxr DO kkf = bottomz, topz eps = ( zuf(kkf) - zuc(k) ) / dzc alpha = ( ( dzf / dzc )**2.0 - 1.0 ) / 24.0 eminus = eps * ( eps - 1.0 ) / 2.0 + alpha edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha eplus = eps * ( eps + 1.0 ) / 2.0 + alpha kh(kkf,jjf,iif) = eminus * ptprs(k-1,jjf,iif) & + edot * ptprs(k,jjf,iif) & + eplus * ptprs(k+1,jjf,iif) END DO END DO END DO END DO DO jjf = nys, nyn DO iif = nxl, nxr eps = ( zuf(nzt+1) - zuc(nztop+1) ) / dzc alpha = ( ( dzf / dzc )**2.0 - 1.0 ) / 24.0 eminus = eps * ( eps - 1.0 ) / 2.0 + alpha edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha eplus = eps * ( eps + 1.0 ) / 2.0 + alpha kh(nzt+1,jjf,iif) = eminus * ptprs(nztop,jjf,iif) & + edot * ptprs(nztop+1,jjf,iif) & + eplus * ptprs(nztop+2,jjf,iif) END DO END DO DEALLOCATE( ptprf, ptprs ) END SUBROUTINE interpolate_to_fine_kh SUBROUTINE interpolate_to_fine_km USE arrays_3d USE control_parameters USE grid_variables USE indices USE pegrid IMPLICIT NONE INTEGER(iwp) :: i INTEGER(iwp) :: j INTEGER(iwp) :: k INTEGER(iwp) :: iif INTEGER(iwp) :: jjf INTEGER(iwp) :: kkf INTEGER(iwp) :: nzbottom INTEGER(iwp) :: nztop INTEGER(iwp) :: bottomx INTEGER(iwp) :: bottomy INTEGER(iwp) :: bottomz INTEGER(iwp) :: topx INTEGER(iwp) :: topy INTEGER(iwp) :: topz REAL(wp) :: eps REAL(wp) :: alpha REAL(wp) :: eminus REAL(wp) :: edot REAL(wp) :: eplus REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: ptprs REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: ptprf nzbottom = bdims_rem (3,1) nztop = bdims_rem (3,2) ! nztop = blk_dim_rem (3,2)+1 ALLOCATE( ptprf(nzbottom:nztop+2,bdims_rem(2,1)-1:bdims_rem(2,2)+1,nxl:nxr) ) ALLOCATE( ptprs(nzbottom:nztop+2,nys:nyn,nxl:nxr) ) ! !-- Initialisation of scalar variables ! !-- Interpolation in x-direction (quadratic, Clark and Farley) DO k = nzbottom, nztop+2 DO j = bdims_rem(2,1)-1, bdims_rem(2,2)+1 DO i = bdims_rem(1,1), bdims_rem(1,2) bottomx = (nxf+1)/(nxc+1) * i topx = (nxf+1)/(nxc+1) * (i+1) - 1 DO iif = bottomx, topx eps = ( iif * dxf + 0.5 * dxf - i * dxc - 0.5 * dxc ) / dxc alpha = ( ( dxf / dxc )**2.0 - 1.0 ) / 24.0 eminus = eps * ( eps - 1.0 ) / 2.0 + alpha edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha eplus = eps * ( eps + 1.0 ) / 2.0 + alpha ptprf(k,j,iif) = eminus * work3d(k,j,i-1) & + edot * work3d(k,j,i) & + eplus * work3d(k,j,i+1) END DO END DO END DO END DO ! !-- Interpolation in y-direction (quadratic, Clark and Farley) DO k = nzbottom, nztop+2 DO j = bdims_rem(2,1), bdims_rem(2,2) bottomy = (nyf+1)/(nyc+1) * j topy = (nyf+1)/(nyc+1) * (j+1) - 1 DO iif = nxl, nxr DO jjf = bottomy, topy eps = ( jjf * dyf + 0.5 * dyf - j * dyc - 0.5 * dyc ) / dyc alpha = ( ( dyf / dyc )**2.0 - 1.0 ) / 24.0 eminus = eps * ( eps - 1.0 ) / 2.0 + alpha edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha eplus = eps * ( eps + 1.0 ) / 2.0 + alpha ptprs(k,jjf,iif) = eminus * ptprf(k,j-1,iif) & + edot * ptprf(k,j,iif) & + eplus * ptprf(k,j+1,iif) END DO END DO END DO END DO ! !-- Interpolation in z-direction (quadratic, Clark and Farley) DO k = nzbottom+1, nztop bottomz = (dzc/dzf) * (k-1) + 1 topz = (dzc/dzf) * k DO jjf = nys, nyn DO iif = nxl, nxr DO kkf = bottomz, topz eps = ( zuf(kkf) - zuc(k) ) / dzc alpha = ( ( dzf / dzc )**2.0 - 1.0 ) / 24.0 eminus = eps * ( eps - 1.0 ) / 2.0 + alpha edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha eplus = eps * ( eps + 1.0 ) / 2.0 + alpha km(kkf,jjf,iif) = eminus * ptprs(k-1,jjf,iif) & + edot * ptprs(k,jjf,iif) & + eplus * ptprs(k+1,jjf,iif) END DO END DO END DO END DO DO jjf = nys, nyn DO iif = nxl, nxr eps = ( zuf(nzt+1) - zuc(nztop+1) ) / dzc alpha = ( ( dzf / dzc )**2.0 - 1.0 ) / 24.0 eminus = eps * ( eps - 1.0 ) / 2.0 + alpha edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha eplus = eps * ( eps + 1.0 ) / 2.0 + alpha km(nzt+1,jjf,iif) = eminus * ptprs(nztop,jjf,iif) & + edot * ptprs(nztop+1,jjf,iif) & + eplus * ptprs(nztop+2,jjf,iif) END DO END DO DEALLOCATE( ptprf, ptprs ) END SUBROUTINE interpolate_to_fine_km ! SUBROUTINE interpolate_to_fine_flux ! ! ! USE arrays_3d ! USE control_parameters ! USE grid_variables ! USE indices ! USE pegrid ! ! ! IMPLICIT NONE ! ! INTEGER(iwp) :: i ! INTEGER(iwp) :: j ! INTEGER(iwp) :: iif ! INTEGER(iwp) :: jjf ! INTEGER(iwp) :: bottomx ! INTEGER(iwp) :: bottomy ! INTEGER(iwp) :: topx ! INTEGER(iwp) :: topy ! REAL(wp) :: eps ! REAL(wp) :: alpha ! REAL(wp) :: eminus ! REAL(wp) :: edot ! REAL(wp) :: eplus ! REAL(wp), DIMENSION(:,:), ALLOCATABLE :: uswspr ! REAL(wp), DIMENSION(:,:), ALLOCATABLE :: vswspr ! REAL(wp), DIMENSION(:,:), ALLOCATABLE :: tspr ! REAL(wp), DIMENSION(:,:), ALLOCATABLE :: uspr ! ! ALLOCATE( uswspr(bdims_rem(2,1)-1:bdims_rem(2,2)+1,nxl:nxr) ) ! ALLOCATE( vswspr(bdims_rem(2,1)-1:bdims_rem(2,2)+1,nxl:nxr) ) ! ALLOCATE( tspr (bdims_rem(2,1)-1:bdims_rem(2,2)+1,nxl:nxr) ) ! ALLOCATE( uspr (bdims_rem(2,1)-1:bdims_rem(2,2)+1,nxl:nxr) ) ! ! ! ! !-- Initialisation of scalar variables (2D) ! ! ! ! !-- Interpolation in x-direction (quadratic, Clark and Farley) ! ! DO j = bdims_rem(2,1)-1, bdims_rem(2,2)+1 ! DO i = bdims_rem(1,1), bdims_rem(1,2) ! ! bottomx = (nxf+1)/(nxc+1) * i ! topx = (nxf+1)/(nxc+1) * (i+1) - 1 ! ! DO iif = bottomx, topx ! ! eps = ( iif * dxf + 0.5 * dxf - i * dxc - 0.5 * dxc ) / dxc ! alpha = ( ( dxf / dxc )**2.0 - 1.0 ) / 24.0 ! eminus = eps * ( eps - 1.0 ) / 2.0 + alpha ! edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha ! eplus = eps * ( eps + 1.0 ) / 2.0 + alpha ! ! uswspr(j,iif) = eminus * work2dusws(j,i-1) & ! + edot * work2dusws(j,i) & ! + eplus * work2dusws(j,i+1) ! ! vswspr(j,iif) = eminus * work2dvsws(j,i-1) & ! + edot * work2dvsws(j,i) & ! + eplus * work2dvsws(j,i+1) ! ! tspr(j,iif) = eminus * work2dts(j,i-1) & ! + edot * work2dts(j,i) & ! + eplus * work2dts(j,i+1) ! ! uspr(j,iif) = eminus * work2dus(j,i-1) & ! + edot * work2dus(j,i) & ! + eplus * work2dus(j,i+1) ! ! END DO ! ! END DO ! END DO ! ! ! ! !-- Interpolation in y-direction (quadratic, Clark and Farley) ! ! DO j = bdims_rem(2,1), bdims_rem(2,2) ! ! bottomy = (nyf+1)/(nyc+1) * j ! topy = (nyf+1)/(nyc+1) * (j+1) - 1 ! ! DO iif = nxl, nxr ! DO jjf = bottomy, topy ! ! eps = ( jjf * dyf + 0.5 * dyf - j * dyc - 0.5 * dyc ) / dyc ! alpha = ( ( dyf / dyc )**2.0 - 1.0 ) / 24.0 ! eminus = eps * ( eps - 1.0 ) / 2.0 + alpha ! edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha ! eplus = eps * ( eps + 1.0 ) / 2.0 + alpha ! !! !!-- TODO !-- variables are not compatible with the new surface layer module ! ! surf_def_h(0)%usws(jjf,iif) = eminus * uswspr(j-1,if) & ! + edot * uswspr(j,iif) & ! + eplus * uswspr(j+1,iif) ! ! surf_def_h(0)%vsws(jjf,iif) = eminus * vswspr(j-1,if) & ! + edot * vswspr(j,iif) & ! + eplus * vswspr(j+1,iif) ! ! ts(jjf,iif) = eminus * tspr(j-1,if) & ! + edot * tspr(j,iif) & ! + eplus * tspr(j+1,iif) ! ! us(jjf,iif) = eminus * uspr(j-1,if) & ! + edot * uspr(j,iif) & ! + eplus * uspr(j+1,iif) ! ! END DO ! END DO ! ! END DO ! ! ! DEALLOCATE( uswspr, vswspr ) ! DEALLOCATE( tspr, uspr ) ! ! ! END SUBROUTINE interpolate_to_fine_flux #endif END SUBROUTINE vnest_init_fine SUBROUTINE vnest_boundary_conds #if defined( __parallel ) !------------------------------------------------------------------------------! ! Description: ! ------------ ! Boundary conditions for the prognostic quantities. ! One additional bottom boundary condition is applied for the TKE (=(u*)**2) ! in prandtl_fluxes. The cyclic lateral boundary conditions are implicitly ! handled in routine exchange_horiz. Pressure boundary conditions are ! explicitly set in routines pres, poisfft, poismg and sor. !