conv_func_compute count_distinct_var6d display_volgrid6d_var index_var6d map_distinct_var6d map_inv_distinct_var6d pack_distinct_var6d varbufr2vargrib varbufr2vargrib_convert vargrib2varbufr vargrib2varbufr_convert volgrid6d_var_hor_comp_index volgrid6d_var_is_hor_comp volgrid6d_var_miss volgrid6d_var_new volgrid6d_var_normalize

varbufr2vargrib() subroutine, public varbufr2vargrib ( type(vol7d_var), dimension(:), intent(in) varbufr, type(volgrid6d_var), dimension(:), intent(out) vargrib, type(conv_func), dimension(:), pointer c_func, type(grid_id), intent(in), optional grid_id_template ) Convert a vol7d_var array object into a physically equivalent volgrid6d_var array object. This method converts a bufr-like array of physical variables vargrib, to an array of grib-like variables varbufr. Unlike the opposite method vargrib2varbufr, in this case the conversion is not uniqe and at the moment the first matching grib-like variable is chosen, without any control over the choice process. The output array must have enough room for the converted variables. The method additionally allocates a conv_func array object of the same size, which can successively be used to convert the numerical values of the fields associated to varbufr to the corresponding fields in the grib-like representation. c_func will have to be deallocated by the calling procedure. If a conversion is not successful, the corresponding output variable is set to volgrid6d_var_miss and the conversion function to conv_func_miss. Parametri [in] varbufr array of input bufr-like variables [out] vargrib array of output grib-like variables c_func pointer to an array of the corresponding conv_func objects, allocated in the method [in] grid_id_template a template (typically grib_api) to which data will be finally exported, it helps in improving variable conversion Definizione alla linea 1170 del file volgrid6d_var_class.F90. ! Copyright (C) 2010 ARPA-SIM <urpsim@smr.arpa.emr.it> ! authors: ! Davide Cesari <dcesari@arpa.emr.it> ! Paolo Patruno <ppatruno@arpa.emr.it> ! This program 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 2 of ! the License, or (at your option) any later version. ! This program 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 this program. If not, see <http://www.gnu.org/licenses/>. #include "config.h" !> Class for managing physical variables in a grib 1/2 fashion. !! This module defines a class which can represent Earth-science !! related physical variables, following the classification scheme !! adopted by WMO for grib1 and grib2 parameter definition. It also !! defines some methods for mapping \a volgrid6d_var variables and !! converting the corresponding fields to a matching \a vol7d_var !! object defined in \a vol7d_var_class module, which, unlike the !! variables defined here, defines univocally a physical quantity. !! !! \ingroup volgrid6d MODULE volgrid6d_var_class USE kinds USE missing_values USE err_handling USE vol7d_var_class USE file_utilities USE grid_id_class IMPLICIT NONE !> Definition of a physical variable in grib coding style. !! \a volgrid6d_var members are public, thus they can be freely !! altered, but it is advisable to set them through the !! volgrid6d_var_class::init constructor. TYPE volgrid6d_var integer :: centre !< centre integer :: category !< grib2: category / grib1: grib table version number integer :: number !< parameter number integer :: discipline !< grib2: discipline / grib1: 255 CHARACTER(len=65) :: description !< optional textual description of the variable CHARACTER(len=24) :: unit !< optional textual description of the variable's unit END TYPE volgrid6d_var TYPE(volgrid6d_var),PARAMETER :: volgrid6d_var_miss= & volgrid6d_var(imiss,imiss,imiss,imiss,cmiss,cmiss) !