MODULE module_polarfft USE module_model_constants USE module_wrf_error CHARACTER (LEN=256) , PRIVATE :: a_message CONTAINS SUBROUTINE couple_scalars_for_filter ( field & ,mu,mub,c1,c2 & ,ids,ide,jds,jde,kds,kde & ,ims,ime,jms,jme,kms,kme & ,ips,ipe,jps,jpe,kps,kpe ) IMPLICIT NONE INTEGER, INTENT(IN) :: ids,ide,jds,jde,kds,kde & ,ims,ime,jms,jme,kms,kme & ,ips,ipe,jps,jpe,kps,kpe REAL , DIMENSION(ims:ime,kms:kme,jms:jme) , INTENT(INOUT) :: field REAL , DIMENSION(ims:ime,jms:jme) , INTENT(IN) :: mu,mub REAL , DIMENSION(kms:kme) , INTENT(IN) :: c1,c2 INTEGER :: i , j , k DO j = jps, MIN(jpe,jde-1) DO k = kps, kpe-1 DO i = ips, MIN(ipe,ide-1) field(i,k,j)=field(i,k,j)*( (c1(k)*mu(i,j))+(c1(k)*mub(i,j)+c2(k))) END DO END DO END DO END SUBROUTINE couple_scalars_for_filter SUBROUTINE uncouple_scalars_for_filter ( field & ,mu,mub,c1,c2 & ,ids,ide,jds,jde,kds,kde & ,ims,ime,jms,jme,kms,kme & ,ips,ipe,jps,jpe,kps,kpe ) IMPLICIT NONE INTEGER, INTENT(IN) :: ids,ide,jds,jde,kds,kde & ,ims,ime,jms,jme,kms,kme & ,ips,ipe,jps,jpe,kps,kpe REAL , DIMENSION(ims:ime,kms:kme,jms:jme) , INTENT(INOUT) :: field REAL , DIMENSION(ims:ime,jms:jme) , INTENT(IN) :: mu,mub REAL , DIMENSION(kms:kme) , INTENT(IN) :: c1,c2 INTEGER :: i , j , k DO j = jps, MIN(jpe,jde-1) DO k = kps, kpe-1 DO i = ips, MIN(ipe,ide-1) field(i,k,j)=field(i,k,j)/( (c1(k)*mu(i,j))+(c1(k)*mub(i,j)+c2(k))) END DO END DO END DO END SUBROUTINE uncouple_scalars_for_filter SUBROUTINE pxft ( grid & ,lineno & ,flag_uv,flag_rurv & ,flag_wph,flag_ww & ,flag_t & ,flag_mu,flag_mut & ,flag_moist & ,flag_chem & ,flag_tracer & ,flag_scalar & ,fft_filter_lat, dclat & ,actual_distance_average & ,pos_def & ,swap_pole_with_next_j & ,moist,chem,tracer,scalar & ,ids,ide,jds,jde,kds,kde & ,ims,ime,jms,jme,kms,kme & ,ips,ipe,jps,jpe,kps,kpe & ,imsx,imex,jmsx,jmex,kmsx,kmex & ,ipsx,ipex,jpsx,jpex,kpsx,kpex ) USE module_state_description USE module_domain, ONLY : domain #ifdef DM_PARALLEL USE module_dm, ONLY : local_communicator, mytask, ntasks, ntasks_x, ntasks_y & , local_communicator_periodic, itrace & , local_communicator_x USE module_driver_constants #if 0 USE module_comm_dm, ONLY : & XPOSE_POLAR_FILTER_V_z2x_sub & ,XPOSE_POLAR_FILTER_V_x2z_sub & ,XPOSE_POLAR_FILTER_U_z2x_sub & ,XPOSE_POLAR_FILTER_U_x2z_sub & ,XPOSE_POLAR_FILTER_T_z2x_sub & ,XPOSE_POLAR_FILTER_T_x2z_sub & ,XPOSE_POLAR_FILTER_W_z2x_sub & ,XPOSE_POLAR_FILTER_W_x2z_sub & ,XPOSE_POLAR_FILTER_PH_z2x_sub & ,XPOSE_POLAR_FILTER_PH_x2z_sub & ,XPOSE_POLAR_FILTER_WW_z2x_sub & ,XPOSE_POLAR_FILTER_WW_x2z_sub & ,XPOSE_POLAR_FILTER_RV_z2x_sub & ,XPOSE_POLAR_FILTER_RV_x2z_sub & ,XPOSE_POLAR_FILTER_RU_z2x_sub & ,XPOSE_POLAR_FILTER_RU_x2z_sub & ,XPOSE_POLAR_FILTER_MOIST_z2x_sub & ,XPOSE_POLAR_FILTER_MOIST_x2z_sub & ,XPOSE_POLAR_FILTER_CHEM_z2x_sub & ,XPOSE_POLAR_FILTER_MOIST_x2z_sub & ,XPOSE_POLAR_FILTER_TRACER_z2x_sub & ,XPOSE_POLAR_FILTER_TRACER_x2z_sub & ,XPOSE_POLAR_FILTER_SCALAR_z2x_sub & ,XPOSE_POLAR_FILTER_SCALAR_x2z_sub #endif #else USE module_dm #endif IMPLICIT NONE ! Input data. TYPE(domain) , TARGET :: grid integer, intent(in) :: lineno integer myproc, i, j, k LOGICAL, INTENT(IN) :: actual_distance_average LOGICAL, INTENT(IN) :: pos_def LOGICAL, INTENT(IN) :: swap_pole_with_next_j INTEGER, INTENT(IN) :: ids,ide,jds,jde,kds,kde & ,ims,ime,jms,jme,kms,kme & ,ips,ipe,jps,jpe,kps,kpe & ,imsx,imex,jmsx,jmex,kmsx,kmex & ,ipsx,ipex,jpsx,jpex,kpsx,kpex REAL , INTENT(IN) :: fft_filter_lat REAL, INTENT(IN) :: dclat INTEGER, INTENT(IN) :: flag_uv & ,flag_rurv & ,flag_ww & ,flag_t,flag_wph & ,flag_mu,flag_mut & ,flag_moist & ,flag_chem & ,flag_tracer & ,flag_scalar REAL, DIMENSION(ims:ime,kms:kme,jms:jme,*) , INTENT(INOUT) :: moist, chem, scalar,tracer ! Local LOGICAL piggyback_mu, piggyback_mut INTEGER ij, k_end #ifdef DM_PARALLEL #else INTEGER itrace #endif piggyback_mu = flag_mu .EQ. 1 piggyback_mut = flag_mut .EQ. 1 ! ! ! The idea is that