------------------------------------------------------------------------------! USE arrays_3d USE control_parameters USE grid_variables USE indices USE pegrid IMPLICIT NONE INTEGER(iwp) :: i INTEGER(iwp) :: j INTEGER(iwp) :: iif INTEGER(iwp) :: jjf ! !-- vnest: top boundary conditions IF ( coupling_mode == 'vnested_crse' ) THEN !-- Send data to fine grid for TOP BC offset(1) = ( pdims_partner(1) / pdims(1) ) * pcoord(1) offset(2) = ( pdims_partner(2) / pdims(2) ) * pcoord(2) do j = 0, ( pdims_partner(2) / pdims(2) ) - 1 do i = 0, ( pdims_partner(1) / pdims(1) ) - 1 map_coord(1) = i+offset(1) map_coord(2) = j+offset(2) target_idex = f_rnk_lst(map_coord(1),map_coord(2)) + numprocs bdims (1,1) = c2f_dims_cg (0,map_coord(1),map_coord(2)) bdims (1,2) = c2f_dims_cg (1,map_coord(1),map_coord(2)) bdims (2,1) = c2f_dims_cg (2,map_coord(1),map_coord(2)) bdims (2,2) = c2f_dims_cg (3,map_coord(1),map_coord(2)) bdims (3,1) = c2f_dims_cg (4,map_coord(1),map_coord(2)) bdims (3,2) = c2f_dims_cg (5,map_coord(1),map_coord(2)) n_cell_c = ( (bdims(1,2)-bdims(1,1)) + 3 ) * & ( (bdims(2,2)-bdims(2,1)) + 3 ) * & ( (bdims(3,2)-bdims(3,1)) + 1 ) CALL MPI_SEND(u (bdims(3,1), bdims(2,1)-1, bdims(1,1)-1), & 1, TYPE_VNEST_BC, target_idex, & 201, comm_inter, ierr) CALL MPI_SEND(v(bdims(3,1), bdims(2,1)-1, bdims(1,1)-1),& 1, TYPE_VNEST_BC, target_idex, & 202, comm_inter, ierr) CALL MPI_SEND(w(bdims(3,1), bdims(2,1)-1, bdims(1,1)-1),& 1, TYPE_VNEST_BC, target_idex, & 203, comm_inter, ierr) CALL MPI_SEND(pt(bdims(3,1), bdims(2,1)-1, bdims(1,1)-1),& 1, TYPE_VNEST_BC, target_idex, & 205, comm_inter, ierr) IF ( humidity ) THEN CALL MPI_SEND(q(bdims(3,1), bdims(2,1)-1, bdims(1,1)-1),& 1, TYPE_VNEST_BC, target_idex, & 209, comm_inter, ierr) ENDIF end do end do ELSEIF ( coupling_mode == 'vnested_fine' ) THEN !-- Receive data from coarse grid for TOP BC offset(1) = pcoord(1) / ( pdims(1)/pdims_partner(1) ) offset(2) = pcoord(2) / ( pdims(2)/pdims_partner(2) ) map_coord(1) = offset(1) map_coord(2) = offset(2) target_idex = c_rnk_lst(map_coord(1),map_coord(2)) bdims_rem (1,1) = c2f_dims_fg(0) bdims_rem (1,2) = c2f_dims_fg(1) bdims_rem (2,1) = c2f_dims_fg(2) bdims_rem (2,2) = c2f_dims_fg(3) bdims_rem (3,1) = c2f_dims_fg(4) bdims_rem (3,2) = c2f_dims_fg(5) n_cell_c = & ( (bdims_rem(1,2)-bdims_rem(1,1)) + 3 ) * & ( (bdims_rem(2,2)-bdims_rem(2,1)) + 3 ) * & ( (bdims_rem(3,2)-bdims_rem(3,1)) + 1 ) ALLOCATE( work3d ( & bdims_rem(3,1) :bdims_rem(3,2) , & bdims_rem(2,1)-1:bdims_rem(2,2)+1, & bdims_rem(1,1)-1:bdims_rem(1,2)+1)) CALL MPI_RECV( work3d ,n_cell_c, MPI_REAL, target_idex, 201, & comm_inter,status, ierr ) interpol3d => u call vnest_set_topbc_u CALL MPI_RECV( work3d ,n_cell_c, MPI_REAL, target_idex, 202, & comm_inter,status, ierr ) interpol3d => v call vnest_set_topbc_v CALL MPI_RECV( work3d ,n_cell_c, MPI_REAL, target_idex, 203, & comm_inter,status, ierr ) interpol3d => w call vnest_set_topbc_w CALL MPI_RECV( work3d,n_cell_c, MPI_REAL, target_idex, 205, & comm_inter,status, ierr ) interpol3d => pt call vnest_set_topbc_s IF ( humidity ) THEN CALL MPI_RECV( work3d,n_cell_c, MPI_REAL, target_idex, 209, & comm_inter,status, ierr ) interpol3d => q call vnest_set_topbc_s CALL exchange_horiz_2d(q (nzt+1,:,:) ) ENDIF !-- TKE Neumann BC for FG top DO jjf = nys, nyn DO iif = nxl, nxr e(nzt+1,jjf,iif) = e(nzt,jjf,iif) END DO END DO ! !-- w level nzt+1 does not impact results. Only to avoid jumps while !-- plotting profiles w(nzt+1,:,:) = w(nzt,:,:) CALL exchange_horiz_2d(u (nzt+1,:,:) ) CALL exchange_horiz_2d(v (nzt+1,:,:) ) CALL exchange_horiz_2d(pt(nzt+1,:,:) ) CALL exchange_horiz_2d(e (nzt+1,:,:) ) CALL exchange_horiz_2d(w (nzt+1,:,:) ) CALL exchange_horiz_2d(w (nzt ,:,:) ) NULLIFY ( interpol3d ) DEALLOCATE ( work3d ) ENDIF CONTAINS SUBROUTINE vnest_set_topbc_w USE arrays_3d USE control_parameters USE grid_variables USE indices USE pegrid IMPLICIT NONE INTEGER(iwp) :: i INTEGER(iwp) :: j INTEGER(iwp) :: iif INTEGER(iwp) :: jjf INTEGER(iwp) :: bottomx INTEGER(iwp) :: bottomy INTEGER(iwp) :: topx INTEGER(iwp) :: topy REAL(wp) :: eps REAL(wp) :: alpha REAL(wp) :: eminus REAL(wp) :: edot REAL(wp) :: eplus REAL(wp), DIMENSION(:,:), ALLOCATABLE :: wprf ALLOCATE( wprf(bdims_rem(2,1)-1:bdims_rem(2,2)+1,nxl:nxr) ) ! !-- Determination of a boundary condition for the vertical velocity component w: !-- In this case only interpolation in x- and y- direction is necessary, as the !-- boundary w-node of the fine grid coincides with a w-node in the coarse grid. !-- For both interpolations the scheme of Clark and Farley is used. ! !-- Interpolation in x-direction DO j = bdims_rem(2,1)-1, bdims_rem(2,2)+1 DO i = bdims_rem(1,1), bdims_rem(1,2) bottomx = (nxf+1)/(nxc+1) * i topx = (nxf+1)/(nxc+1) * (i+1) - 1 DO iif = bottomx, topx eps = (iif * dxf + 0.5 * dxf - i * dxc - 0.5 * dxc) / dxc alpha = ( (dxf/dxc)**2.0 - 1.0) / 24.0 eminus = eps * ( eps - 1.0 ) / 2.0 + alpha edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha eplus = eps * ( eps + 1.0 ) / 2.0 + alpha wprf(j,iif) = eminus * work3d(bdims_rem(3,1),j,i-1) & + edot * work3d(bdims_rem(3,1),j,i) & + eplus * work3d(bdims_rem(3,1),j,i+1) END DO END DO END DO ! !-- Interpolation in y-direction DO j = bdims_rem(2,1), bdims_rem(2,2) bottomy = (nyf+1)/(nyc+1) * j topy = (nyf+1)/(nyc+1) * (j+1) - 1 DO iif = nxl, nxr DO jjf = bottomy, topy eps = (jjf * dyf + 0.5 * dyf - j * dyc - 0.5 * dyc) / dyc alpha = ( (dyf/dyc)**2.0 - 1.0) / 24.0 eminus = eps * ( eps - 1.0 ) / 2.0 + alpha edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha eplus = eps * ( eps + 1.0 ) / 2.0 + alpha w(nzt,jjf,iif) = eminus * wprf(j-1,iif) & + edot * wprf(j,iif) & + eplus * wprf(j+1,iif) END DO END DO END DO DEALLOCATE( wprf ) END SUBROUTINE vnest_set_topbc_w SUBROUTINE vnest_set_topbc_u USE arrays_3d USE control_parameters USE grid_variables USE indices USE pegrid IMPLICIT NONE INTEGER(iwp) :: i INTEGER(iwp) :: j INTEGER(iwp) :: k INTEGER(iwp) :: iif INTEGER(iwp) :: jjf INTEGER(iwp) :: bottomx INTEGER(iwp) :: bottomy INTEGER(iwp) :: topx INTEGER(iwp) :: topy REAL(wp) :: eps REAL(wp) :: alpha REAL(wp) :: eminus REAL(wp) :: edot REAL(wp) :: eplus REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: uprf REAL(wp), DIMENSION(:,:), ALLOCATABLE :: uprs ALLOCATE( uprf(bdims_rem(3,1):bdims_rem(3,2),nys:nyn,bdims_rem(1,1)-1:bdims_rem(1,2)+1) ) ALLOCATE( uprs(nys:nyn,bdims_rem(1,1)-1:bdims_rem(1,2)+1) ) ! !-- Interpolation in y-direction DO k = bdims_rem(3,1), bdims_rem(3,2) DO j = bdims_rem(2,1), bdims_rem(2,2) bottomy = (nyf+1)/(nyc+1) * j topy = (nyf+1)/(nyc+1) * (j+1) - 1 DO i = bdims_rem(1,1)-1, bdims_rem(1,2)+1 DO jjf = bottomy, topy eps = (jjf * dyf + 0.5 * dyf - j * dyc - 0.5 * dyc) / dyc alpha = ( (dyf/dyc)**2.0 - 1.0) / 24.0 eminus = eps * ( eps - 1.0 ) / 2.0 + alpha edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha eplus = eps * ( eps + 1.0 ) / 2.0 + alpha uprf(k,jjf,i) = eminus * work3d(k,j-1,i) & + edot * work3d(k,j,i) & + eplus * work3d(k,j+1,i) END DO END DO END DO END DO ! !-- Interpolation in z-direction DO jjf = nys, nyn DO i = bdims_rem(1,1)-1, bdims_rem(1,2)+1 eps = ( zuf(nzt+1) - zuc(bdims_rem(3,1)+1) ) / dzc alpha = ( (dzf/dzc)**2.0 - 1.0) / 24.0 eminus = eps * ( eps - 1.0 ) / 2.0 + alpha edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha eplus = eps * ( eps + 1.0 ) / 2.0 + alpha uprs(jjf,i) = eminus * uprf(bdims_rem(3,1),jjf,i) & + edot * uprf(bdims_rem(3,1)+1,jjf,i) & + eplus * uprf(bdims_rem(3,1)+2,jjf,i) END DO END DO ! !