< missing value volgrid6d_var. TYPE(vol7d_var),PARAMETER :: vol7d_var_horstag(2) = (/ & vol7d_var('B11003', '', '', 0, 0, 0, 0, 0, 0), & vol7d_var('B11004', '', '', 0, 0, 0, 0, 0, 0) & /) TYPE(vol7d_var),PARAMETER :: vol7d_var_horcomp(4) = (/ &! RESHAPE( (/ & vol7d_var('B11003', '', '', 0, 0, 0, 0, 0, 0), & vol7d_var('B11004', '', '', 0, 0, 0, 0, 0, 0), & vol7d_var('B11200', '', '', 0, 0, 0, 0, 0, 0), & vol7d_var('B11201', '', '', 0, 0, 0, 0, 0, 0) & /) !/), (/2,2/)) ! bug in gfortran !> Class defining a real conversion function between units. It is !! used to numerically convert a value expressed as a \a volgrid6d_var !! variable in a value expressed as a \a vol7d_var variable and !! vice-versa. At the moment only a linear conversion is !! supported. Objects of this class are returned only by the \a !! vargrib2varbufr \a varbufr2vargrib, and \a convert methods and are !! used in the \a convert and \a compute methods defined in this !! MODULE. TYPE conv_func PRIVATE REAL :: a, b END TYPE conv_func TYPE(conv_func), PARAMETER :: conv_func_miss=conv_func(rmiss,rmiss) TYPE(conv_func), PARAMETER :: conv_func_identity=conv_func(1.0,0.0) TYPE vg6d_v7d_var_conv TYPE(volgrid6d_var) :: vg6d_var TYPE(vol7d_var) :: v7d_var TYPE(conv_func) :: c_func ! aggiungere informazioni ad es. su rotazione del vento END TYPE vg6d_v7d_var_conv TYPE(vg6d_v7d_var_conv), PARAMETER :: vg6d_v7d_var_conv_miss= & vg6d_v7d_var_conv(volgrid6d_var_miss, vol7d_var_miss, conv_func_miss) TYPE(vg6d_v7d_var_conv), ALLOCATABLE :: conv_fwd(:), conv_bwd(:) !> Initialize a \a volgrid6d_var object with the optional arguments provided. !! If an argument is not provided, the corresponding object member and !! those depending on it will be set to missing. For grib1-style !! variables, the \a discipline argument must be omitted, it will be !! set to 255 (grib missing value). !! !! \param this TYPE(volgrid6d_var),INTENT(INOUT) object to be initialized !! \param centre INTEGER,INTENT(in),OPTIONAL centre !! \param category INTEGER,INTENT(in),OPTIONAL grib2: category / grib1: grib table version number !! \param number INTEGER,INTENT(in),OPTIONAL parameter number !! \param discipline INTEGER,INTENT(in),OPTIONAL grib2: discipline / grib1: 255 !! \param description CHARACTER(len=*),INTENT(in),OPTIONAL optional textual description of the variable !! \param unit CHARACTER(len=*),INTENT(in),OPTIONAL optional textual description of the variable's unit INTERFACE init MODULE PROCEDURE volgrid6d_var_init END INTERFACE !> Destructor for the corresponding object, it assigns it to a missing value. !! \param this TYPE(volgrid6d_var) object to be destroyed INTERFACE delete MODULE PROCEDURE volgrid6d_var_delete END INTERFACE INTERFACE c_e MODULE PROCEDURE volgrid6d_var_c_e END INTERFACE !> Logical equality operators for objects of the classes \a !! volgrid6d_var and \a conv_func. !! They are all defined as \c ELEMENTAL thus work also on arrays of !! any shape. INTERFACE OPERATOR (==) MODULE PROCEDURE volgrid6d_var_eq, conv_func_eq END INTERFACE !> Logical inequality operators for objects of the classes \a !! volgrid6d_var and \a conv_func. !! They are all defined as \c ELEMENTAL thus work also on arrays of !! any shape. INTERFACE OPERATOR (/=) MODULE PROCEDURE volgrid6d_var_ne, conv_func_ne END INTERFACE #define VOL7D_POLY_TYPE TYPE(volgrid6d_var) #define VOL7D_POLY_TYPES _var6d #include "array_utilities_pre.F90" !> Display on the screen a brief content of object INTERFACE display MODULE PROCEDURE display_volgrid6d_var END INTERFACE !> Compose two conversions into a single one. !! Unlike scalar multiplication (and like matrix multiplication) here !! a*b /= b*a. By convention, the second factor is applied first in !! the result. INTERFACE OPERATOR (*) MODULE PROCEDURE conv_func_mult END INTERFACE OPERATOR (*) !> Apply the conversion function \a this to \a values. !! function version INTERFACE compute MODULE PROCEDURE conv_func_compute END INTERFACE !