this parallel transpose fft routine can be ! called at various points in the solver (solve_em) and it will filter ! the correct fields based on the flag arguments. There are two 2d ! fields mu_2 and mut that may need to be filtered too. Since a two-d ! parallel transpose would be inefficient and because the fields that are ! not staggered in z have an extra layer anyway, these can be ! piggybacked. This is handled using latches to makes sure that *somebody* ! carries one or both of these on their back through the filtering and then ! copies them back afterwards. IMPORTANT NOTE: for simplicity, this routine ! is not completely general. It makes the following assumptions: ! ! 1) If both flag_mu and flag_mut are specified then flag_uv is also specified ! ! 2) If flag_uv is not specified, then only flag_mu and not flag_mut can be ! ! 3) If flag_mu is specified than either flag_uv or flag_t must be ! ! This is designed to keep the clutter to a minimum in solve_em. ! This is not intended to be a general abstraction of the polar filtering ! calls in in WRF solvers or if the solve_em algorithms change. ! If the needs of the calling solver change, this logic may have to be ! rewritten. ! ! !write(0,*)"",__LINE__,' short circuit ' !return !write(0,*)'pxft called from ',lineno call wrf_get_myproc(myproc) !write(20+myproc,*)ipex-ipsx+1,jpex-jpsx+1,' clat_xxx ' !do j = jpsx, jpex !do i = ipsx, ipex !write(20+myproc,*)grid%clat_xxx(i,j) !enddo !enddo !!!!!!!!!!!!!!!!!!!!!!! ! U & V IF ( flag_uv .EQ. 1 ) THEN IF ( piggyback_mu ) THEN grid%u_2(ips:ipe,kde,jps:jpe) = grid%mu_2(ips:ipe,jps:jpe) ENDIF #if ( defined( DM_PARALLEL ) && ( ! defined( STUBMPI ) ) ) # include "XPOSE_POLAR_FILTER_V_z2x.inc" CALL polar_filter_3d( grid%v_xxx, grid%clat_xxx, .false., & fft_filter_lat, dclat, & ids, ide, jds, jde, kds, kde-1, & imsx, imex, jmsx, jmex, kmsx, kmex, & ipsx, ipex, jpsx, jpex, kpsx, MIN(kde-1,kpex ) ) # include "XPOSE_POLAR_FILTER_V_x2z.inc" # include "XPOSE_POLAR_FILTER_U_z2x.inc" k_end = MIN(kde-1,kpex) IF ( piggyback_mu ) k_end = MIN(kde,kpex) CALL polar_filter_3d( grid%u_xxx, grid%clat_xxx, piggyback_mu, & fft_filter_lat, 0., & ids, ide, jds, jde, kds, kde, & imsx, imex, jmsx, jmex, kmsx, kmex, & ipsx, ipex, jpsx, jpex, kpsx, k_end ) # include "XPOSE_POLAR_FILTER_U_x2z.inc" #else CALL polar_filter_3d( grid%v_2, grid%clat, .false., & fft_filter_lat, dclat, & ids, ide, jds, jde, kds, kde, & ims, ime, jms, jme, kms, kme, & ips, ipe, jps, jpe, kps, MIN(kde-1,kpe) ) k_end = MIN(kde-1,kpe) IF ( piggyback_mu ) k_end = MIN(kde,kpe) CALL polar_filter_3d( grid%u_2, grid%clat, piggyback_mu, & fft_filter_lat, 0., & ids, ide, jds, jde, kds, kde-1, & ims, ime, jms, jme, kms, kme, & ips, ipe, jps, jpe, kps, k_end ) #endif IF ( piggyback_mu ) THEN grid%mu_2(ips:ipe,jps:jpe) = grid%u_2(ips:ipe,kde,jps:jpe) piggyback_mu = .FALSE. ENDIF ENDIF !!!!!!!!!!!!!!!!!!!!!!! ! T IF ( flag_t .EQ. 1 ) THEN IF ( piggyback_mu ) THEN grid%t_2(ips:ipe,kde,jps:jpe) = grid%mu_2(ips:ipe,jps:jpe) ENDIF #if ( defined( DM_PARALLEL ) && ( ! defined( STUBMPI ) ) ) # include "XPOSE_POLAR_FILTER_T_z2x.inc" k_end = MIN(kde-1,kpex) IF ( piggyback_mu ) k_end = MIN(kde,kpex) CALL polar_filter_3d( grid%t_xxx, grid%clat_xxx,piggyback_mu, & fft_filter_lat, 0., & ids, ide, jds, jde, kds, kde-1, & imsx, imex, jmsx, jmex, kmsx, kmex, & ipsx, ipex, jpsx, jpex, kpsx, k_end ) IF ( actual_distance_average ) THEN CALL filter_tracer ( grid%t_xxx , grid%clat_xxx , grid%mf_xxx , & grid%fft_filter_lat , grid%mf_fft , & pos_def, swap_pole_with_next_j , & ids, ide, jds, jde, kds, kde , & imsx, imex, jmsx, jmex, kmsx, kmex, & ipsx, ipex, jpsx, jpex, kpsx, kpex ) END IF # include "XPOSE_POLAR_FILTER_T_x2z.inc" #else k_end = MIN(kde-1,kpe) IF ( piggyback_mu ) k_end = MIN(kde,kpe) CALL polar_filter_3d( grid%t_2, grid%clat, piggyback_mu, & fft_filter_lat, 