-- Interpolation in x-direction DO jjf = nys, nyn DO i = bdims_rem(1,1), bdims_rem(1,2) bottomx = (nxf+1)/(nxc+1) * i topx = (nxf+1)/(nxc+1) * (i+1) - 1 DO iif = bottomx, topx u(nzt+1,jjf,iif) = uprs(jjf,i) + ( iif * dxf - i * dxc ) * ( uprs(jjf,i+1) - uprs(jjf,i) ) / dxc END DO END DO END DO DEALLOCATE ( uprf, uprs ) END SUBROUTINE vnest_set_topbc_u SUBROUTINE vnest_set_topbc_v USE arrays_3d USE control_parameters USE grid_variables USE indices USE pegrid IMPLICIT NONE INTEGER(iwp) :: i INTEGER(iwp) :: j INTEGER(iwp) :: k INTEGER(iwp) :: iif INTEGER(iwp) :: jjf INTEGER(iwp) :: bottomx INTEGER(iwp) :: bottomy INTEGER(iwp) :: topx INTEGER(iwp) :: topy REAL(wp) :: eps REAL(wp) :: alpha REAL(wp) :: eminus REAL(wp) :: edot REAL(wp) :: eplus REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: vprf REAL(wp), DIMENSION(:,:), ALLOCATABLE :: vprs ALLOCATE( vprf(bdims_rem(3,1):bdims_rem(3,2),bdims_rem(2,1)-1:bdims_rem(2,2)+1,nxl:nxr) ) ALLOCATE( vprs(bdims_rem(2,1)-1:bdims_rem(2,2)+1,nxl:nxr) ) ! !-- Determination of a boundary condition for the horizontal velocity component v: !-- Interpolation in x- and z-direction is carried out by using the scheme, !-- which was derived by Clark and Farley (1984). In y-direction a !-- linear interpolation is carried out. ! !-- Interpolation in x-direction DO k = bdims_rem(3,1), bdims_rem(3,2) DO j = bdims_rem(2,1)-1, bdims_rem(2,2)+1 DO i = bdims_rem(1,1), bdims_rem(1,2) bottomx = (nxf+1)/(nxc+1) * i topx = (nxf+1)/(nxc+1) * (i+1) - 1 DO iif = bottomx, topx eps = (iif * dxf + 0.5 * dxf - i * dxc - 0.5 * dxc) / dxc alpha = ( (dxf/dxc)**2.0 - 1.0) / 24.0 eminus = eps * ( eps - 1.0 ) / 2.0 + alpha edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha eplus = eps * ( eps + 1.0 ) / 2.0 + alpha vprf(k,j,iif) = eminus * work3d(k,j,i-1) & + edot * work3d(k,j,i) & + eplus * work3d(k,j,i+1) END DO END DO END DO END DO ! !-- Interpolation in z-direction DO j = bdims_rem(2,1)-1, bdims_rem(2,2)+1 DO iif = nxl, nxr eps = ( zuf(nzt+1) - zuc(bdims_rem(3,1)+1) ) / dzc alpha = ( (dzf/dzc)**2.0 - 1.0) / 24.0 eminus = eps * ( eps - 1.0 ) / 2.0 + alpha edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha eplus = eps * ( eps + 1.0 ) / 2.0 + alpha vprs(j,iif) = eminus * vprf(bdims_rem(3,1),j,iif) & + edot * vprf(bdims_rem(3,1)+1,j,iif) & + eplus * vprf(bdims_rem(3,1)+2,j,iif) END DO END DO ! !-- Interpolation in y-direction DO j = bdims_rem(2,1), bdims_rem(2,2) DO iif = nxl, nxr bottomy = (nyf+1)/(nyc+1) * j topy = (nyf+1)/(nyc+1) * (j+1) - 1 DO jjf = bottomy, topy v(nzt+1,jjf,iif) = vprs(j,iif) + ( jjf * dyf - j * dyc ) * ( vprs(j+1,iif) - vprs(j,iif) ) / dyc END DO END DO END DO DEALLOCATE ( vprf, vprs) END SUBROUTINE vnest_set_topbc_v SUBROUTINE vnest_set_topbc_s USE arrays_3d USE control_parameters USE grid_variables USE indices USE pegrid IMPLICIT NONE INTEGER(iwp) :: i INTEGER(iwp) :: j INTEGER(iwp) :: k INTEGER(iwp) :: iif INTEGER(iwp) :: jjf INTEGER(iwp) :: bottomx INTEGER(iwp) :: bottomy INTEGER(iwp) :: topx INTEGER(iwp) :: topy REAL(wp) :: eps REAL(wp) :: alpha REAL(wp) :: eminus REAL(wp) :: edot REAL(wp) :: eplus REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: ptprf REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: ptprs ALLOCATE( ptprf(bdims_rem(3,1):bdims_rem(3,2),bdims_rem(2,1)-1:bdims_rem(2,2)+1,nxl:nxr) ) ALLOCATE( ptprs(bdims_rem(3,1):bdims_rem(3,2),nys:nyn,nxl:nxr) ) ! !-- Determination of a boundary condition for the potential temperature pt: !-- The scheme derived by Clark and Farley can be used in all three dimensions. ! !-- Interpolation in x-direction DO k = bdims_rem(3,1), bdims_rem(3,2) DO j = bdims_rem(2,1)-1, bdims_rem(2,2)+1 DO i = bdims_rem(1,1), bdims_rem(1,2) bottomx = (nxf+1)/(nxc+1) * i topx = (nxf+1)/(nxc+1) *(i+1) - 1 DO iif = bottomx, topx eps = (iif * dxf + 0.5 * dxf - i * dxc - 0.5 * dxc) / dxc alpha = ( (dxf/dxc)**2.0 - 1.0) / 24.0 eminus = eps * (eps - 1.0 ) / 2.0 + alpha edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha eplus = eps * ( eps + 1.0 ) / 2.0 + alpha ptprf(k,j,iif) = eminus * work3d(k,j,i-1) & + edot * work3d(k,j,i) & + eplus * work3d(k,j,i+1) END DO END DO END DO END DO ! !-- Interpolation in y-direction DO k = bdims_rem(3,1), bdims_rem(3,2) DO j = bdims_rem(2,1), bdims_rem(2,2) bottomy = (nyf+1)/(nyc+1) * j topy = (nyf+1)/(nyc+1) * (j+1) - 1 DO iif = nxl, nxr DO jjf = bottomy, topy eps = (jjf * dyf + 0.5 * dyf - j * dyc - 0.5 * dyc) / dyc alpha = ( (dyf/dyc)**2.0 - 1.0) / 24.0 eminus = eps * (eps - 1.0) / 2.0 + alpha edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha eplus = eps * ( eps + 1.0 ) / 2.0 + alpha ptprs(k,jjf,iif) = eminus * ptprf(k,j-1,iif) & + edot * ptprf(k,j,iif) & + eplus * ptprf(k,j+1,iif) END DO END DO END DO END DO ! !-- Interpolation in z-direction DO jjf = nys, nyn DO iif = nxl, nxr eps = ( zuf(nzt+1) - zuc(bdims_rem(3,1)+1) ) / dzc alpha = ( (dzf/dzc)**2.0 - 1.0) / 24.0 eminus = eps * ( eps - 1.0 ) / 2.0 + alpha edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha eplus = eps * ( eps + 1.0 ) / 2.0 + alpha interpol3d (nzt+1,jjf,iif) = eminus * ptprs(bdims_rem(3,1),jjf,iif) & + edot * ptprs(bdims_rem(3,1)+1,jjf,iif) & + eplus * ptprs(bdims_rem(3,1)+2,jjf,iif) END DO END DO DEALLOCATE ( ptprf, ptprs ) END SUBROUTINE vnest_set_topbc_s #endif END SUBROUTINE vnest_boundary_conds SUBROUTINE vnest_boundary_conds_khkm #if defined( __parallel ) !--------------------------------------------------------------------------------! ! Description: ! ------------ ! Boundary conditions for the prognostic quantities. ! One additional bottom boundary condition is applied for the TKE (=(u*)**2) ! in prandtl_fluxes. The cyclic lateral boundary conditions are implicitly ! handled in routine exchange_horiz. Pressure boundary conditions are ! explicitly set in routines pres, poisfft, poismg and sor. !------------------------------------------------------------------------------! USE arrays_3d USE control_parameters USE grid_variables USE indices USE pegrid IMPLICIT NONE INTEGER(iwp) :: i INTEGER(iwp) :: j INTEGER(iwp) :: iif INTEGER(iwp) :: jjf IF ( coupling_mode == 'vnested_crse' ) THEN ! Send data to fine grid for TOP BC offset(1) = ( pdims_partner(1) / pdims(1) ) * pcoord(1) offset(2) = ( pdims_partner(2) / pdims(2) ) * pcoord(2) do j = 0, ( pdims_partner(2) / pdims(2) ) - 1 do i = 0, ( pdims_partner(1) / pdims(1) ) - 1 map_coord(1) = i+offset(1) map_coord(2) = j+offset(2) target_idex = f_rnk_lst(map_coord(1),map_coord(2)) + numprocs CALL MPI_RECV( bdims_rem, 6, MPI_INTEGER, target_idex, 10, & comm_inter,status, ierr ) bdims (1,1) = bdims_rem (1,1) / cfratio(1) bdims (1,2) = bdims_rem (1,2) / cfratio(1) bdims (2,1) = bdims_rem (2,1) / cfratio(2) bdims (2,2) = bdims_rem (2,2) / cfratio(2) bdims (3,1) = bdims_rem (3,2) / cfratio(3) bdims (3,2) = bdims (3,1) + 2 CALL MPI_SEND( bdims, 6, MPI_INTEGER, target_idex, 9, & comm_inter, ierr ) n_cell_c = ( (bdims(1,2)-bdims(1,1)) + 3 ) * & ( (bdims(2,2)-bdims(2,1)) + 3 ) * & ( (bdims(3,2)-bdims(3,1)) + 1 ) CALL MPI_SEND(kh(bdims(3,1) :bdims(3,2) , & bdims(2,1)-1:bdims(2,2)+1, & bdims(1,1)-1:bdims(1,2)+1),& n_cell_c, MPI_REAL, target_idex, & 207, comm_inter, ierr) CALL MPI_SEND(km(bdims(3,1) :bdims(3,2) , & bdims(2,1)-1:bdims(2,2)+1, & bdims(1,1)-1:bdims(1,2)+1),& n_cell_c, MPI_REAL, target_idex, & 208, comm_inter, ierr) end do end do ELSEIF ( coupling_mode == 'vnested_fine' ) THEN ! Receive data from coarse grid for TOP BC offset(1) = pcoord(1) / ( pdims(1)/pdims_partner(1) ) offset(2) = pcoord(2) / ( pdims(2)/pdims_partner(2) ) map_coord(1) = offset(1) map_coord(2) = offset(2) target_idex = c_rnk_lst(map_coord(1),map_coord(2)) bdims (1,1) = nxl bdims (1,2) = nxr bdims (2,1) = nys bdims (2,2) = nyn bdims (3,1) = nzb bdims (3,2) = nzt CALL MPI_SEND( bdims, 6, MPI_INTEGER, target_idex, 10, & comm_inter, ierr ) CALL MPI_RECV( bdims_rem, 6, MPI_INTEGER, target_idex, 9, & comm_inter,status, ierr ) n_cell_c = ( (bdims_rem(1,2)-bdims_rem(1,1)) + 3 ) * & ( (bdims_rem(2,2)-bdims_rem(2,1)) + 3 ) * & ( (bdims_rem(3,2)-bdims_rem(3,1)) + 1 ) ALLOCATE( work3d ( bdims_rem(3,1) :bdims_rem(3,2) , & bdims_rem(2,1)-1:bdims_rem(2,2)+1, & bdims_rem(1,1)-1:bdims_rem(1,2)+1)) CALL MPI_RECV( work3d,n_cell_c, MPI_REAL, target_idex, 207, & comm_inter,status, ierr ) ! Neumann BC for FG kh DO jjf = nys, nyn DO iif = nxl, nxr kh(nzt+1,jjf,iif) = kh(nzt,jjf,iif) END DO END DO CALL MPI_RECV( work3d,n_cell_c, MPI_REAL, target_idex, 208, & comm_inter,status, ierr ) ! Neumann BC for FG kh DO jjf = nys, nyn DO iif = nxl, nxr km(nzt+1,jjf,iif) = km(nzt,jjf,iif) END DO END DO ! !