> Apply the conversion function \a this to \a values. !! subroutine version INTERFACE convert MODULE PROCEDURE varbufr2vargrib_convert, vargrib2varbufr_convert END INTERFACE PRIVATE PUBLIC volgrid6d_var, volgrid6d_var_miss, volgrid6d_var_new, init, delete, & c_e, volgrid6d_var_normalize, & OPERATOR(==), OPERATOR(/=), OPERATOR(*), & count_distinct, pack_distinct, count_and_pack_distinct, & map_distinct, map_inv_distinct, & index, display, & vargrib2varbufr, varbufr2vargrib, & conv_func, conv_func_miss, compute, convert, & volgrid6d_var_hor_comp_index, volgrid6d_var_is_hor_comp CONTAINS ELEMENTAL FUNCTION volgrid6d_var_new(centre, category, number, & discipline, description, unit) RESULT(this) integer,INTENT(in),OPTIONAL :: centre !< centre integer,INTENT(in),OPTIONAL :: category !< grib2: category / grib1: grib table version number integer,INTENT(in),OPTIONAL :: number !< parameter number integer,INTENT(in),OPTIONAL :: discipline !< grib2: discipline CHARACTER(len=*),INTENT(in),OPTIONAL :: description !< textual description of the variable CHARACTER(len=*),INTENT(in),OPTIONAL :: unit !< textual description of the variable's unit TYPE(volgrid6d_var) :: this !< object to be initialised CALL init(this, centre, category, number, discipline, description, unit) END FUNCTION volgrid6d_var_new ! documented in the interface ELEMENTAL SUBROUTINE volgrid6d_var_init(this, centre, category, number, discipline,description,unit) TYPE(volgrid6d_var),INTENT(INOUT) :: this ! object to be initialized INTEGER,INTENT(in),OPTIONAL :: centre ! centre INTEGER,INTENT(in),OPTIONAL :: category ! grib2: category / grib1: grib table version number INTEGER,INTENT(in),OPTIONAL :: number ! parameter number INTEGER,INTENT(in),OPTIONAL :: discipline ! grib2: discipline / grib1: 255 CHARACTER(len=*),INTENT(in),OPTIONAL :: description ! optional textual description of the variable CHARACTER(len=*),INTENT(in),OPTIONAL :: unit ! optional textual description of the variable's unit IF (PRESENT(centre)) THEN this%centre = centre ELSE this%centre = imiss this%category = imiss this%number = imiss this%discipline = imiss RETURN ENDIF IF (PRESENT(category)) THEN this%category = category ELSE this%category = imiss this%number = imiss this%discipline = imiss RETURN ENDIF IF (PRESENT(number)) THEN this%number = number ELSE this%number = imiss this%discipline = imiss RETURN ENDIF ! se sono arrivato fino a qui ho impostato centre, category e number !per il grib 1 manca discipline e imposto 255 (missing del grib2) IF (PRESENT(discipline)) THEN this%discipline = discipline ELSE this%discipline = 255 ENDIF IF (PRESENT(description)) THEN this%description = description ELSE this%description = cmiss ENDIF IF (PRESENT(unit)) THEN this%unit = unit ELSE this%unit = cmiss ENDIF END SUBROUTINE volgrid6d_var_init ! documented in the interface SUBROUTINE volgrid6d_var_delete(this) TYPE(volgrid6d_var),INTENT(INOUT) :: this this%centre = imiss this%category = imiss this%number = imiss this%discipline = imiss this%description = cmiss this%unit = cmiss END SUBROUTINE volgrid6d_var_delete ELEMENTAL FUNCTION volgrid6d_var_c_e(this) RESULT(c_e) TYPE(volgrid6d_var),INTENT(IN) :: this LOGICAL :: c_e c_e = this /= volgrid6d_var_miss END FUNCTION volgrid6d_var_c_e ELEMENTAL FUNCTION volgrid6d_var_eq(this, that) RESULT(res) TYPE(volgrid6d_var),INTENT(IN) :: this, that LOGICAL :: res IF (this%discipline == that%discipline) THEN IF (this%discipline == 255) THEN ! grib1, WMO tables are all equivalent res = ((this%category == that%category) .OR. & (this%category >= 1 .AND. this%category <=3 .AND. & that%category >= 1 .AND. that%category <=3)) .AND. & this%number == that%number IF ((this%category >= 128 .AND. this%category <= 254) .OR. & (this%number >= 128 .AND. this%number <= 254)) THEN res = res .AND. this%centre == that%centre ! local definition, centre matters ENDIF ELSE ! grib2 res = this%category == that%category .AND. & this%number == that%number IF ((this%discipline >= 192 .AND. this%discipline <= 254) .OR. & (this%category >= 192 .AND. this%category <= 254) .OR. & (this%number >= 192 .AND. this%number <= 254)) THEN res = res .AND. this%centre == that%centre ! local definition, centre matters ENDIF ENDIF ELSE ! different edition or different discipline res = .false. ENDIF END FUNCTION volgrid6d_var_eq ELEMENTAL FUNCTION volgrid6d_var_ne(this, that) RESULT(res) TYPE(volgrid6d_var),INTENT(IN) :: this, that LOGICAL :: res res = .NOT.