0., & ids, ide, jds, jde, kds, kde-1, & ims, ime, jms, jme, kms, kme, & ips, ipe, jps, jpe, kps, k_end ) IF ( actual_distance_average ) THEN CALL filter_tracer ( grid%t_2 , grid%clat , grid%msft , & grid%fft_filter_lat , grid%mf_fft , & pos_def, swap_pole_with_next_j , & ids, ide, jds, jde, kds, kde , & ims, ime, jms, jme, kms, kme, & ips, ipe, jps, jpe, kps, k_end ) END IF #endif IF ( piggyback_mu ) THEN grid%mu_2(ips:ipe,jps:jpe) = grid%t_2(ips:ipe,kde,jps:jpe) piggyback_mu = .FALSE. ENDIF ENDIF !!!!!!!!!!!!!!!!!!!!!!! ! W and PH IF ( flag_wph .EQ. 1 ) THEN ! W AND PH USE ALL LEVELS SO NEVER PIGGYBACK, MU IS OUT OF LUCK HERE #if ( defined( DM_PARALLEL ) && ( ! defined( STUBMPI ) ) ) # include "XPOSE_POLAR_FILTER_W_z2x.inc" CALL polar_filter_3d( grid%w_xxx, grid%clat_xxx, .false., & fft_filter_lat, 0., & ids, ide, jds, jde, kds, kde, & imsx, imex, jmsx, jmex, kmsx, kmex, & ipsx, ipex, jpsx, jpex, kpsx, kpex ) # include "XPOSE_POLAR_FILTER_W_x2z.inc" # include "XPOSE_POLAR_FILTER_PH_z2x.inc" CALL polar_filter_3d( grid%ph_xxx, grid%clat_xxx, .false., & fft_filter_lat, 0., & ids, ide, jds, jde, kds, kde, & imsx, imex, jmsx, jmex, kmsx, kmex, & ipsx, ipex, jpsx, jpex, kpsx, kpex ) # include "XPOSE_POLAR_FILTER_PH_x2z.inc" #else CALL polar_filter_3d( grid%w_2, grid%clat, .false., & fft_filter_lat, 0., & ids, ide, jds, jde, kds, kde, & ims, ime, jms, jme, kms, kme, & ips, ipe, jps, jpe, kps, kpe ) CALL polar_filter_3d( grid%ph_2, grid%clat, .false., & fft_filter_lat, 0., & ids, ide, jds, jde, kds, kde, & ims, ime, jms, jme, kms, kme, & ips, ipe, jps, jpe, kps, kpe ) #endif ENDIF !!!!!!!!!!!!!!!!!!!!!!! ! WW IF ( flag_ww .EQ. 1 ) THEN ! WW USES ALL LEVELS SO NEVER PIGGYBACK, MU IS OUT OF LUCK HERE #if ( defined( DM_PARALLEL ) && ( ! defined( STUBMPI ) ) ) # include "XPOSE_POLAR_FILTER_WW_z2x.inc" CALL polar_filter_3d( grid%ww_xxx, grid%clat_xxx, .false., & fft_filter_lat, 0., & ids, ide, jds, jde, kds, kde, & imsx, imex, jmsx, jmex, kmsx, kmex, & ipsx, ipex, jpsx, jpex, kpsx, kpex ) # include "XPOSE_POLAR_FILTER_WW_x2z.inc" #else CALL polar_filter_3d( grid%ww_m, grid%clat, .false., & fft_filter_lat, 0., & ids, ide, jds, jde, kds, kde, & ims, ime, jms, jme, kms, kme, & ips, ipe, jps, jpe, kps, kpe ) #endif ENDIF !!!!!!!!!!!!!!!!!!!!!!! ! RU AND RV IF ( flag_rurv .EQ. 1 ) THEN IF ( piggyback_mut ) THEN grid%ru_m(ips:ipe,kde,jps:jpe) = grid%mut(ips:ipe,jps:jpe) ENDIF #if ( defined( DM_PARALLEL ) && ( ! defined( STUBMPI ) ) ) # include "XPOSE_POLAR_FILTER_RV_z2x.inc" CALL polar_filter_3d( grid%rv_xxx, grid%clat_xxx, .false., & fft_filter_lat, dclat, & ids, ide, jds, jde, kds, kde, & imsx, imex, jmsx, jmex, kmsx, kmex, & ipsx, ipex, jpsx, jpex, kpsx, MIN(kpex,kde-1) ) # include "XPOSE_POLAR_FILTER_RV_x2z.inc" # include "XPOSE_POLAR_FILTER_RU_z2x.inc" k_end = MIN(kde-1,kpex) IF ( piggyback_mut ) k_end = MIN(kde,kpex) CALL polar_filter_3d( grid%ru_xxx, grid%clat_xxx, piggyback_mut, & fft_filter_lat, 0., & ids, ide, jds, jde, kds, kde, & imsx, imex, jmsx, jmex, kmsx, kmex, & ipsx, ipex, jpsx, jpex, kpsx, k_end ) #include "XPOSE_POLAR_FILTER_RU_x2z.inc" #else CALL polar_filter_3d( grid%rv_m, grid%clat, .false., & fft_filter_lat, dclat, & ids, ide, jds, jde, kds, kde, & ims, ime, jms, jme, kms, kme, & ips, ipe, jps, jpe, kps, MIN(kde-1,kpe) ) k_end = MIN(kde-1,kpe) IF ( piggyback_mut ) k_end = MIN(kde,kpe) CALL polar_filter_3d( grid%ru_m, grid%clat, piggyback_mut, & fft_filter_lat, 0., & ids, ide, jds, jde, kds, kde-1, & ims, ime, jms, jme, kms, kme, & ips, ipe, jps, jpe, kps, k_end ) #endif IF ( piggyback_mut ) THEN grid%mut(ips:ipe,jps:jpe) = grid%ru_m(ips:ipe,kde,jps:jpe) piggyback_mut = .FALSE. ENDIF ENDIF !!!!!!!!!!!!!!!!!!!!!!! ! MOIST IF ( flag_moist .GE. PARAM_FIRST_SCALAR ) THEN itrace = flag_moist #if ( defined( DM_PARALLEL ) && ( ! defined( STUBMPI ) ) ) # include "XPOSE_POLAR_FILTER_MOIST_z2x.inc" CALL polar_filter_3d( grid%fourd_xxx, grid%clat_xxx, .false. , & fft_filter_lat, 