-- The following evaluation can only be performed, if the fine grid is situated below the inversion !! DO jjf = nys-1, nyn+1 !! DO iif = nxl-1, nxr+1 !! !! km(nzt+1,jjf,iif) = 0.1 * l_grid(nzt+1) * SQRT( e(nzt+1,jjf,iif) ) !! kh(nzt+1,jjf,iif) = 3.0 * km(nzt+1,jjf,iif) !! !! END DO !! END DO CALL exchange_horiz_2d(km(nzt+1,:,:) ) CALL exchange_horiz_2d(kh(nzt+1,:,:) ) DEALLOCATE ( work3d ) ENDIF ! CONTAINS ! ! SUBROUTINE vnest_set_topbc_kh ! ! ! USE arrays_3d ! USE control_parameters ! USE grid_variables ! USE indices ! USE pegrid ! ! ! IMPLICIT NONE ! ! INTEGER(iwp) :: i ! INTEGER(iwp) :: j ! INTEGER(iwp) :: k ! INTEGER(iwp) :: iif ! INTEGER(iwp) :: jjf ! INTEGER(iwp) :: bottomx ! INTEGER(iwp) :: bottomy ! INTEGER(iwp) :: topx ! INTEGER(iwp) :: topy ! REAL(wp) :: eps ! REAL(wp) :: alpha ! REAL(wp) :: eminus ! REAL(wp) :: edot ! REAL(wp) :: eplus ! REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: ptprf ! REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: ptprs ! ! ! ! ALLOCATE( ptprf(bdims_rem(3,1):bdims_rem(3,2),bdims_rem(2,1)-1:bdims_rem(2,2)+1,nxl:nxr) ) ! ALLOCATE( ptprs(bdims_rem(3,1):bdims_rem(3,2),nys:nyn,nxl:nxr) ) ! ! ! ! !-- Determination of a boundary condition for the potential temperature pt: ! !-- The scheme derived by Clark and Farley can be used in all three dimensions. ! ! ! ! !-- Interpolation in x-direction ! ! DO k = bdims_rem(3,1), bdims_rem(3,2) ! ! DO j = bdims_rem(2,1)-1, bdims_rem(2,2)+1 ! ! DO i = bdims_rem(1,1), bdims_rem(1,2) ! ! bottomx = (nxf+1)/(nxc+1) * i ! topx = (nxf+1)/(nxc+1) *(i+1) - 1 ! ! DO iif = bottomx, topx ! ! eps = (iif * dxf + 0.5 * dxf - i * dxc - 0.5 * dxc) / dxc ! ! alpha = ( (dxf/dxc)**2.0 - 1.0) / 24.0 ! ! eminus = eps * (eps - 1.0 ) / 2.0 + alpha ! ! edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha ! ! eplus = eps * ( eps + 1.0 ) / 2.0 + alpha ! ! ptprf(k,j,iif) = eminus * work3d(k,j,i-1) & ! + edot * work3d(k,j,i) & ! + eplus * work3d(k,j,i+1) ! END DO ! ! END DO ! ! END DO ! ! END DO ! ! ! ! !-- Interpolation in y-direction ! ! DO k = bdims_rem(3,1), bdims_rem(3,2) ! ! DO j = bdims_rem(2,1), bdims_rem(2,2) ! ! bottomy = (nyf+1)/(nyc+1) * j ! topy = (nyf+1)/(nyc+1) * (j+1) - 1 ! ! DO iif = nxl, nxr ! ! DO jjf = bottomy, topy ! ! eps = (jjf * dyf + 0.5 * dyf - j * dyc - 0.5 * dyc) / dyc ! ! alpha = ( (dyf/dyc)**2.0 - 1.0) / 24.0 ! ! eminus = eps * (eps - 1.0) / 2.0 + alpha ! ! edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha ! ! eplus = eps * ( eps + 1.0 ) / 2.0 + alpha ! ! ptprs(k,jjf,iif) = eminus * ptprf(k,j-1,iif) & ! + edot * ptprf(k,j,iif) & ! + eplus * ptprf(k,j+1,iif) ! END DO ! ! END DO ! ! END DO ! ! END DO ! ! ! ! !-- Interpolation in z-direction ! ! DO jjf = nys, nyn ! DO iif = nxl, nxr ! ! eps = ( zuf(nzt+1) - zuc(bdims_rem(3,1)+1) ) / dzc ! ! alpha = ( (dzf/dzc)**2.0 - 1.0) / 24.0 ! ! eminus = eps * ( eps - 1.0 ) / 2.0 + alpha ! ! edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha ! ! eplus = eps * ( eps + 1.0 ) / 2.0 + alpha ! ! kh (nzt+1,jjf,iif) = eminus * ptprs(bdims_rem(3,1),jjf,iif) & ! + edot * ptprs(bdims_rem(3,1)+1,jjf,iif) & ! + eplus * ptprs(bdims_rem(3,1)+2,jjf,iif) ! ! END DO ! END DO ! ! DEALLOCATE ( ptprf, ptprs ) ! ! ! ! END SUBROUTINE vnest_set_topbc_kh ! SUBROUTINE vnest_set_topbc_km ! ! ! USE arrays_3d ! USE control_parameters ! USE grid_variables ! USE indices ! USE pegrid ! ! ! IMPLICIT NONE ! ! INTEGER(iwp) :: i ! INTEGER(iwp) :: j ! INTEGER(iwp) :: k ! INTEGER(iwp) :: iif ! INTEGER(iwp) :: jjf ! INTEGER(iwp) :: bottomx ! INTEGER(iwp) :: bottomy ! INTEGER(iwp) :: topx ! INTEGER(iwp) :: topy ! REAL(wp) :: eps ! REAL(wp) :: alpha ! REAL(wp) :: eminus ! REAL(wp) :: edot ! REAL(wp) :: eplus ! REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: ptprf ! REAL(wp), DIMENSION(:,:,:), ALLOCATABLE :: ptprs ! ! ! ! ALLOCATE( ptprf(bdims_rem(3,1):bdims_rem(3,2),bdims_rem(2,1)-1:bdims_rem(2,2)+1,nxl:nxr) ) ! ALLOCATE( ptprs(bdims_rem(3,1):bdims_rem(3,2),nys:nyn,nxl:nxr) ) ! ! ! ! !-- Determination of a boundary condition for the potential temperature pt: ! !-- The scheme derived by Clark and Farley can be used in all three dimensions. ! ! ! ! !-- Interpolation in x-direction ! ! DO k = bdims_rem(3,1), bdims_rem(3,2) ! ! DO j = bdims_rem(2,1)-1, bdims_rem(2,2)+1 ! ! DO i = bdims_rem(1,1), bdims_rem(1,2) ! ! bottomx = (nxf+1)/(nxc+1) * i ! topx = (nxf+1)/(nxc+1) *(i+1) - 1 ! ! DO iif = bottomx, topx ! ! eps = (iif * dxf + 0.5 * dxf - i * dxc - 0.5 * dxc) / dxc ! ! alpha = ( (dxf/dxc)**2.0 - 1.0) / 24.0 ! ! eminus = eps * (eps - 1.0 ) / 2.0 + alpha ! ! edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha ! ! eplus = eps * ( eps + 1.0 ) / 2.0 + alpha ! ! ptprf(k,j,iif) = eminus * work3d(k,j,i-1) & ! + edot * work3d(k,j,i) & ! + eplus * work3d(k,j,i+1) ! END DO ! ! END DO ! ! END DO ! ! END DO ! ! ! ! !-- Interpolation in y-direction ! ! DO k = bdims_rem(3,1), bdims_rem(3,2) ! ! DO j = bdims_rem(2,1), bdims_rem(2,2) ! ! bottomy = (nyf+1)/(nyc+1) * j ! topy = (nyf+1)/(nyc+1) * (j+1) - 1 ! ! DO iif = nxl, nxr ! ! DO jjf = bottomy, topy ! ! eps = (jjf * dyf + 0.5 * dyf - j * dyc - 0.5 * dyc) / dyc ! ! alpha = ( (dyf/dyc)**2.0 - 1.0) / 24.0 ! ! eminus = eps * (eps - 1.0) / 2.0 + alpha ! ! edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha ! ! eplus = eps * ( eps + 1.0 ) / 2.0 + alpha ! ! ptprs(k,jjf,iif) = eminus * ptprf(k,j-1,iif) & ! + edot * ptprf(k,j,iif) & ! + eplus * ptprf(k,j+1,iif) ! END DO ! ! END DO ! ! END DO ! ! END DO ! ! ! ! !-- Interpolation in z-direction ! ! DO jjf = nys, nyn ! DO iif = nxl, nxr ! ! eps = ( zuf(nzt+1) - zuc(bdims_rem(3,1)+1) ) / dzc ! ! alpha = ( (dzf/dzc)**2.0 - 1.0) / 24.0 ! ! eminus = eps * ( eps - 1.0 ) / 2.0 + alpha ! ! edot = ( 1.0 - eps**2.0 ) - 2.0 * alpha ! ! eplus = eps * ( eps + 1.0 ) / 2.0 + alpha ! ! km (nzt+1,jjf,iif) = eminus * ptprs(bdims_rem(3,1),jjf,iif) & ! + edot * ptprs(bdims_rem(3,1)+1,jjf,iif) & ! + eplus * ptprs(bdims_rem(3,1)+2,jjf,iif) ! ! END DO ! END DO ! ! DEALLOCATE ( ptprf, ptprs ) ! ! ! ! END SUBROUTINE vnest_set_topbc_km #endif END SUBROUTINE vnest_boundary_conds_khkm SUBROUTINE vnest_anterpolate #if defined( __parallel ) !--------------------------------------------------------------------------------! ! Description: ! ------------ ! Anterpolate data from fine grid to coarse grid. !------------------------------------------------------------------------------! USE arrays_3d USE control_parameters USE grid_variables USE indices USE interfaces USE pegrid USE surface_mod, & ONLY : bc_h IMPLICIT NONE REAL(wp) :: time_since_reference_point_rem INTEGER(iwp) :: i INTEGER(iwp) :: j INTEGER(iwp) :: k INTEGER(iwp) :: l !< running index boundary type, for up- and downward-facing walls INTEGER(iwp) :: m !< running index surface elements ! !-- In case of model termination initiated by the remote model !-- (terminate_coupled_remote > 0), initiate termination of the local model. !-- The rest of the coupler must then be skipped because it would cause an MPI !