(this == that) END FUNCTION volgrid6d_var_ne #include "array_utilities_inc.F90" !> Display on the screen a brief content of \a volgrid6d_var object. SUBROUTINE display_volgrid6d_var(this) TYPE(volgrid6d_var),INTENT(in) :: this !< volgrid6d_var object to display print*,"GRIDVAR: ",this%centre,this%discipline,this%category,this%number END SUBROUTINE display_volgrid6d_var !> Convert a \a volgrid6d_var array object into a physically equivalent !! \a vol7d_var array object. This method converts a grib-like array !! of physical variables \a vargrib, to an array of unique, physically !! based, bufr-like variables \a varbufr. The output array must have !! enough room for the converted variables. The method additionally !! allocates a \a conv_func array object of the same size, which can !! successively be used to convert the numerical values of the fields !! associated to \a vargrib to the corresponding fields in the \a !! bufr-like representation. \a c_func will have to be deallocated by !! the calling procedure. If a conversion is not successful, the !! corresponding output variable is set to \a vol7d_var_miss and the !! conversion function to \a conv_func_miss. SUBROUTINE vargrib2varbufr(vargrib, varbufr, c_func) TYPE(volgrid6d_var),INTENT(in) :: vargrib(:) !< array of input grib-like variables TYPE(vol7d_var),INTENT(out) :: varbufr(:) !< array of output bufr-like variables TYPE(conv_func),POINTER :: c_func(:) !< pointer to an array of the corresponding \a conv_func objects, allocated in the method INTEGER :: i, n, stallo n = min(SIZE(varbufr), SIZE(vargrib)) ALLOCATE(c_func(n),stat=stallo) IF (stallo /= 0) THEN call l4f_log(l4f_fatal,"allocating memory") call raise_fatal_error() ENDIF DO i = 1, n varbufr(i) = convert(vargrib(i), c_func(i)) ENDDO END SUBROUTINE vargrib2varbufr !> Convert a \a volgrid6d_var object into a physically equivalent !! \a vol7d_var object. This method returns a physically based, !! bufr-like representation of type \a vol7d_var of the grib-like !! input physical variable \a vargrib. The method optionally returns !! a \a conv_func object which can successively be used to convert the !! numerical values of the field associated to \a vargrib to the !! corresponding fields in the bufr-like representation. If the !! conversion is not successful, the output variable is !! set to \a vol7d_var_miss and the conversion function to \a !! conv_func_miss. FUNCTION vargrib2varbufr_convert(vargrib, c_func) RESULT(convert) TYPE(volgrid6d_var),INTENT(in) :: vargrib !< input grib-like variable TYPE(conv_func),INTENT(out),OPTIONAL :: c_func !< corresponding \a conv_func object TYPE(vol7d_var) :: convert INTEGER :: i IF (.NOT. ALLOCATED(conv_fwd)) CALL vg6d_v7d_var_conv_setup() DO i = 1, SIZE(conv_fwd) IF (vargrib == conv_fwd(i)%vg6d_var) THEN convert = conv_fwd(i)%v7d_var IF (PRESENT(c_func)) c_func = conv_fwd(i)%c_func RETURN ENDIF ENDDO ! not found convert = vol7d_var_miss IF (PRESENT(c_func)) c_func = conv_func_miss ! set hint for backwards conversion convert%gribhint(:) = (/vargrib%centre, vargrib%category, vargrib%number, & vargrib%discipline/) CALL l4f_log(l4f_warn, 'vargrib2varbufr: variable '// & trim(to_char(vargrib%centre))//':'//trim(to_char(vargrib%category))//':'// & trim(to_char(vargrib%number))//':'//trim(to_char(vargrib%discipline))// & ' not found in table') END FUNCTION vargrib2varbufr_convert !