0., & ids, ide, jds, jde, kds, kde, & imsx, imex, jmsx, jmex, kmsx, kmex, & ipsx, ipex, jpsx, jpex, kpsx, MIN(kpex,kde-1) ) IF ( actual_distance_average ) THEN CALL filter_tracer ( grid%fourd_xxx , grid%clat_xxx , grid%mf_xxx , & grid%fft_filter_lat , grid%mf_fft , & pos_def, swap_pole_with_next_j , & ids, ide, jds, jde, kds, kde , & imsx, imex, jmsx, jmex, kmsx, kmex, & ipsx, ipex, jpsx, jpex, kpsx, kpex ) END IF # include "XPOSE_POLAR_FILTER_MOIST_x2z.inc" #else CALL polar_filter_3d( moist(ims,kms,jms,itrace), grid%clat, .false., & fft_filter_lat, 0., & ids, ide, jds, jde, kds, kde, & ims, ime, jms, jme, kms, kme, & ips, ipe, jps, jpe, kps, MIN(kpe,kde-1) ) IF ( actual_distance_average ) THEN CALL filter_tracer ( moist(ims,kms,jms,itrace) , grid%clat , grid%msft , & grid%fft_filter_lat , grid%mf_fft , & pos_def, swap_pole_with_next_j , & ids, ide, jds, jde, kds, kde , & ims, ime, jms, jme, kms, kme, & ips, ipe, jps, jpe, kps, MIN(kpe,kde-1) ) END IF #endif ENDIF !!!!!!!!!!!!!!!!!!!!!!! ! CHEM IF ( flag_chem .GE. PARAM_FIRST_SCALAR ) THEN itrace = flag_chem #if ( defined( DM_PARALLEL ) && ( ! defined( STUBMPI ) ) ) # include "XPOSE_POLAR_FILTER_CHEM_z2x.inc" CALL polar_filter_3d( grid%fourd_xxx, grid%clat_xxx, .false. , & fft_filter_lat, 0., & ids, ide, jds, jde, kds, kde, & imsx, imex, jmsx, jmex, kmsx, kmex, & ipsx, ipex, jpsx, jpex, kpsx, MIN(kpex,kde-1) ) IF ( actual_distance_average ) THEN CALL filter_tracer ( grid%fourd_xxx , grid%clat_xxx , grid%mf_xxx , & grid%fft_filter_lat , grid%mf_fft , & pos_def, swap_pole_with_next_j , & ids, ide, jds, jde, kds, kde , & imsx, imex, jmsx, jmex, kmsx, kmex, & ipsx, ipex, jpsx, jpex, kpsx, kpex ) END IF # include "XPOSE_POLAR_FILTER_CHEM_x2z.inc" #else CALL polar_filter_3d( chem(ims,kms,jms,itrace), grid%clat, .false. , & fft_filter_lat, 0., & ids, ide, jds, jde, kds, kde, & ims, ime, jms, jme, kms, kme, & ips, ipe, jps, jpe, kps, MIN(kpe,kde-1) ) IF ( actual_distance_average ) THEN CALL filter_tracer ( chem(ims,kms,jms,itrace) , grid%clat , grid%msft , & grid%fft_filter_lat , grid%mf_fft , & pos_def, swap_pole_with_next_j , & ids, ide, jds, jde, kds, kde , & ims, ime, jms, jme, kms, kme, & ips, ipe, jps, jpe, kps, MIN(kpe,kde-1) ) END IF #endif ENDIF !!!!!!!!!!!!!!!!!!!!!!! ! TRACER IF ( flag_tracer .GE. PARAM_FIRST_SCALAR ) THEN itrace = flag_tracer #if ( defined( DM_PARALLEL ) && ( ! defined( STUBMPI ) ) ) # include "XPOSE_POLAR_FILTER_TRACER_z2x.inc" CALL polar_filter_3d( grid%fourd_xxx, grid%clat_xxx, .false. , & fft_filter_lat, 0., & ids, ide, jds, jde, kds, kde, & imsx, imex, jmsx, jmex, kmsx, kmex, & ipsx, ipex, jpsx, jpex, kpsx, MIN(kpex,kde-1) ) IF ( actual_distance_average ) THEN CALL filter_tracer ( grid%fourd_xxx , grid%clat_xxx , grid%mf_xxx , & grid%fft_filter_lat , grid%mf_fft , & pos_def, swap_pole_with_next_j , & ids, ide, jds, jde, kds, kde , & imsx, imex, jmsx, jmex, kmsx, kmex, & ipsx, ipex, jpsx, jpex, kpsx, kpex ) END IF # include "XPOSE_POLAR_FILTER_TRACER_x2z.inc" #else CALL polar_filter_3d( tracer(ims,kms,jms,itrace), grid%clat, .false. , & fft_filter_lat, 0., & ids, ide, jds, jde, kds, kde, & ims, ime, jms, jme, kms, kme, & ips, ipe, jps, jpe, kps, MIN(kpe,kde-1) ) IF ( actual_distance_average ) THEN CALL filter_tracer ( tracer(ims,kms,jms,itrace) , grid%clat , grid%msft , & grid%fft_filter_lat , grid%mf_fft , & pos_def, swap_pole_with_next_j , & ids, ide, jds, jde, kds, kde , & ims, ime, jms, jme, kms, kme, & ips, ipe, jps, jpe, kps, MIN(kpe,kde-1) ) END IF #endif ENDIF !!!!!!!!!!!!!!!!!!!!!!! ! SCALAR IF ( flag_scalar .GE. PARAM_FIRST_SCALAR ) THEN itrace = flag_scalar #if ( defined( DM_PARALLEL ) && ( ! defined( STUBMPI ) ) ) # include "XPOSE_POLAR_FILTER_SCALAR_z2x.inc" CALL polar_filter_3d( grid%fourd_xxx , grid%clat_xxx, .false. , & fft_filter_lat, 0., & ids, ide, jds, jde, kds, kde, & imsx, imex, jmsx, jmex, kmsx, kmex, & ipsx, ipex, jpsx, jpex, kpsx, MIN(kpex,kde-1) ) IF ( actual_distance_average ) THEN CALL filter_tracer ( grid%fourd_xxx , grid%clat_xxx , grid%mf_xxx , & grid%fft_filter_lat , grid%mf_fft , & pos_def, swap_pole_with_next_j , & ids, ide, jds, jde, kds, kde , & imsx, imex, jmsx, jmex, kmsx, kmex, & ipsx, ipex, jpsx, jpex, kpsx, kpex ) END IF # include "XPOSE_POLAR_FILTER_SCALAR_x2z.inc" #else CALL polar_filter_3d( scalar(ims,kms,jms,itrace) , grid%clat, .false. , & fft_filter_lat, 0., & ids, ide, jds, jde, kds, kde, & ims, ime, jms, jme, kms, kme, & ips, ipe, jps, jpe, kps, MIN(kpe,kde-1) ) IF ( actual_distance_average ) THEN CALL filter_tracer ( scalar(ims,kms,jms,itrace) , grid%clat , grid%msft , & grid%fft_filter_lat , grid%mf_fft , & pos_def, swap_pole_with_next_j , & ids, ide, jds, jde, kds, kde , & ims, ime, jms, jme, kms, kme, & ips, ipe, jps, jpe, kps, MIN(kpe,kde-1) ) END IF #endif ENDIF IF ( flag_mu .EQ. 1 .AND. piggyback_mu ) THEN CALL wrf_error_fatal("mu needed to get piggybacked on a transpose and did not") ENDIF IF ( flag_mut .EQ. 1 .AND. piggyback_mut ) THEN CALL wrf_error_fatal("mut needed to get piggybacked on a transpose and did not") ENDIF !write(0,*)'pxft back to ',lineno RETURN END SUBROUTINE pxft SUBROUTINE polar_filter_3d( f, xlat, piggyback, fft_filter_lat, dvlat, & ids, ide, jds, jde, kds, kde, & ims, ime, jms, jme, kms, kme, & its, ite, jts, jte, kts, kte ) IMPLICIT NONE INTEGER , INTENT(IN ) :: ids, ide, jds, jde, kds, kde, & ims, ime, jms, jme, kms, kme, & its, ite, jts, jte, kts, kte REAL , INTENT(IN ) :: fft_filter_lat REAL , DIMENSION( ims:ime , kms:kme, jms:jme ) , INTENT(INOUT) :: f REAL , DIMENSION( ims:ime , jms:jme ) , INTENT(IN) :: xlat REAL , INTENT(IN) :: dvlat LOGICAL , INTENT(IN) :: piggyback REAL , DIMENSION(1:ide-ids,1:kte-kts+1) :: sheet REAL , DIMENSION(1:kte-kts+1) :: sheet_total REAL :: lat, avg, rnboxw INTEGER :: ig, jg, i, j, j_end, nx, ny, nmax, kw INTEGER :: k, nboxw, nbox2, istart, iend, overlap INTEGER, DIMENSION(6) :: wavenumber = (/ 1, 3, 7, 10, 13, 16 /) INTEGER :: fftflag ! Variables will stay in domain form since this routine is meaningless ! unless tile extent is the same as domain extent in E/W direction, i.e., ! the processor has access to all grid points in E/W direction. ! There may be other ways of doing FFTs, but we have not learned them yet... ! Check to make sure we have full access to all E/W points IF ((its /= ids) .OR. (ite /= ide)) THEN WRITE ( wrf_err_message , * ) 'module_polarfft: 3d: (its /= ids) or (ite /= ide)',its,ids,ite,ide CALL wrf_error_fatal ( TRIM( wrf_err_message ) ) END IF fftflag= 1 ! call double-precision fft ! fftflag= 0 ! call single-precision fft nx = ide-ids ! "U" stagger variables will be repeated by periodic BCs ny = kte-kts+1 ! we can filter extra level for variables that are non-Z-staggered lat = 0. j_end = MIN(jte, jde-1) IF (dvlat /= 0. .and. j_end == jde-1) j_end = jde DO j = jts, j_end ! jg is the J index in the global (domain) span of the array. jg = j-jds+1 ! determine whether or not to filter the data if(xlat(ids,j).gt.0.) then lat = xlat(ids,j)-dvlat else lat = xlat(ids,j)+dvlat endif IF (abs(lat) >= fft_filter_lat) THEN DO k=kts,kte DO i=ids,ide-1 sheet(i-ids+1,k-kts+1) = f(i,k,j) END DO END DO call polar_filter_fft_2d_ncar(nx,ny,sheet,lat,fft_filter_lat,piggyback,fftflag) DO k=kts,kte DO i=ids,ide-1 f(i,k,j) = sheet(i-ids+1,k-kts+1) END DO ! setting up ims-ime with x periodicity: ! enforce periodicity as in set_physical_bc3d DO i=1,ids-ims f(ids-i,k,j)=f(ide-i,k,j) END DO DO i=1,ime-ide+1 f(ide+i-1,k,j)=f(ids+i-1,k,j) END DO END DO END IF END DO ! outer j (latitude) loop END SUBROUTINE polar_filter_3d !