-- intercomminucation hang. !-- If necessary, the coupler will be called at the beginning of the next !-- restart run. IF ( myid == 0) THEN CALL MPI_SENDRECV( terminate_coupled, 1, MPI_INTEGER, & target_id, 0, & terminate_coupled_remote, 1, MPI_INTEGER, & target_id, 0, & comm_inter, status, ierr ) ENDIF CALL MPI_BCAST( terminate_coupled_remote, 1, MPI_INTEGER, 0, comm2d, & ierr ) IF ( terminate_coupled_remote > 0 ) THEN WRITE( message_string, * ) 'remote model "', & TRIM( coupling_mode_remote ), & '" terminated', & '&with terminate_coupled_remote = ', & terminate_coupled_remote, & '&local model "', TRIM( coupling_mode ), & '" has', & '&terminate_coupled = ', & terminate_coupled CALL message( 'vnest_anterpolate', 'PA0310', 1, 2, 0, 6, 0 ) RETURN ENDIF ! !-- Exchange the current simulated time between the models IF ( myid == 0 ) THEN CALL MPI_SEND( time_since_reference_point, 1, MPI_REAL, target_id, & 11, comm_inter, ierr ) CALL MPI_RECV( time_since_reference_point_rem, 1, MPI_REAL, & target_id, 11, comm_inter, status, ierr ) ENDIF CALL MPI_BCAST( time_since_reference_point_rem, 1, MPI_REAL, 0, comm2d, & ierr ) IF ( coupling_mode == 'vnested_crse' ) THEN ! Receive data from fine grid for anterpolation offset(1) = ( pdims_partner(1) / pdims(1) ) * pcoord(1) offset(2) = ( pdims_partner(2) / pdims(2) ) * pcoord(2) do j = 0, ( pdims_partner(2) / pdims(2) ) - 1 do i = 0, ( pdims_partner(1) / pdims(1) ) - 1 map_coord(1) = i+offset(1) map_coord(2) = j+offset(2) target_idex = f_rnk_lst(map_coord(1),map_coord(2)) + numprocs CALL MPI_RECV( bdims_rem, 6, MPI_INTEGER, target_idex, 10, & comm_inter,status, ierr ) bdims (1,1) = bdims_rem (1,1) / cfratio(1) bdims (1,2) = bdims_rem (1,2) / cfratio(1) bdims (2,1) = bdims_rem (2,1) / cfratio(2) bdims (2,2) = bdims_rem (2,2) / cfratio(2) bdims (3,1) = bdims_rem (3,1) bdims (3,2) = bdims_rem (3,2) / cfratio(3) CALL MPI_SEND( bdims, 6, MPI_INTEGER, target_idex, 9, & comm_inter, ierr ) n_cell_c = & (bdims(1,2)-bdims(1,1)+1) * & (bdims(2,2)-bdims(2,1)+1) * & (bdims(3,2)-bdims(3,1)+0) CALL MPI_RECV( u( & bdims(3,1)+1:bdims(3,2), & bdims(2,1) :bdims(2,2), & bdims(1,1) :bdims(1,2)),& n_cell_c, MPI_REAL, target_idex, 101, & comm_inter,status, ierr ) CALL MPI_RECV( v( & bdims(3,1)+1:bdims(3,2), & bdims(2,1) :bdims(2,2), & bdims(1,1) :bdims(1,2)),& n_cell_c, MPI_REAL, target_idex, 102, & comm_inter,status, ierr ) CALL MPI_RECV(pt( & bdims(3,1)+1:bdims(3,2), & bdims(2,1) :bdims(2,2), & bdims(1,1) :bdims(1,2)),& n_cell_c, MPI_REAL, target_idex, 105, & comm_inter,status, ierr ) IF ( humidity ) THEN CALL MPI_RECV(q( & bdims(3,1)+1:bdims(3,2), & bdims(2,1) :bdims(2,2), & bdims(1,1) :bdims(1,2)),& n_cell_c, MPI_REAL, target_idex, 106, & comm_inter,status, ierr ) ENDIF CALL MPI_RECV( w( & bdims(3,1) :bdims(3,2)-1, & bdims(2,1) :bdims(2,2), & bdims(1,1) :bdims(1,2)), & n_cell_c, MPI_REAL, target_idex, 103, & comm_inter,status, ierr ) end do end do ! !-- Boundary conditions for the velocity components u and v IF ( ibc_uv_b == 0 ) THEN u(nzb,:,:) = 0.0_wp v(nzb,:,:) = 0.0_wp ELSE u(nzb,:,:) = u(nzb+1,:,:) v(nzb,:,:) = v(nzb+1,:,:) END IF ! !-- Boundary conditions for the velocity components w w(nzb,:,:) = 0.0_wp ! !-- Temperature at bottom boundary. !-- Neumann, zero-gradient IF ( ibc_pt_b == 1 ) THEN DO l = 0, 1 DO m = 1, bc_h(l)%ns i = bc_h(l)%i(m) j = bc_h(l)%j(m) k = bc_h(l)%k(m) pt(k+bc_h(l)%koff,j,i) = pt(k,j,i) ENDDO ENDDO ENDIF CALL exchange_horiz( u, nbgp ) CALL exchange_horiz( v, nbgp ) CALL exchange_horiz( w, nbgp ) CALL exchange_horiz( pt, nbgp ) ELSEIF ( coupling_mode == 'vnested_fine' ) THEN ! Send data to coarse grid for anterpolation offset(1) = pcoord(1) / ( pdims(1)/pdims_partner(1) ) offset(2) = pcoord(2) / ( pdims(2)/pdims_partner(2) ) map_coord(1) = offset(1) map_coord(2) = offset(2) target_idex = c_rnk_lst(map_coord(1),map_coord(2)) !-- Limit anterpolation level to nzt - z nesting ratio (a pseudo-buffer layer) bdims (1,1) = nxl bdims (1,2) = nxr bdims (2,1) = nys bdims (2,2) = nyn bdims (3,1) = nzb bdims (3,2) = nzt-cfratio(3) CALL MPI_SEND( bdims, 6, MPI_INTEGER, target_idex, 10, & comm_inter, ierr ) CALL MPI_RECV( bdims_rem, 6, MPI_INTEGER, target_idex, 9, & comm_inter,status, ierr ) ALLOCATE( work3d ( & bdims_rem(3,1)+1:bdims_rem(3,2), & bdims_rem(2,1) :bdims_rem(2,2), & bdims_rem(1,1) :bdims_rem(1,2))) anterpol3d => u CALL anterpolate_to_crse_u CALL MPI_SEND( work3d, 1, TYPE_VNEST_ANTER, target_idex, & 101, comm_inter, ierr) anterpol3d => v CALL anterpolate_to_crse_v CALL MPI_SEND( work3d, 1, TYPE_VNEST_ANTER, target_idex, & 102, comm_inter, ierr) anterpol3d => pt CALL anterpolate_to_crse_s CALL MPI_SEND( work3d, 1, TYPE_VNEST_ANTER, target_idex, & 105, comm_inter, ierr) IF ( humidity ) THEN anterpol3d => q CALL anterpolate_to_crse_s CALL MPI_SEND( work3d, 1, TYPE_VNEST_ANTER, target_idex, & 106, comm_inter, ierr) ENDIF DEALLOCATE( work3d ) ALLOCATE( work3d ( bdims_rem(3,1) :bdims_rem(3,2)-1, & bdims_rem(2,1) :bdims_rem(2,2), & bdims_rem(1,1) :bdims_rem(1,2))) anterpol3d => w CALL anterpolate_to_crse_w CALL MPI_SEND( work3d, 1, TYPE_VNEST_ANTER, target_idex, & 103, comm_inter, ierr) NULLIFY ( anterpol3d ) DEALLOCATE( work3d ) ENDIF CONTAINS SUBROUTINE anterpolate_to_crse_u USE arrays_3d USE control_parameters USE grid_variables USE indices USE pegrid IMPLICIT NONE INTEGER(iwp) :: i INTEGER(iwp) :: j INTEGER(iwp) :: k INTEGER(iwp) :: iif INTEGER(iwp) :: jjf INTEGER(iwp) :: kkf INTEGER(iwp) :: bottomy INTEGER(iwp) :: bottomz INTEGER(iwp) :: topy INTEGER(iwp) :: topz REAL(wp) :: aweight ! !-- Anterpolation of the velocity components u !-- only values in yz-planes that coincide in the fine and !-- the coarse grid are considered DO k = bdims_rem(3,1)+1, bdims_rem(3,2) bottomz = (dzc/dzf) * (k-1) + 1 topz = (dzc/dzf) * k DO j = bdims_rem(2,1),bdims_rem(2,2) bottomy = (nyf+1) / (nyc+1) * j topy = (nyf+1) / (nyc+1) * (j+1) - 1 DO i = bdims_rem(1,1),bdims_rem(1,2) iif = (nxf+1) / (nxc+1) * i aweight = 0.0 DO kkf = bottomz, topz DO jjf = bottomy, topy aweight = aweight + anterpol3d(kkf,jjf,iif) * & (dzf/dzc) * (dyf/dyc) END DO END DO work3d(k,j,i) = aweight END DO END DO END DO END SUBROUTINE anterpolate_to_crse_u SUBROUTINE anterpolate_to_crse_v USE arrays_3d USE control_parameters USE grid_variables USE indices USE pegrid IMPLICIT NONE INTEGER(iwp) :: i INTEGER(iwp) :: j INTEGER(iwp) :: k INTEGER(iwp) :: iif INTEGER(iwp) :: jjf INTEGER(iwp) :: kkf INTEGER(iwp) :: bottomx INTEGER(iwp) :: bottomz INTEGER(iwp) :: topx INTEGER(iwp) :: topz REAL(wp) :: aweight ! !-- Anterpolation of the velocity components v !-- only values in xz-planes that coincide in the fine and !-- the coarse grid are considered DO k = bdims_rem(3,1)+1, bdims_rem(3,2) bottomz = (dzc/dzf) * (k-1) + 1 topz = (dzc/dzf) * k DO j = bdims_rem(2,1), bdims_rem(2,2) jjf = (nyf+1) / (nyc+1) * j DO i = bdims_rem(1,1), bdims_rem(1,2) bottomx = (nxf+1) / (nxc+1) * i topx = (nxf+1) / (nxc+1) * (i+1) - 1 aweight = 0.0 DO kkf = bottomz, topz DO iif = bottomx, topx aweight = aweight + anterpol3d(kkf,jjf,iif) * & (dzf/dzc) * (dxf/dxc) END DO END DO work3d(k,j,i) = aweight END DO END DO END DO END SUBROUTINE anterpolate_to_crse_v SUBROUTINE anterpolate_to_crse_w USE arrays_3d USE control_parameters USE grid_variables USE indices USE pegrid IMPLICIT NONE INTEGER(iwp) :: i INTEGER(iwp) :: j INTEGER(iwp) :: k INTEGER(iwp) :: iif INTEGER(iwp) :: jjf INTEGER(iwp) :: kkf INTEGER(iwp) :: bottomx INTEGER(iwp) :: bottomy INTEGER(iwp) :: topx INTEGER(iwp) :: topy REAL(wp) :: aweight ! !-- Anterpolation of the velocity components w !-- only values in xy-planes that coincide in the fine and !