> Convert a \a vol7d_var array object into a physically equivalent !! \a volgrid6d_var array object. This method converts a bufr-like !! array of physical variables \a vargrib, to an array of grib-like !! variables \a varbufr. Unlike the opposite method \a !! vargrib2varbufr, in this case the conversion is not uniqe and at !! the moment the first matching grib-like variable is chosen, without !! any control over the choice process. The output array must have !! enough room for the converted variables. The method additionally !! allocates a \a conv_func array object of the same size, which can !! successively be used to convert the numerical values of the fields !! associated to \a varbufr to the corresponding fields in the \a !! grib-like representation. \a c_func will have to be deallocated by !! the calling procedure. If a conversion is not successful, the !! corresponding output variable is set to \a volgrid6d_var_miss and !! the conversion function to \a conv_func_miss. SUBROUTINE varbufr2vargrib(varbufr, vargrib, c_func, grid_id_template) TYPE(vol7d_var),INTENT(in) :: varbufr(:) !< array of input bufr-like variables TYPE(volgrid6d_var),INTENT(out) :: vargrib(:) !< array of output grib-like variables TYPE(conv_func),POINTER :: c_func(:) !< pointer to an array of the corresponding \a conv_func objects, allocated in the method TYPE(grid_id),INTENT(in),OPTIONAL :: grid_id_template !< a template (typically grib_api) to which data will be finally exported, it helps in improving variable conversion INTEGER :: i, n, stallo n = min(SIZE(varbufr), SIZE(vargrib)) ALLOCATE(c_func(n),stat=stallo) IF (stallo /= 0) THEN CALL l4f_log(l4f_fatal,"allocating memory") CALL raise_fatal_error() ENDIF DO i = 1, n vargrib(i) = convert(varbufr(i), c_func(i), grid_id_template) ENDDO END SUBROUTINE varbufr2vargrib !> Convert a \a vol7d_var object into a physically equivalent !! \a volgrid6d_var object. This method returns a grib-like !! representation of type \a volgrid6d_var of the bufr-like input !! physical variable \a varbufr. Unlike the opposite \a convert !! method, in this case the conversion is not uniqe and at the moment !! the first matching grib-like variable is chosen, without any !! control over the choice process. The method optionally returns a !! \a conv_func object which can successively be used to convert the !! numerical values of the field associated to \a varbufr to the !! corresponding fields in the grib-like representation. If the !! conversion is not successful, the output variable is set to \a !! volgrid6d_var_miss and the conversion function to \a !! conv_func_miss. FUNCTION varbufr2vargrib_convert(varbufr, c_func, grid_id_template) RESULT(convert) TYPE(vol7d_var),INTENT(in) :: varbufr !< input bufr-like variable TYPE(conv_func),INTENT(out),OPTIONAL :: c_func !< corresponding \a conv_func object TYPE(grid_id),INTENT(in),OPTIONAL :: grid_id_template !< a template (typically grib_api) to which data will be finally exported, it helps in improving variable conversion TYPE(volgrid6d_var) :: convert INTEGER :: i #ifdef HAVE_LIBGRIBAPI INTEGER :: gaid, editionnumber, category, centre #endif IF (.NOT. ALLOCATED(conv_bwd)) CALL vg6d_v7d_var_conv_setup() #ifdef HAVE_LIBGRIBAPI editionnumber=255; category=255; centre=255 #endif IF (PRESENT(grid_id_template)) THEN #ifdef HAVE_LIBGRIBAPI gaid = grid_id_get_gaid(grid_id_template) IF (c_e(gaid)) THEN CALL grib_get(gaid, 'GRIBEditionNumber', editionnumber) IF (editionnumber == 1) THEN CALL grib_get(gaid,'gribTablesVersionNo',category) ENDIF CALL grib_get(gaid,'centre',centre) ENDIF #endif ENDIF DO i = 1, SIZE(conv_bwd) IF (varbufr == conv_bwd(i)%v7d_var) THEN #ifdef HAVE_LIBGRIBAPI IF (editionnumber /= 255) THEN ! further check required (gaid present) IF (editionnumber == 1) THEN IF (conv_bwd(i)%vg6d_var%discipline /= 255) cycle ! wrong edition ELSE IF (editionnumber == 2) THEN IF (conv_bwd(i)%vg6d_var%discipline == 255) cycle ! wrong edition ENDIF IF (conv_bwd(i)%vg6d_var%centre /= 255 .AND. & conv_bwd(i)%vg6d_var%centre /= centre) cycle ! wrong centre ENDIF #endif convert = conv_bwd(i)%vg6d_var IF (PRESENT(c_func)) c_func = conv_bwd(i)%c_func RETURN ENDIF ENDDO ! not found convert = volgrid6d_var_miss IF (PRESENT(c_func)) c_func = conv_func_miss ! if hint available use it as a fallback IF (any(varbufr%gribhint /= imiss)) THEN convert%centre = varbufr%gribhint(1) convert%category = varbufr%gribhint(2) convert%number = varbufr%gribhint(3) convert%discipline = varbufr%gribhint(4) ENDIF CALL l4f_log(l4f_warn, 'varbufr2vargrib: variable '// & trim(varbufr%btable)//" : "//trim(varbufr%description)//" : "//trim(varbufr%unit)// & ' not found in table') END FUNCTION varbufr2vargrib_convert !> Normalize a variable definition converting it to the !! format (grib edition) specified in the (grib) template provided. !! This allows a basic grib1 <-> grib2 conversion provided that !! entries for both grib editions of the related variable are present !! in the static file \a vargrib2ufr.csv. If the \a c_func variable !! returned is not missing (i.e. /= conv_func_miss) the field value !! should be converted as well using the conv_func::compute method . SUBROUTINE volgrid6d_var_normalize(this, c_func, grid_id_template) TYPE(volgrid6d_var),INTENT(inout) :: this !< variable to normalize TYPE(conv_func),INTENT(out) :: c_func !< \a conv_func object to convert data TYPE(grid_id),INTENT(in) :: grid_id_template !< a template (typically grib_api) to which data will be finally exported, it helps in improving variable conversion LOGICAL :: eqed, eqcentre INTEGER :: gaid, editionnumber, centre TYPE(volgrid6d_var) :: tmpgrib TYPE(vol7d_var) :: tmpbufr TYPE(conv_func) tmpc_func1, tmpc_func2 eqed = .true. eqcentre = .true. c_func = conv_func_miss #ifdef HAVE_LIBGRIBAPI gaid = grid_id_get_gaid(grid_id_template) IF (c_e(gaid)) THEN CALL grib_get(gaid, 'GRIBEditionNumber', editionnumber) CALL grib_get(gaid, 'centre', centre) eqed = editionnumber == 1 .EQV. this%discipline == 255 eqcentre = centre == this%centre ENDIF #endif IF (eqed .AND. eqcentre) RETURN ! nothing to do tmpbufr = convert(this, tmpc_func1) tmpgrib = convert(tmpbufr, tmpc_func2, grid_id_template) IF (tmpgrib /= volgrid6d_var_miss) THEN ! conversion back and forth successful, set also conversion function this = tmpgrib c_func = tmpc_func1 * tmpc_func2 ! set to missing in common case to avoid useless computation IF (c_func == conv_func_identity) c_func = conv_func_miss ELSE IF (.NOT.eqed) THEN ! conversion back and forth unsuccessful and grib edition incompatible, set to miss this = tmpgrib ENDIF END SUBROUTINE volgrid6d_var_normalize ! Private subroutine for reading forward and backward conversion tables ! todo: better error handling SUBROUTINE vg6d_v7d_var_conv_setup() INTEGER :: un, i, n, stallo ! forward, grib to bufr un = open_package_file('vargrib2bufr.csv', filetype_data) n=0 DO WHILE(.true.) READ(un,*,END=100) n = n + 1 ENDDO 100 CONTINUE rewind(un) ALLOCATE(conv_fwd(n),stat=stallo) IF (stallo /= 0) THEN CALL l4f_log(l4f_fatal,"allocating memory") CALL raise_fatal_error() ENDIF conv_fwd(:) = vg6d_v7d_var_conv_miss CALL import_var_conv(un, conv_fwd) CLOSE(un) ! backward, bufr to grib un = open_package_file('vargrib2bufr.csv', filetype_data) ! use the same file for now !un = open_package_file('varbufr2grib.csv', filetype_data) n=0 DO WHILE(.true.) READ(un,*,END=300) n = n + 1 ENDDO 300 CONTINUE rewind(un) ALLOCATE(conv_bwd(n),stat=stallo) IF (stallo /= 0) THEN CALL l4f_log(l4f_fatal,"allocating memory") CALL raise_fatal_error() end if conv_bwd(:) = vg6d_v7d_var_conv_miss CALL import_var_conv(un, conv_bwd) DO i = 1, n conv_bwd(i)%c_func%a = 1./conv_bwd(i)%c_func%a