------------------------------------------------------------------------------ SUBROUTINE polar_filter_fft_2d_ncar(nx,ny,fin,lat,filter_latitude,piggyback,fftflag) IMPLICIT NONE INTEGER , INTENT(IN) :: nx, ny REAL , DIMENSION(nx,ny), INTENT(INOUT) :: fin REAL , INTENT(IN) :: lat, filter_latitude LOGICAL, INTENT(IN) :: piggyback REAL :: pi, rcosref, freq, c, cf INTEGER :: k, fftflag REAL, DIMENSION(NX,NY) :: work REAL, DIMENSION(NX+15) :: wsave REAL(KIND=8), DIMENSION(NX,NY) :: fin8, work8 REAL(KIND=8), DIMENSION(NX+15) :: wsave8 INTEGER :: i, j REAL, dimension(nx,ny) :: fp INTEGER :: lensave, ier, nh, n1 INTEGER :: lot, jump, n, inc, lenr, lensav, lenwrk REAL, PARAMETER :: alpha = 0.0 REAL :: factor_k INTEGER :: ntop pi = ACOS(-1.) rcosref = 1./COS(filter_latitude*pi/180.) ! we are following the naming convention of the fftpack5 routines n = nx lot = ny lensav = n+15 inc = 1 lenr = nx*ny jump = nx lenwrk = lenr ntop = ny IF(piggyback) ntop = ny-1 ! forward transform ! initialize coefficients, place in wsave ! (should place this in init and save wsave at program start) if(fftflag.eq.0) then call rfftmi(n,wsave,lensav,ier) else call dfft1i(n,wsave8,lensav,ier) endif IF(ier /= 0) THEN write(a_message,*) ' error in rfftmi ',ier CALL wrf_message ( a_message ) END IF ! do the forward transform if(fftflag.eq.0) then call rfftmf(lot, jump, n, inc, fin, lenr, wsave, lensav, work, lenwrk, ier) else fin8 = fin do k=1,ny call dfft1f(n, inc, fin8(1,k), lenr, wsave8, lensav, work8, lenwrk, ier) enddo endif IF(ier /= 0) THEN write(a_message,*) ' error in rfftmf ',ier CALL wrf_message ( a_message ) END IF if(MOD(n,2) == 0) then nh = n/2 - 1 else nh = (n-1)/2 end if DO j=1,ny fp(1,j) = 1. ENDDO DO i=2,nh+1 freq=REAL(i-1)/REAL(n) c = (rcosref*COS(lat*pi/180.)/SIN(freq*pi))**2 ! c = MAX(0.,MIN(1.,c)) do j=1,ntop factor_k = (1.-alpha)+alpha*min(1.,float(ntop - j)/10.) cf = c*factor_k*factor_k cf = MAX(0.,MIN(1.,cf)) fp(2*(i-1),j) = cf fp(2*(i-1)+1,j) = cf enddo if(piggyback) then cf = MAX(0.,MIN(1.,c)) fp(2*(i-1),ny) = cf fp(2*(i-1)+1,ny) = cf endif END DO IF(MOD(n,2) == 0) THEN c = (rcosref*COS(lat*pi/180.))**2 ! c = MAX(0.,MIN(1.,c)) do j=1,ntop factor_k = (1.-alpha)+alpha*min(1.,float(ntop - j)/10.) cf = c*factor_k*factor_k cf = MAX(0.,MIN(1.,cf)) fp(n,j) = cf enddo if(piggyback) then cf = MAX(0.,MIN(1.,c)) fp(n,ny) = cf endif END IF if(fftflag.eq.0) then do j=1,ny do i=1,nx fin(i,j) = fp(i,j)*fin(i,j) enddo enddo else do j=1,ny do i=1,nx fin8(i,j) = fp(i,j)*fin8(i,j) enddo enddo endif ! do the backward transform if(fftflag.eq.0) then call rfftmb(lot, jump, n, inc, fin, lenr, wsave, lensav, work, lenwrk, ier) else do k=1,ny call dfft1b(n, inc, fin8(1,k), lenr, wsave8, lensav, work8, lenwrk, ier) enddo fin= fin8 endif IF(ier /= 0) THEN write(a_message,*) ' error in rfftmb ',ier CALL wrf_message ( a_message ) END IF END SUBROUTINE polar_filter_fft_2d_ncar !--------------------------------------------------------------------- SUBROUTINE filter_tracer ( tr3d_in , xlat , msftx , & fft_filter_lat , mf_fft , & pos_def , swap_pole_with_next_j , & ids , ide , jds , jde , kds , kde , & ims , ime , jms , jme , kms , kme , & its , ite , jts , jte , kts , kte ) IMPLICIT NONE INTEGER , INTENT(IN) :: ids , ide , jds , jde , kds , kde , & ims , ime , jms , jme , kms , kme , & its , ite , jts , jte , kts , kte REAL , INTENT(IN) :: fft_filter_lat , mf_fft REAL , DIMENSION(ims:ime,kms:kme,jms:jme) , INTENT(INOUT) :: tr3d_in REAL , DIMENSION(ims:ime,jms:jme) , INTENT(IN) :: xlat , msftx LOGICAL , INTENT(IN) :: pos_def , swap_pole_with_next_j ! Local vars INTEGER :: i , j , j_lat_pos , j_lat_neg , k INTEGER :: i_kicker , ik , i1, i2, i3, i4 INTEGER :: i_left , i_right , ii , count REAL :: length_scale , sum REAL , DIMENSION(its:ite,jts:jte) :: tr_in, tr_out CHARACTER (LEN=256) :: message ! The filtering is a simple average on a latitude loop. Possibly a LONG list of ! numbers. We assume that ALL of the 2d arrays have been transposed so that ! each patch has the entire domain size of the i-dimension available. IF ( ( its .NE. ids ) .OR. ( ite .NE. ide ) ) THEN