-- the coarse grid are considered DO k = bdims_rem(3,1), bdims_rem(3,2)-1 kkf = cfratio(3) * k DO j = bdims_rem(2,1), bdims_rem(2,2) bottomy = (nyf+1) / (nyc+1) * j topy = (nyf+1) / (nyc+1) * (j+1) - 1 DO i = bdims_rem(1,1), bdims_rem(1,2) bottomx = (nxf+1) / (nxc+1) * i topx = (nxf+1) / (nxc+1) * (i+1) - 1 aweight = 0.0 DO jjf = bottomy, topy DO iif = bottomx, topx aweight = aweight + anterpol3d (kkf,jjf,iif) * & (dxf/dxc) * (dyf/dyc) END DO END DO work3d(k,j,i) = aweight END DO END DO END DO END SUBROUTINE anterpolate_to_crse_w SUBROUTINE anterpolate_to_crse_s USE arrays_3d USE control_parameters USE grid_variables USE indices USE pegrid IMPLICIT NONE INTEGER(iwp) :: i INTEGER(iwp) :: j INTEGER(iwp) :: k INTEGER(iwp) :: iif INTEGER(iwp) :: jjf INTEGER(iwp) :: kkf INTEGER(iwp) :: bottomx INTEGER(iwp) :: bottomy INTEGER(iwp) :: bottomz INTEGER(iwp) :: topx INTEGER(iwp) :: topy INTEGER(iwp) :: topz REAL(wp) :: aweight ! !-- Anterpolation of the potential temperature pt !-- all fine grid values are considered DO k = bdims_rem(3,1)+1, bdims_rem(3,2) bottomz = (dzc/dzf) * (k-1) + 1 topz = (dzc/dzf) * k DO j = bdims_rem(2,1), bdims_rem(2,2) bottomy = (nyf+1) / (nyc+1) * j topy = (nyf+1) / (nyc+1) * (j+1) - 1 DO i = bdims_rem(1,1), bdims_rem(1,2) bottomx = (nxf+1) / (nxc+1) * i topx = (nxf+1) / (nxc+1) * (i+1) - 1 aweight = 0.0 DO kkf = bottomz, topz DO jjf = bottomy, topy DO iif = bottomx, topx aweight = aweight + anterpol3d(kkf,jjf,iif) * & (dzf/dzc) * (dyf/dyc) * (dxf/dxc) END DO END DO END DO work3d(k,j,i) = aweight END DO END DO END DO END SUBROUTINE anterpolate_to_crse_s #endif END SUBROUTINE vnest_anterpolate SUBROUTINE vnest_anterpolate_e #if defined( __parallel ) !--------------------------------------------------------------------------------! ! Description: ! ------------ ! Anterpolate TKE from fine grid to coarse grid. !------------------------------------------------------------------------------! USE arrays_3d USE control_parameters USE grid_variables USE indices USE interfaces USE pegrid IMPLICIT NONE REAL(wp) :: time_since_reference_point_rem INTEGER(iwp) :: i INTEGER(iwp) :: j ! !-- In case of model termination initiated by the remote model !-- (terminate_coupled_remote > 0), initiate termination of the local model. !-- The rest of the coupler must then be skipped because it would cause an MPI !-- intercomminucation hang. !-- If necessary, the coupler will be called at the beginning of the next !-- restart run. IF ( myid == 0) THEN CALL MPI_SENDRECV( terminate_coupled, 1, MPI_INTEGER, & target_id, 0, & terminate_coupled_remote, 1, MPI_INTEGER, & target_id, 0, & comm_inter, status, ierr ) ENDIF CALL MPI_BCAST( terminate_coupled_remote, 1, MPI_INTEGER, 0, comm2d, & ierr ) IF ( terminate_coupled_remote > 0 ) THEN WRITE( message_string, * ) 'remote model "', & TRIM( coupling_mode_remote ), & '" terminated', & '&with terminate_coupled_remote = ', & terminate_coupled_remote, & '&local model "', TRIM( coupling_mode ), & '" has', & '&terminate_coupled = ', & terminate_coupled CALL message( 'vnest_anterpolate_e', 'PA0310', 1, 2, 0, 6, 0 ) RETURN ENDIF ! !-- Exchange the current simulated time between the models IF ( myid == 0 ) THEN CALL MPI_SEND( time_since_reference_point, 1, MPI_REAL, target_id, & 11, comm_inter, ierr ) CALL MPI_RECV( time_since_reference_point_rem, 1, MPI_REAL, & target_id, 11, comm_inter, status, ierr ) ENDIF CALL MPI_BCAST( time_since_reference_point_rem, 1, MPI_REAL, 0, comm2d, & ierr ) IF ( coupling_mode == 'vnested_crse' ) THEN ! Receive data from fine grid for anterpolation offset(1) = ( pdims_partner(1) / pdims(1) ) * pcoord(1) offset(2) = ( pdims_partner(2) / pdims(2) ) * pcoord(2) do j = 0, ( pdims_partner(2) / pdims(2) ) - 1 do i = 0, ( pdims_partner(1) / pdims(1) ) - 1 map_coord(1) = i+offset(1) map_coord(2) = j+offset(2) target_idex = f_rnk_lst(map_coord(1),map_coord(2)) + numprocs bdims (1,1) = f2c_dims_cg (0,map_coord(1),map_coord(2)) bdims (1,2) = f2c_dims_cg (1,map_coord(1),map_coord(2)) bdims (2,1) = f2c_dims_cg (2,map_coord(1),map_coord(2)) bdims (2,2) = f2c_dims_cg (3,map_coord(1),map_coord(2)) bdims (3,1) = f2c_dims_cg (4,map_coord(1),map_coord(2)) bdims (3,2) = f2c_dims_cg (5,map_coord(1),map_coord(2)) n_cell_c = (bdims(1,2)-bdims(1,1)+1) * & (bdims(2,2)-bdims(2,1)+1) * & (bdims(3,2)-bdims(3,1)+0) CALL MPI_RECV( e( bdims(3,1)+1:bdims(3,2), & bdims(2,1) :bdims(2,2), & bdims(1,1) :bdims(1,2)),& n_cell_c, MPI_REAL, target_idex, 104, & comm_inter,status, ierr ) end do end do ! !-- Boundary conditions IF ( .NOT. constant_diffusion ) THEN e(nzb,:,:) = e(nzb+1,:,:) END IF IF ( .NOT. constant_diffusion ) CALL exchange_horiz( e, nbgp ) ELSEIF ( coupling_mode == 'vnested_fine' ) THEN ! Send data to coarse grid for anterpolation offset(1) = pcoord(1) / ( pdims(1)/pdims_partner(1) ) offset(2) = pcoord(2) / ( pdims(2)/pdims_partner(2) ) map_coord(1) = offset(1) map_coord(2) = offset(2) target_idex = c_rnk_lst(map_coord(1),map_coord(2)) bdims_rem (1,1) = f2c_dims_fg (0) bdims_rem (1,2) = f2c_dims_fg (1) bdims_rem (2,1) = f2c_dims_fg (2) bdims_rem (2,2) = f2c_dims_fg (3) bdims_rem (3,1) = f2c_dims_fg (4) bdims_rem (3,2) = f2c_dims_fg (5) ALLOCATE( work3d ( & bdims_rem(3,1)+1:bdims_rem(3,2), & bdims_rem(2,1) :bdims_rem(2,2), & bdims_rem(1,1) :bdims_rem(1,2))) anterpol3d => e CALL anterpolate_to_crse_e CALL MPI_SEND( work3d, 1, TYPE_VNEST_ANTER, target_idex, & 104, comm_inter, ierr) NULLIFY ( anterpol3d ) DEALLOCATE( work3d ) ENDIF CONTAINS SUBROUTINE anterpolate_to_crse_e USE arrays_3d USE control_parameters USE grid_variables USE indices USE pegrid IMPLICIT NONE INTEGER(iwp) :: i INTEGER(iwp) :: j INTEGER(iwp) :: k INTEGER(iwp) :: iif INTEGER(iwp) :: jjf INTEGER(iwp) :: kkf INTEGER(iwp) :: bottomx INTEGER(iwp) :: bottomy INTEGER(iwp) :: bottomz INTEGER(iwp) :: topx INTEGER(iwp) :: topy INTEGER(iwp) :: topz REAL(wp) :: aweight_a REAL(wp) :: aweight_b REAL(wp) :: aweight_c REAL(wp) :: aweight_d REAL(wp) :: aweight_e REAL(wp) :: energ DO k = bdims_rem(3,1)+1, bdims_rem(3,2) bottomz = (dzc/dzf) * (k-1) + 1 topz = (dzc/dzf) * k DO j = bdims_rem(2,1), bdims_rem(2,2) bottomy = (nyf+1) / (nyc+1) * j topy = (nyf+1) / (nyc+1) * (j+1) - 1 DO i = bdims_rem(1,1), bdims_rem(1,2) bottomx = (nxf+1) / (nxc+1) * i topx = (nxf+1) / (nxc+1) * (i+1) - 1 aweight_a = 0.0 aweight_b = 0.0 aweight_c = 0.0 aweight_d = 0.0 aweight_e = 0.0 DO kkf = bottomz, topz DO jjf = bottomy, topy DO iif = bottomx, topx aweight_a = aweight_a + anterpol3d(kkf,jjf,iif) * & (dzf/dzc) * (dyf/dyc) * (dxf/dxc) energ = ( 0.5 * ( u(kkf,jjf,iif) + u(kkf,jjf,iif+1) ) )**2.0 + & ( 0.5 * ( v(kkf,jjf,iif) + v(kkf,jjf+1,iif) ) )**2.0 + & ( 0.5 * ( w(kkf-1,jjf,iif) + w(kkf,jjf,iif) ) )**2.0 aweight_b = aweight_b + energ * & (dzf/dzc) * (dyf/dyc) * (dxf/dxc) aweight_c = aweight_c + 0.5 * ( u(kkf,jjf,iif) + u(kkf,jjf,iif+1) ) * & (dzf/dzc) * (dyf/dyc) * (dxf/dxc) aweight_d = aweight_d + 0.5 * ( v(kkf,jjf,iif) + v(kkf,jjf+1,iif) ) * & (dzf/dzc) * (dyf/dyc) * (dxf/dxc) aweight_e = aweight_e + 0.5 * ( w(kkf-1,jjf,iif) + w(kkf,jjf,iif) ) * & (dzf/dzc) * (dyf/dyc) * (dxf/dxc) END DO END DO END DO work3d(k,j,i) = aweight_a + 0.5 * ( aweight_b - & aweight_c**2.0 - & aweight_d**2.0 - & aweight_e**2.0 ) END DO END DO END DO END SUBROUTINE anterpolate_to_crse_e #endif END SUBROUTINE vnest_anterpolate_e SUBROUTINE vnest_init_pegrid_rank #if defined( __parallel ) ! Domain decomposition and exchange of grid variables between coarse and fine ! Given processor coordinates as index f_rnk_lst(pcoord(1), pcoord(2)) ! returns the rank. A single coarse block will have to send data to multiple ! fine blocks. In the coarse grid the pcoords of the remote block is first found and then using ! f_rnk_lst the target_idex is identified. ! blk_dim stores the index limits of a given block. blk_dim_remote is received ! from the asscoiated nest partner. ! cf_ratio(1:3) is the ratio between fine and coarse grid: nxc/nxf, nyc/nyf and ! ceiling(dxc/dxf) USE control_parameters, & ONLY: coupling_mode USE kinds USE pegrid IMPLICIT NONE INTEGER(iwp) :: dest_rnk INTEGER(iwp) :: i !