conv_bwd(i)%c_func%b = - conv_bwd(i)%c_func%b ENDDO CLOSE(un) CONTAINS SUBROUTINE import_var_conv(un, conv_type) INTEGER, INTENT(in) :: un TYPE(vg6d_v7d_var_conv), INTENT(out) :: conv_type(:) INTEGER :: i TYPE(csv_record) :: csv CHARACTER(len=1024) :: line CHARACTER(len=10) :: btable INTEGER :: centre, category, number, discipline DO i = 1, SIZE(conv_type) READ(un,'(A)',END=200)line CALL init(csv, line) CALL csv_record_getfield(csv, btable) CALL csv_record_getfield(csv) ! skip fields for description and unit, CALL csv_record_getfield(csv) ! they correspond to grib information, not bufr Btable CALL init(conv_type(i)%v7d_var, btable=btable) CALL csv_record_getfield(csv, centre) CALL csv_record_getfield(csv, category) CALL csv_record_getfield(csv, number) CALL csv_record_getfield(csv, discipline) CALL init(conv_type(i)%vg6d_var, centre=centre, category=category, & number=number, discipline=discipline) ! controllare l'ordine CALL csv_record_getfield(csv, conv_type(i)%c_func%a) CALL csv_record_getfield(csv, conv_type(i)%c_func%b) CALL delete(csv) ENDDO 200 CONTINUE END SUBROUTINE import_var_conv END SUBROUTINE vg6d_v7d_var_conv_setup ELEMENTAL FUNCTION conv_func_eq(this, that) RESULT(res) TYPE(conv_func),INTENT(IN) :: this, that LOGICAL :: res res = this%a == that%a .AND. this%b == that%b END FUNCTION conv_func_eq ELEMENTAL FUNCTION conv_func_ne(this, that) RESULT(res) TYPE(conv_func),INTENT(IN) :: this, that LOGICAL :: res res = .NOT.(this == that) END FUNCTION conv_func_ne FUNCTION conv_func_mult(this, that) RESULT(mult) TYPE(conv_func),INTENT(in) :: this TYPE(conv_func),INTENT(in) :: that TYPE(conv_func) :: mult IF (this == conv_func_miss .OR. that == conv_func_miss) THEN mult = conv_func_miss ELSE mult%a = this%a*that%a mult%b = this%a*that%b+this%b ENDIF END FUNCTION conv_func_mult !> Apply the conversion function \a this to \a values. !! The numerical conversion (only linear at the moment) defined by the !! \a conv_func object \a this is applied to the \a values argument; !! the converted result is stored in place; missing values remain !! missing; if the conversion function is undefined (\a !! conv_func_miss) the values are unchanged. The method is \c !! ELEMENTAL, thus \a values can be also an array of any shape. ELEMENTAL SUBROUTINE conv_func_compute(this, values) TYPE(conv_func),INTENT(in) :: this !< object defining the conversion function REAL,INTENT(inout) :: values !< value to be converted in place IF (this /= conv_func_miss) THEN IF (c_e(values)) values = values*this%a + this%b ELSE values=rmiss ENDIF END SUBROUTINE conv_func_compute !> Locate variables which are horizontal components of a vector field. !! This method scans the \a volgrid6d_var array provided and locates !! pairs of variables which are x and y component of the same vector !! field. On exit, the arrays \x xind(:) and \a yind(:) are allocated !! to a size equal to the number of vector fields detected and their !! corresponding elements will point to x and y components of the same !! vector field. If inconsistencies are found, e.g. only one component !! is detected for a field, or more than one input variable define !! the same component, then \a xind and \a xind are nullified, thus an !! error condition can be tested as \c .NOT.ASSOCIATED(xind). If no !! vector fields are found then \a xind and \a xind are allocated to !! zero size. If \a xind and \a yind are \c ASSOCIATED() after return, !! they should be \c DEALLOCATEd by the calling procedure. SUBROUTINE volgrid6d_var_hor_comp_index(this, xind, yind) TYPE(volgrid6d_var),INTENT(in) :: this(:) !< array of volgrid6d_var objects (grib variable) to test INTEGER,POINTER :: xind(:), yind(:) !