CALL wrf_error_fatal ( 'filtering assumes all values on X' ) END IF ! Starting at the south pole, we find where the ! grid distance is big enough, then go back a point. Continuing to the ! north pole, we find the first small grid distance. These are the ! computational latitude loops and the associated computational poles. j_lat_neg = 0 j_lat_pos = jde + 1 loop_neg : DO j = MIN(jde-1,jte) , jts , -1 IF ( xlat(its,j) .LT. 0.0 ) THEN IF ( ABS(xlat(its,j)) .GE. fft_filter_lat ) THEN j_lat_neg = j EXIT loop_neg END IF END IF END DO loop_neg loop_pos : DO j = jts , MIN(jde-1,jte) IF ( xlat(its,j) .GT. 0.0 ) THEN IF ( xlat(its,j) .GE. fft_filter_lat ) THEN j_lat_pos = j EXIT loop_pos END IF END IF END DO loop_pos ! Initialize the starting values for the averages. DO k = kts, kte DO j = jts , MIN(jde-1,jte) DO i = its , MIN(ide-1,ite) tr_in(i,j) = tr3d_in(i,k,j) tr_out(i,j) = tr_in(i,j) END DO END DO ! Filter the fields at the negative lats. DO j = MIN(j_lat_neg,jte) , jts , -1 ! i_kicker = MIN( MAX ( NINT(msftx(its,j)/2) , 1 ) , (ide - ids) / 2 ) i_kicker = MIN( MAX ( NINT(msftx(its,j)/mf_fft/2) , 1 ) , (ide - ids) / 2 ) ! WRITE (message,FMT='(a,i4,a,i4,f6.2,1x,f6.1,f6.1)') 'SOUTH j = ' , j, ', kicker = ',i_kicker,xlat(its,j),msftx(its,j),mf_fft ! CALL wrf_debug ( 0 , TRIM(message) ) DO i = its , MIN(ide-1,ite) sum = 0. count = 0 DO ik = 1 , i_kicker/2 ii = i-ik IF ( ii .GE. ids ) THEN i_left = ii ELSE i_left = ( ii - ids ) + (ide-1)+1 END IF ii = i+ik IF ( ii .LE. ide-1 ) THEN i_right = ii ELSE i_right = ( ii - (ide-1) ) + its-1 END IF sum = sum + tr_in(i_left,j) + tr_in(i_right,j) count = count + 1 END DO tr_out(i,j) = ( tr_in(i,j) + sum ) / REAL ( 2 * count + 1 ) END DO END DO ! Filter the fields at the positive lats. DO j = MAX(j_lat_pos,jts) , MIN(jde-1,jte) ! i_kicker = MIN( MAX ( NINT(msftx(its,j)/2) , 1 ) , (ide - ids) / 2 ) i_kicker = MIN( MAX ( NINT(msftx(its,j)/mf_fft/2) , 1 ) , (ide - ids) / 2 ) ! WRITE (message,FMT='(a,i4,a,i4,f6.2,1x,f6.1,f6.1)') 'NORTH j = ' , j, ', kicker = ',i_kicker,xlat(its,j),msftx(its,j),mf_fft ! CALL wrf_debug ( 0 , TRIM(message) ) DO i = its , MIN(ide-1,ite) count = 0 sum = 0. DO ik = 1 , i_kicker/2 ii = i-ik IF ( ii .GE. ids ) THEN i_left = ii ELSE i_left = ( ii - ids ) + (ide-1)+1 END IF ii = i+ik IF ( ii .LE. ide-1 ) THEN i_right = ii ELSE i_right = ( ii - (ide-1) ) + its-1 END IF sum = sum + tr_in(i_left,j) + tr_in(i_right,j) count = count + 1 END DO tr_out(i,j) = ( tr_in(i,j) + sum ) / REAL ( 2 * count + 1 ) END DO END DO ! Set output values for whole patch. DO j = jts , MIN(jde-1,jte) DO i = its , MIN(ide-1,ite) tr3d_in(i,k,j) = tr_out(i,j) END DO END DO ! Positive definite on scalars? IF ( pos_def ) THEN DO j = jts , MIN(jde-1,jte) DO i = its , MIN(ide-1,ite) tr3d_in(i,k,j) = MAX( tr3d_in(i,k,j) , 0. ) END DO END DO END IF ! Remove values at j=1 and j=jde-1 locations, set them to the rows just next to them. IF ( swap_pole_with_next_j ) THEN IF ( jts .EQ. jds ) THEN DO i = its , MIN(ide-1,ite) tr3d_in(i,k,jts) = tr3d_in(i,k,jts+1) END DO END IF IF ( jte .EQ. jde ) THEN DO i = its , MIN(ide-1,ite) tr3d_in(i,k,jte-1) = tr3d_in(i,k,jte-2) END DO END IF END IF END DO ! k-loop END SUBROUTINE filter_tracer !--------------------------------------------------------------------- SUBROUTINE filter_tracer_old ( tr3d_in , xlat , msftx , fft_filter_lat , & ids , ide , jds , jde , kds , kde , & ims , ime , jms , jme , kms , kme , & its , ite , jts , jte , kts , kte ) IMPLICIT NONE INTEGER , INTENT(IN) :: ids , ide , jds , jde , kds , kde , & ims , ime , jms , jme , kms , kme , & its , ite , jts , jte , kts , kte REAL , INTENT(IN) :: fft_filter_lat REAL , DIMENSION(ims:ime,kms:kme,jms:jme) , INTENT(INOUT) :: tr3d_in REAL , DIMENSION(ims:ime,jms:jme) , INTENT(IN) :: xlat , msftx ! Local vars INTEGER :: i , j , j_lat_pos , j_lat_neg , k INTEGER :: i_kicker , ik , i1, i2, i3, i4 REAL :: length_scale , sum REAL , DIMENSION(its:ite,jts:jte) :: tr_in, tr_out ! The filtering is a simple average on a latitude loop. Possibly a LONG list of ! numbers. We assume that ALL of the 2d arrays have been transposed so that ! each patch has the entire domain size of the i-dim local. IF ( ( its .NE. ids ) .OR. ( ite .NE. ide ) ) THEN CALL wrf_error_fatal ( 'filtering assumes all values on X' ) END IF ! Starting at the south pole, we find where the ! grid distance is big enough, then go back a point. Continuing to the ! north pole, we find the first small grid distance. These are the ! computational latitude loops and the associated computational poles. j_lat_neg = 0 j_lat_pos = jde + 1 loop_neg : DO j = jts , MIN(jde-1,jte) IF ( xlat(its,j) .LT. 0.0 ) THEN IF ( ABS(xlat(its,j)) .LT. fft_filter_lat ) THEN j_lat_neg = j - 1 EXIT loop_neg END IF END IF END DO loop_neg loop_pos : DO j = jts , MIN(jde-1,jte) IF ( xlat(its,j) .GT. 0.0 ) THEN IF ( xlat(its,j) .GE. fft_filter_lat ) THEN j_lat_pos = j EXIT loop_pos END IF END IF END DO loop_pos ! Set output values to initial input topo values for whole patch. DO k = kts, kte DO j = jts , MIN(jde-1,jte) DO i = its , MIN(ide-1,ite) tr_in(i,j) = tr3d_in(i,k,j) tr_out(i,j) = tr_in(i,j) END DO END DO ! Filter the topo at the negative lats. DO j = j_lat_neg , jts , -1 i_kicker = MIN( MAX ( NINT(msftx(its,j)) , 1 ) , (ide - ids) / 2 ) ! print *,'j = ' , j, ', kicker = ',i_kicker DO i = its , MIN(ide-1,ite) IF ( ( i - i_kicker .GE. its ) .AND. ( i + i_kicker .LE. ide-1 ) ) THEN sum = 0.0 DO ik = 1 , i_kicker sum = sum + tr_in(i+ik,j) + tr_in(i-ik,j) END DO tr_out(i,j) = ( tr_in(i,j) + sum ) / REAL ( 2 * i_kicker + 1 ) ELSE IF ( ( i - i_kicker .LT. its ) .AND. ( i + i_kicker .LE. ide-1 ) ) THEN sum = 0.0 DO ik = 1 , i_kicker sum = sum + tr_in(i+ik,j) END DO i1 = i - i_kicker + ide -1 i2 = ide-1 i3 = ids i4 = i-1 DO ik = i1 , i2 sum = sum + tr_in(ik,j) END DO DO ik = i3 , i4 sum = sum + tr_in(ik,j) END DO tr_out(i,j) = ( tr_in(i,j) + sum ) / REAL ( 2 * i_kicker + 1 ) ELSE IF ( ( i - i_kicker .GE. its ) .AND. ( i + i_kicker .GT. ide-1 ) ) THEN sum = 0.0 DO ik = 1 , i_kicker sum = sum + tr_in(i-ik,j) END DO i1 = i+1 i2 = ide-1 i3 = ids i4 = ids + ( i_kicker+i ) - ide DO ik = i1 , i2 sum = sum + tr_in(ik,j) END DO DO ik = i3 , i4 sum = sum + tr_in(ik,j) END DO tr_out(i,j) = ( tr_in(i,j) + sum ) / REAL ( 2 * i_kicker + 1 ) END IF END DO END DO ! Filter the topo at the positive lats. DO j = j_lat_pos , MIN(jde-1,jte) i_kicker = MIN( MAX ( NINT(msftx(its,j)) , 1 ) , (ide - ids) / 2 ) ! print *,'j = ' , j, ', kicker = ',i_kicker DO i = its , MIN(ide-1,ite) IF ( ( i - i_kicker .GE. its ) .AND. ( i + i_kicker .LE. ide-1 ) ) THEN sum = 0.0 DO ik = 1 , i_kicker sum = sum + tr_in(i+ik,j) + tr_in(i-ik,j) END DO tr_out(i,j) = ( tr_in(i,j) + sum ) / REAL ( 2 * i_kicker + 1 ) ELSE IF ( ( i - i_kicker .LT. its ) .AND. ( i + i_kicker .LE. ide-1 ) ) THEN sum = 0.0 DO ik = 1 , i_kicker sum = sum + tr_in(i+ik,j) END DO i1 = i - i_kicker + ide -1 i2 = ide-1 i3 = ids i4 = i-1 DO ik = i1 , i2 sum = sum + tr_in(ik,j) END DO DO ik = i3 , i4 sum = sum + tr_in(ik,j) END DO tr_out(i,j) = ( tr_in(i,j) + sum ) / REAL ( 2 * i_kicker + 1 ) ELSE IF ( ( i - i_kicker .GE. its ) .AND. ( i + i_kicker .GT. ide-1 ) ) THEN sum = 0.0 DO ik = 1 , i_kicker sum = sum + tr_in(i-ik,j) END DO i1 = i+1 i2 = ide-1 i3 = ids i4 = ids + ( i_kicker+i ) - ide DO ik = i1 , i2 sum = sum + tr_in(ik,j) END DO DO ik = i3 , i4 sum = sum + tr_in(ik,j) END DO tr_out(i,j) = ( tr_in(i,j) + sum ) / REAL ( 2 * i_kicker + 1 ) END IF END DO END DO ! Set output values to initial input topo values for whole patch. DO j = jts , MIN(jde-1,jte) DO i = its , MIN(ide-1,ite) tr3d_in(i,k,j) = tr_out(i,j) END DO END DO END DO ! k-loop END SUBROUTINE filter_tracer_old !--------------------------------------------------------------------- END MODULE module_polarfft