< IF (myid == 0) THEN IF ( coupling_mode == 'vnested_crse') THEN CALL MPI_SEND( pdims, 2, MPI_INTEGER, numprocs, 33, comm_inter, & ierr ) CALL MPI_RECV( pdims_partner, 2, MPI_INTEGER, numprocs, 66, & comm_inter, status, ierr ) ELSEIF ( coupling_mode == 'vnested_fine') THEN CALL MPI_RECV( pdims_partner, 2, MPI_INTEGER, 0, 33, & comm_inter, status, ierr ) CALL MPI_SEND( pdims, 2, MPI_INTEGER, 0, 66, comm_inter, & ierr ) ENDIF ENDIF IF ( coupling_mode == 'vnested_crse') THEN CALL MPI_BCAST( pdims_partner, 2, MPI_INTEGER, 0, comm2d, ierr ) ALLOCATE( c_rnk_lst( 0:(pdims(1)-1) ,0:(pdims(2)-1) ) ) ALLOCATE( f_rnk_lst( 0:(pdims_partner(1)-1) ,0:(pdims_partner(2)-1) ) ) do i=0,numprocs-1 CALL MPI_CART_COORDS( comm2d, i, ndim, pcoord, ierr ) call MPI_Cart_rank(comm2d, pcoord, dest_rnk, ierr) c_rnk_lst(pcoord(1),pcoord(2)) = dest_rnk end do ELSEIF ( coupling_mode == 'vnested_fine') THEN CALL MPI_BCAST( pdims_partner, 2, MPI_INTEGER, 0, comm2d, ierr ) ALLOCATE( c_rnk_lst( 0:(pdims_partner(1)-1) ,0:(pdims_partner(2)-1) ) ) ALLOCATE( f_rnk_lst( 0:(pdims(1)-1) ,0:(pdims(2)-1) ) ) do i=0,numprocs-1 CALL MPI_CART_COORDS( comm2d, i, ndim, pcoord, ierr ) call MPI_Cart_rank(comm2d, pcoord, dest_rnk, ierr) f_rnk_lst(pcoord(1),pcoord(2)) = dest_rnk enddo ENDIF IF ( coupling_mode == 'vnested_crse') THEN if (myid == 0) then CALL MPI_SEND( c_rnk_lst, pdims(1)*pdims(2), MPI_INTEGER, numprocs, 0, comm_inter, ierr ) CALL MPI_RECV( f_rnk_lst, pdims_partner(1)*pdims_partner(2), MPI_INTEGER, numprocs, 4, comm_inter,status, ierr ) end if CALL MPI_BCAST( f_rnk_lst, pdims_partner(1)*pdims_partner(2), MPI_INTEGER, 0, comm2d, ierr ) ELSEIF ( coupling_mode == 'vnested_fine') THEN if (myid == 0) then CALL MPI_RECV( c_rnk_lst, pdims_partner(1)*pdims_partner(2), MPI_INTEGER, 0, 0, comm_inter,status, ierr ) CALL MPI_SEND( f_rnk_lst, pdims(1)*pdims(2), MPI_INTEGER, 0, 4, comm_inter, ierr ) end if CALL MPI_BCAST( c_rnk_lst, pdims_partner(1)*pdims_partner(2), MPI_INTEGER, 0, comm2d, ierr ) ENDIF !-- Reason for MPI error unknown; solved if three lines duplicated CALL MPI_CART_COORDS( comm2d, myid, ndim, pcoord, ierr ) CALL MPI_CART_SHIFT( comm2d, 0, 1, pleft, pright, ierr ) CALL MPI_CART_SHIFT( comm2d, 1, 1, psouth, pnorth, ierr ) #endif END SUBROUTINE vnest_init_pegrid_rank SUBROUTINE vnest_init_pegrid_domain #if defined( __parallel ) USE control_parameters, & ONLY: coupling_mode, coupling_topology, dz, & dz_stretch_level_start, message_string USE grid_variables, & ONLY: dx, dy USE indices, & ONLY: nbgp, nx, ny, nz, nxl, nxr, nys, nyn, nzb, nzt USE kinds USE pegrid IMPLICIT NONE INTEGER(iwp) :: i !< INTEGER(iwp) :: j !< INTEGER(iwp) :: tempx INTEGER(iwp) :: tempy INTEGER(iwp) :: TYPE_INT_YZ INTEGER(iwp) :: SIZEOFREAL INTEGER(iwp) :: MTV_X INTEGER(iwp) :: MTV_Y INTEGER(iwp) :: MTV_Z INTEGER(iwp) :: MTV_RX INTEGER(iwp) :: MTV_RY INTEGER(iwp) :: MTV_RZ ! !-- Pass the number of grid points of the coarse model to !-- the nested model and vice versa IF ( coupling_mode == 'vnested_crse' ) THEN nxc = nx nyc = ny nzc = nz dxc = dx dyc = dy dzc = dz(1) cg_nprocs = numprocs IF ( myid == 0 ) THEN CALL MPI_SEND( nxc, 1, MPI_INTEGER , numprocs, 1, comm_inter, & ierr ) CALL MPI_SEND( nyc, 1, MPI_INTEGER , numprocs, 2, comm_inter, & ierr ) CALL MPI_SEND( nzc, 1, MPI_INTEGER , numprocs, 3, comm_inter, & ierr ) CALL MPI_SEND( dxc, 1, MPI_REAL , numprocs, 4, comm_inter, & ierr ) CALL MPI_SEND( dyc, 1, MPI_REAL , numprocs, 5, comm_inter, & ierr ) CALL MPI_SEND( dzc, 1, MPI_REAL , numprocs, 6, comm_inter, & ierr ) CALL MPI_SEND( pdims, 2, MPI_INTEGER, numprocs, 7, comm_inter, & ierr ) CALL MPI_SEND( cg_nprocs, 1, MPI_INTEGER, numprocs, 8, comm_inter, & ierr ) CALL MPI_RECV( nxf, 1, MPI_INTEGER, numprocs, 21, comm_inter, & status, ierr ) CALL MPI_RECV( nyf, 1, MPI_INTEGER, numprocs, 22, comm_inter, & status, ierr ) CALL MPI_RECV( nzf, 1, MPI_INTEGER, numprocs, 23, comm_inter, & status, ierr ) CALL MPI_RECV( dxf, 1, MPI_REAL, numprocs, 24, comm_inter, & status, ierr ) CALL MPI_RECV( dyf, 1, MPI_REAL, numprocs, 25, comm_inter, & status, ierr ) CALL MPI_RECV( dzf, 1, MPI_REAL, numprocs, 26, comm_inter, & status, ierr ) CALL MPI_RECV( pdims_partner, 2, MPI_INTEGER, & numprocs, 27, comm_inter, status, ierr ) CALL MPI_RECV( fg_nprocs, 1, MPI_INTEGER, & numprocs, 28, comm_inter, status, ierr ) ENDIF CALL MPI_BCAST( nxf, 1, MPI_INTEGER, 0, comm2d, ierr ) CALL MPI_BCAST( nyf, 1, MPI_INTEGER, 0, comm2d, ierr ) CALL MPI_BCAST( nzf, 1, MPI_INTEGER, 0, comm2d, ierr ) CALL MPI_BCAST( dxf, 1, MPI_REAL, 0, comm2d, ierr ) CALL MPI_BCAST( dyf, 1, MPI_REAL, 0, comm2d, ierr ) CALL MPI_BCAST( dzf, 1, MPI_REAL, 0, comm2d, ierr ) CALL MPI_BCAST( pdims_partner, 2, MPI_INTEGER, 0, comm2d, ierr ) CALL MPI_BCAST( fg_nprocs, 1, MPI_INTEGER, 0, comm2d, ierr ) ! !-- Check if stretching is used within the nested domain. ABS(...) is !-- necessary because of the default value of -9999999.9_wp (negative) IF ( ABS( dz_stretch_level_start(1) ) <= (nzf+1)*dzf ) THEN message_string = 'Stretching in the parent domain is '// & 'only allowed above the nested domain' CALL message( 'vnest_init_pegrid_domain', 'PA0497', 1, 2, 0, 6, 0 ) ENDIF ELSEIF ( coupling_mode == 'vnested_fine' ) THEN nxf = nx nyf = ny nzf = nz dxf = dx dyf = dy dzf = dz(1) fg_nprocs = numprocs IF ( myid == 0 ) THEN CALL MPI_RECV( nxc, 1, MPI_INTEGER, 0, 1, comm_inter, status, & ierr ) CALL MPI_RECV( nyc, 1, MPI_INTEGER, 0, 2, comm_inter, status, & ierr ) CALL MPI_RECV( nzc, 1, MPI_INTEGER, 0, 3, comm_inter, status, & ierr ) CALL MPI_RECV( dxc, 1, MPI_REAL, 0, 4, comm_inter, status, & ierr ) CALL MPI_RECV( dyc, 1, MPI_REAL, 0, 5, comm_inter, status, & ierr ) CALL MPI_RECV( dzc, 1, MPI_REAL, 0, 6, comm_inter, status, & ierr ) CALL MPI_RECV( pdims_partner, 2, MPI_INTEGER, 0, 7, comm_inter, & status, ierr ) CALL MPI_RECV( cg_nprocs, 1, MPI_INTEGER, 0, 8, comm_inter, & status, ierr ) CALL MPI_SEND( nxf, 1, MPI_INTEGER, 0, 21, comm_inter, ierr ) CALL MPI_SEND( nyf, 1, MPI_INTEGER, 0, 22, comm_inter, ierr ) CALL MPI_SEND( nzf, 1, MPI_INTEGER, 0, 23, comm_inter, ierr ) CALL MPI_SEND( dxf, 1, MPI_REAL, 0, 24, comm_inter, ierr ) CALL MPI_SEND( dyf, 1, MPI_REAL, 0, 25, comm_inter, ierr ) CALL MPI_SEND( dzf, 1, MPI_REAL, 0, 26, comm_inter, ierr ) CALL MPI_SEND( pdims,2,MPI_INTEGER, 0, 27, comm_inter, ierr ) CALL MPI_SEND( fg_nprocs,1,MPI_INTEGER, 0, 28, comm_inter, ierr ) ENDIF CALL MPI_BCAST( nxc, 1, MPI_INTEGER, 0, comm2d, ierr) CALL MPI_BCAST( nyc, 1, MPI_INTEGER, 0, comm2d, ierr) CALL MPI_BCAST( nzc, 1, MPI_INTEGER, 0, comm2d, ierr) CALL MPI_BCAST( dxc, 1, MPI_REAL, 0, comm2d, ierr) CALL MPI_BCAST( dyc, 1, MPI_REAL, 0, comm2d, ierr) CALL MPI_BCAST( dzc, 1, MPI_REAL, 0, comm2d, ierr) CALL MPI_BCAST( pdims_partner, 2, MPI_INTEGER, 0, comm2d, ierr) CALL MPI_BCAST( cg_nprocs, 1, MPI_INTEGER, 0, comm2d, ierr) ENDIF ngp_c = ( nxc+1 + 2 * nbgp ) * ( nyc+1 + 2 * nbgp ) ngp_f = ( nxf+1 + 2 * nbgp ) * ( nyf+1 + 2 * nbgp ) IF ( coupling_mode(1:8) == 'vnested_') coupling_topology = 1 !-- Nesting Ratio: For each coarse grid cell how many fine grid cells exist cfratio(1) = INT ( (nxf+1) / (nxc+1) ) cfratio(2) = INT ( (nyf+1) / (nyc+1) ) cfratio(3) = CEILING ( dzc / dzf ) !-- target_id is used only for exhange of information like simulated_time !