< output arrays of indices pointing to matching horizontal components, allocated by this method TYPE(vol7d_var) :: varbufr(SIZE(this)) TYPE(conv_func),POINTER :: c_func(:) INTEGER :: i, nv, counts(SIZE(vol7d_var_horcomp)) NULLIFY(xind, yind) counts(:) = 0 CALL vargrib2varbufr(this, varbufr, c_func) DO i = 1, SIZE(vol7d_var_horcomp) counts(i) = count(varbufr(:) == vol7d_var_horcomp(i)) ENDDO IF (any(counts(1::2) > 1)) THEN CALL l4f_log(l4f_warn, '> 1 variable refer to x component of the same field, (un)rotation impossible') DEALLOCATE(c_func) RETURN ENDIF IF (any(counts(2::2) > 1)) THEN CALL l4f_log(l4f_warn, '> 1 variable refer to y component of the same field, (un)rotation impossible') DEALLOCATE(c_func) RETURN ENDIF ! check that variables are paired and count pairs nv = 0 DO i = 1, SIZE(vol7d_var_horcomp), 2 IF (counts(i) == 0 .AND. counts(i+1) > 0) THEN CALL l4f_log(l4f_warn, 'variable '//trim(vol7d_var_horcomp(i+1)%btable)// & ' present but the corresponding x-component '// & trim(vol7d_var_horcomp(i)%btable)//' is missing, (un)rotation impossible') RETURN ELSE IF (counts(i+1) == 0 .AND. counts(i) > 0) THEN CALL l4f_log(l4f_warn, 'variable '//trim(vol7d_var_horcomp(i)%btable)// & ' present but the corresponding y-component '// & trim(vol7d_var_horcomp(i+1)%btable)//' is missing, (un)rotation impossible') RETURN ENDIF IF (counts(i) == 1 .AND. counts(i+1) == 1) nv = nv + 1 ENDDO ! repeat the loop storing indices ALLOCATE(xind(nv), yind(nv)) nv = 0 DO i = 1, SIZE(vol7d_var_horcomp), 2 IF (counts(i) == 1 .AND. counts(i+1) == 1) THEN nv = nv + 1 xind(nv) = index(varbufr(:), vol7d_var_horcomp(i)) yind(nv) = index(varbufr(:), vol7d_var_horcomp(i+1)) ENDIF ENDDO DEALLOCATE(c_func) END SUBROUTINE volgrid6d_var_hor_comp_index !> Tests whether a variable is the horizontal component of a vector field. !! Returns \a .TRUE. if the corresponding variable is recognized as an !! horizontal component of a vector field; if it is the case the !! variable may need rotation in case of coordinate change. FUNCTION volgrid6d_var_is_hor_comp(this) RESULT(is_hor_comp) TYPE(volgrid6d_var),INTENT(in) :: this !< volgrid6d_var object (grib variable) to test LOGICAL :: is_hor_comp TYPE(vol7d_var) :: varbufr varbufr = convert(this) is_hor_comp = any(varbufr == vol7d_var_horcomp(:)) END FUNCTION volgrid6d_var_is_hor_comp ! before unstaggering?? !IF (.NOT. ALLOCATED(conv_fwd)) CALL vg6d_v7d_var_conv_setup() ! !call init(varu,btable="B11003") !call init(varv,btable="B11004") ! ! test about presence of u and v in standard table !if ( index(conv_fwd(:)%v7d_var,varu) == 0 .or. index(conv_fwd(:)%v7d_var,varv) == 0 )then ! call l4f_category_log(this%category,L4F_FATAL, & ! "variables B11003 and/or B11004 (wind components) not defined by vg6d_v7d_var_conv_setup") ! CALL raise_error() ! RETURN !end if ! !if (associated(this%var))then ! nvar=size(this%var) ! allocate(varbufr(nvar),stat=stallo) ! if (stallo /=0)then ! call l4f_log(L4F_FATAL,"allocating memory") ! call raise_fatal_error("allocating memory") ! end if ! ! CALL vargrib2varbufr(this%var, varbufr) !ELSE ! CALL l4f_category_log(this%category, L4F_ERROR, & ! "trying to destagger an incomplete volgrid6d object") ! CALL raise_error() ! RETURN !end if ! !nvaru=COUNT(varbufr==varu) !nvarv=COUNT(varbufr==varv) ! !if (nvaru > 1 )then ! call l4f_category_log(this%category,L4F_WARN, & ! ">1 variables refer to u wind component, destaggering will not be done ") ! DEALLOCATE(varbufr) ! RETURN !endif ! !if (nvarv > 1 )then ! call l4f_category_log(this%category,L4F_WARN, & ! ">1 variables refer to v wind component, destaggering will not be done ") ! DEALLOCATE(varbufr) ! RETURN !endif ! !if (nvaru == 0 .and. nvarv == 0) then ! call l4f_category_log(this%category,L4F_WARN, & ! "no u or v wind component found in volume, nothing to do") ! DEALLOCATE(varbufr) ! RETURN !endif ! !if (COUNT(varbufr/=varu .and. varbufr/=varv) > 0) then ! call l4f_category_log(this%category,L4F_WARN, & ! "there are variables different from u and v wind component in C grid") !endif END MODULE volgrid6d_var_class