-- which are then MPI_BCAST to other processors in the group IF ( myid == 0 ) THEN IF ( TRIM( coupling_mode ) == 'vnested_crse' ) THEN target_id = numprocs ELSE IF ( TRIM( coupling_mode ) == 'vnested_fine' ) THEN target_id = 0 ENDIF ENDIF !-- Store partner grid dimenstions and create MPI derived types IF ( coupling_mode == 'vnested_crse' ) THEN offset(1) = ( pdims_partner(1) / pdims(1) ) * pcoord(1) offset(2) = ( pdims_partner(2) / pdims(2) ) * pcoord(2) tempx = ( pdims_partner(1) / pdims(1) ) - 1 tempy = ( pdims_partner(2) / pdims(2) ) - 1 ALLOCATE( c2f_dims_cg (0:5,offset(1):tempx+offset(1),offset(2):tempy+offset(2) ) ) ALLOCATE( f2c_dims_cg (0:5,offset(1):tempx+offset(1),offset(2):tempy+offset(2) ) ) do j = 0, ( pdims_partner(2) / pdims(2) ) - 1 do i = 0, ( pdims_partner(1) / pdims(1) ) - 1 map_coord(1) = i+offset(1) map_coord(2) = j+offset(2) target_idex = f_rnk_lst(map_coord(1),map_coord(2)) + numprocs CALL MPI_RECV( bdims_rem, 6, MPI_INTEGER, target_idex, 10, & comm_inter,status, ierr ) !-- Store the CG dimensions that correspond to the FG partner; needed for FG top BC !-- One CG can have multiple FG partners. The 3D array is mapped by partner proc co-ord c2f_dims_cg (0,map_coord(1),map_coord(2)) = bdims_rem (1,1) / cfratio(1) c2f_dims_cg (1,map_coord(1),map_coord(2)) = bdims_rem (1,2) / cfratio(1) c2f_dims_cg (2,map_coord(1),map_coord(2)) = bdims_rem (2,1) / cfratio(2) c2f_dims_cg (3,map_coord(1),map_coord(2)) = bdims_rem (2,2) / cfratio(2) c2f_dims_cg (4,map_coord(1),map_coord(2)) = bdims_rem (3,2) / cfratio(3) c2f_dims_cg (5,map_coord(1),map_coord(2)) =(bdims_rem (3,2) / cfratio(3)) + 2 !-- Store the CG dimensions that correspond to the FG partner; needed for anterpolation f2c_dims_cg (0,map_coord(1),map_coord(2)) = bdims_rem (1,1) / cfratio(1) f2c_dims_cg (1,map_coord(1),map_coord(2)) = bdims_rem (1,2) / cfratio(1) f2c_dims_cg (2,map_coord(1),map_coord(2)) = bdims_rem (2,1) / cfratio(2) f2c_dims_cg (3,map_coord(1),map_coord(2)) = bdims_rem (2,2) / cfratio(2) f2c_dims_cg (4,map_coord(1),map_coord(2)) = bdims_rem (3,1) f2c_dims_cg (5,map_coord(1),map_coord(2)) =(bdims_rem (3,2)-cfratio(3))/ cfratio(3) CALL MPI_SEND( c2f_dims_cg (:,map_coord(1),map_coord(2)), 6, & MPI_INTEGER, target_idex, 100, comm_inter, ierr ) CALL MPI_SEND( f2c_dims_cg (:,map_coord(1),map_coord(2)), 6, & MPI_INTEGER, target_idex, 101, comm_inter, ierr ) end do end do !-- A derived data type to pack 3 Z-levels of CG to set FG top BC MTV_X = ( nxr - nxl + 1 ) + 2*nbgp MTV_Y = ( nyn - nys + 1 ) + 2*nbgp MTV_Z = nzt+1 - nzb +1 MTV_RX = ( c2f_dims_cg (1,offset(1),offset(2)) - c2f_dims_cg (0,offset(1),offset(2)) ) +1+2 MTV_RY = ( c2f_dims_cg (3,offset(1),offset(2)) - c2f_dims_cg (2,offset(1),offset(2)) ) +1+2 MTV_RZ = ( c2f_dims_cg (5,offset(1),offset(2)) - c2f_dims_cg (4,offset(1),offset(2)) ) +1 CALL MPI_TYPE_EXTENT(MPI_REAL, SIZEOFREAL, IERR) CALL MPI_TYPE_VECTOR ( MTV_RY, MTV_RZ, MTV_Z, MPI_REAL, TYPE_INT_YZ, IERR) CALL MPI_TYPE_HVECTOR( MTV_RX, 1, MTV_Z*MTV_Y*SIZEOFREAL, & TYPE_INT_YZ, TYPE_VNEST_BC, IERR) CALL MPI_TYPE_FREE(TYPE_INT_YZ, IERR) CALL MPI_TYPE_COMMIT(TYPE_VNEST_BC, IERR) ELSEIF ( coupling_mode == 'vnested_fine' ) THEN ALLOCATE( c2f_dims_fg (0:5) ) ALLOCATE( f2c_dims_fg (0:5) ) offset(1) = pcoord(1) / ( pdims(1)/pdims_partner(1) ) offset(2) = pcoord(2) / ( pdims(2)/pdims_partner(2) ) map_coord(1) = offset(1) map_coord(2) = offset(2) target_idex = c_rnk_lst(map_coord(1),map_coord(2)) bdims (1,1) = nxl bdims (1,2) = nxr bdims (2,1) = nys bdims (2,2) = nyn bdims (3,1) = nzb bdims (3,2) = nzt CALL MPI_SEND( bdims, 6, MPI_INTEGER, target_idex, 10, & comm_inter, ierr ) !-- Store the CG dimensions that correspond to the FG partner; needed for FG top BC !-- One FG can have only one CG partner CALL MPI_RECV( c2f_dims_fg, 6, MPI_INTEGER, target_idex, 100, & comm_inter,status, ierr ) CALL MPI_RECV( f2c_dims_fg, 6, MPI_INTEGER, target_idex, 101, & comm_inter,status, ierr ) !-- Store the CG dimensions that correspond to the FG partner; needed for anterpolation n_cell_c = (f2c_dims_fg(1)-f2c_dims_fg(0)+1) * & (f2c_dims_fg(3)-f2c_dims_fg(2)+1) * & (f2c_dims_fg(5)-f2c_dims_fg(4)+0) CALL MPI_TYPE_CONTIGUOUS(n_cell_c, MPI_REAL, TYPE_VNEST_ANTER, IERR) CALL MPI_TYPE_COMMIT(TYPE_VNEST_ANTER, ierr) ENDIF #endif END SUBROUTINE vnest_init_pegrid_domain SUBROUTINE vnest_init_grid #if defined( __parallel ) USE arrays_3d, & ONLY: zu, zw USE control_parameters, & ONLY: coupling_mode, message_string, number_stretch_level_start USE indices, & ONLY: nzt USE kinds USE pegrid IMPLICIT NONE ! !-- Allocate and Exchange zuc and zuf, zwc and zwf IF ( coupling_mode(1:8) == 'vnested_' ) THEN ALLOCATE( zuc(0:nzc+1), zuf(0:nzf+1) ) ALLOCATE( zwc(0:nzc+1), zwf(0:nzf+1) ) IF ( coupling_mode == 'vnested_crse' ) THEN zuc = zu zwc = zw IF ( myid == 0 ) THEN CALL MPI_SEND( zuc, nzt+2, MPI_REAL, numprocs, 41, comm_inter, & ierr ) CALL MPI_RECV( zuf, nzf+2, MPI_REAL, numprocs, 42, comm_inter, & status, ierr ) CALL MPI_SEND( zwc, nzt+2, MPI_REAL, numprocs, 43, comm_inter, & ierr ) CALL MPI_RECV( zwf, nzf+2, MPI_REAL, numprocs, 44, comm_inter, & status, ierr ) ENDIF CALL MPI_BCAST( zuf,nzf+2,MPI_REAL, 0, comm2d, ierr ) CALL MPI_BCAST( zwf,nzf+2,MPI_REAL, 0, comm2d, ierr ) ELSEIF ( coupling_mode == 'vnested_fine' ) THEN ! !-- Check if stretching is used within the nested domain IF ( number_stretch_level_start > 0 ) THEN message_string = 'Stretching in the nested domain is not '//& 'allowed' CALL message( 'vnest_init_grid', 'PA0498', 1, 2, 0, 6, 0 ) ENDIF zuf = zu zwf = zw IF ( myid == 0 ) THEN CALL MPI_RECV( zuc,nzc+2, MPI_REAL, 0, 41, comm_inter, status, & ierr ) CALL MPI_SEND( zuf,nzt+2, MPI_REAL, 0, 42, comm_inter, ierr ) CALL MPI_RECV( zwc,nzc+2, MPI_REAL, 0, 43, comm_inter, status, & ierr ) CALL MPI_SEND( zwf,nzt+2, MPI_REAL, 0, 44, comm_inter, ierr ) ENDIF CALL MPI_BCAST( zuc,nzc+2,MPI_REAL, 0, comm2d, ierr ) CALL MPI_BCAST( zwc,nzc+2,MPI_REAL, 0, comm2d, ierr ) ENDIF ENDIF #endif END SUBROUTINE vnest_init_grid SUBROUTINE vnest_check_parameters #if defined( __parallel ) USE pegrid, & ONLY: myid IMPLICIT NONE IF (myid==0) PRINT*, '*** vnest: check parameters not implemented yet ***' #endif END SUBROUTINE vnest_check_parameters SUBROUTINE vnest_timestep_sync #if defined( __parallel ) USE control_parameters, & ONLY: coupling_mode, dt_3d USE interfaces USE kinds USE pegrid IMPLICIT NONE IF ( coupling_mode == 'vnested_crse') THEN dtc = dt_3d if (myid == 0) then CALL MPI_SEND( dt_3d, 1, MPI_REAL, target_id, & 31, comm_inter, ierr ) CALL MPI_RECV( dtf, 1, MPI_REAL, & target_id, 32, comm_inter, status, ierr ) endif CALL MPI_BCAST( dtf, 1, MPI_REAL, 0, comm2d, ierr ) ELSE dtf = dt_3d if (myid == 0) then CALL MPI_RECV( dtc, 1, MPI_REAL, & target_id, 31, comm_inter, status, ierr ) CALL MPI_SEND( dt_3d, 1, MPI_REAL, target_id, & 32, comm_inter, ierr ) endif CALL MPI_BCAST( dtc, 1, MPI_REAL, 0, comm2d, ierr ) ENDIF !-- Identical timestep for coarse and fine grids dt_3d = MIN( dtc, dtf ) #endif END SUBROUTINE vnest_timestep_sync SUBROUTINE vnest_deallocate #if defined( __parallel ) USE control_parameters, & ONLY: coupling_mode IMPLICIT NONE IF ( ALLOCATED(c_rnk_lst) ) DEALLOCATE (c_rnk_lst) IF ( ALLOCATED(f_rnk_lst) ) DEALLOCATE (f_rnk_lst) IF ( coupling_mode == 'vnested_crse') THEN IF ( ALLOCATED (c2f_dims_cg) ) DEALLOCATE (c2f_dims_cg) IF ( ALLOCATED (f2c_dims_cg) ) DEALLOCATE (f2c_dims_cg) ELSEIF( coupling_mode == 'vnested_fine' ) THEN IF ( ALLOCATED (c2f_dims_fg) ) DEALLOCATE (c2f_dims_fg) IF ( ALLOCATED (f2c_dims_fg) ) DEALLOCATE (f2c_dims_fg) ENDIF #endif END SUBROUTINE vnest_deallocate END MODULE vertical_nesting_mod