USER'S GUIDE TO ADD NEW SURFACE VARIABLES OR NEW
Contents
1. NG long version of the name NGRUP and that there will be one dimension INGRUP to increment x List of routines to update For the time being we assume that we do not add any new attribute There is a minimal list of routines which should be updated e module surface_fields_mix F90 e setup su _surf_fids F90 e in some cases phys dmn mf_phys F90 x The first questions to answer e The first question to answer is to know if this group contains prognostic variables which must enter SP or diagnostic variables which must enter SD If it must enter SD does this group contain constants diagnostic fields or pseudo prognostic fields e Does this group contain 3D fields several layers or 2D fields one layer x module surface_fields_mix F90 Now we assume that we have answered the previous questions e The new type TYPE_SFL_ NGRUP must be defined paragraph Group specific type definitions for example the new variable is assumed here to be a 2D one entering SD Group NG NGRUP provide comments TYPE TYPE_SFL_ NGRUP TYPE TYPE_SURF_MTL_2D POINTER Y NG END TYPE TYPE_SFL_ NGRUP e In the part Data structures Prognostic multi time level fields if SP or one time level fields if SD add NGRUP REAL KIND JPRB ALLOCATABLE SD_ NG TYPE TYPE_SURF_GEN YSD_ NG D TYPE TYPE_SFL_ NGRUP YSD_ NG e Ifthe new group enters SP routines GPPOPER and GPO2. D_ NG as dummy arguments or using module SURFACE_FIELDS_MIX e more generally all routines involved in FULL POS requiring PSP_ NG or PSD_ NG as dummy arguments or using module SURFACE FIELDS MIX For ordering of the GFP if possible keep the same order in YOMAFN NAMAFN SUAFN1 SUAFN2 SUAFNS More generally look at all routines using YOMAF N to see if some physical surface post processable variables are used and if the new post processable variable requires code updating 5 Add a new surface variable additional features when assimilating it There are several parts of the code to be checked updates may be required at least e the GOMS dataflow e the observation interpolator e canari caprsurf F90 x List of routines to update in the GOMS dataflow There is a minimal list of routines which should be updated e module goms_mod F90 e obs_preproc sugoms F90 e a subset of the routines using the following variables of module gom_mod F90 GID NG _ newvar pointer for array GOM must be checked and in particular those where individual surface variables are mentioned mostly in directories op_obs and pp _obs e op_obs cobs F90 Only a small subset of surface fields are present in the GOMS dataflow For example e in module goms_mod F90 update definition of types TYPE GOM_ID and TYPE GVAR update calculation of variable GID e in obs_preproc sugoms F 90 set up GOM LINTERP GID NG _ newvar and GOM INTERP_TYP
3. E GID NG _ newvar e in op_obs cobs F90 call GOM_ MODEL SET for new surface variable using pointer GID NG _ newvar x Updating TL and AD codes TL and AD codes must be updated for example op_obs cobsad F90 6 Add a new surface variable additional features when applying nudging on it Nudging is used for climatic simulations This topic is not described in detail for the time being it has to be written by people having a good knowledge into nudging Some of the variables XVU of YOMNUD contain reference values for the nudging of surface prognostic variables surface moisture for example When adding a new variable which can be nudged the question of updating the following routines must be answered e climate updsst F90 and its call in adiab cpg_gp F 90 e dia cpnudg F90 and its call in phys dmn mf_phys F 90 A new XVU must be added in YOMNUD and all routines using the XVU must be updated expect modifications in the following routines e dia cpnudg F90 e climate updnud F90 e climate updsst F90 e module yomnud F90 7 Adda new surface attribute We assume that we want to add a new attribute to the following data structures TYPE SURF_MTL_2D resp TYPE_SURF_MTL _3D the generic name of this attribute is NEWATTR x List of routines to update There is a minimal list of routines which should be updated e module surface_fields_mix F90 e setup su _surf_fids F90 x module surface_fields_mix F90 and setup su_surf_f
4. EGE file In the last case see the previous paragraph x Additional remarks New INCLI routines must be called by INCLIO For each new variable it must be questioned if there is only an annual climatological value or if there are different monthly climatological values 9 References e IDGOMU Geer A 2013 GOM user guide interpolation to observation space Internal note 3pp
5. NG D YSD_ NG Y NEWVAR amp amp some optional arguments ENDIF ENDIF set up information that couldn t be set until the total number of NGRUP fields was known CALL FINALISE_SFLP2 YSD_ NG D YSD_ NG Y NG Remarks e Call to INISFLP2 argument after INGRUP is FALSE for diagnostic fields TRUE for prognostic fields e About the optional arguments of SETUP SFLP2 on can specify for example KGRIB GRIB code CDNAME field name for the ARPEGE files KTRAJ KREQIN KREQIN 1 means that the field must be read on a file e For a 3D variable SETUP_SFLP3 must be called instead of SETUP_SFLP2 e Sometimes no field is allocated for LECMWE resp for NOT LECMWE in this case in paragraph set up attributes for active fields we may find empty code under LECMWE resp under NOT LECMWE phys_dmn mf_phys F90 If the new variable is a pseudo historic SD field the same thing as for a new variable in a already existing class must be done x other routines Other routines may have to be modified in particular those requiring dummy arguments YRSURF PSP_ NG or PSD_ NG 2 3 New group creation without specification of individual variables We now consider creation of a new group which may contain several variables and where it is not necessary to give an individual name to each variable This is the case for example for group SFO SFORC We assume that the new group will have a generic code name
6. PER should be updated if the new group enters SD routine GPOPER should be updated GPOPER_2 must be called for 2D fields GPOPER 3 for 3D fields e The new group must be added in most of encapsulated routines the name of which ends by SURF setup su_surf_flds F 90 In SU_LSURF_FLDS we add the corresponding code for a new group in section 1 if this is a prognostic variable in section 2 if this is a diagnostic variable That provides a piece of code which is divided into 3 parts and looks like the new variable is assumed here to be a 2D one entering SD Group NG NGRUP compute number of fields allocated in this group INGRUP 0 Modify value of INGRUP according to some conditions call to INI_SFLP2 allocation IF LL_FIRST_CALL NULLIFY YSD_ NG Y NG IF ASSOCIATED YSD_ NG Y NG DEALLOCATE YSD_ NG Y NG ALLOCATE YSD_ NG 4Y NG JPMAXSFLDS CALL INI_SFLP2 YSD_ NG D YSD_ NG Y NG INGRUP FALSE NGRUP SD_ NG gt set up attributes for active fields IF LECMWF THEN DO JFLD 1 YSD_ NG D NUMFLDS CALL SETUP_SFLP2 YSD_ NG D YSD_ NG Y NG JFLD amp amp some optional arguments ENDDO ELSE DO JFLD 1 YSD_ NG D NUMFLDS CALL SETUP_SFLP2 YSD_ NG D YSD_ NG Y NG JFLD amp amp some optional arguments ENDDO ENDIF set up information that couldn t be set until the total number of NGRUP fields was known CALL FINALISE_SFLP2 YSD_ NG D YSD_ NG
7. USER S GUIDE TO ADD NEW SURFACE VARIABLES OR NEW SURFACE ATTRIBUTES IN ARPEGE IFS ALADIN AROME CYCLE 42 YESSAD K METEO FRANCE CNRM GMAP ALGO July 7 2015 Contents 1 Introduction 2 Add a new surface variable main features 2 1 Nonew group creation 224 6 20 be aM dae eee hae bee Paced be BL hae Pe 2 2 New group creation with specification of individual variables 0 2 3 New group creation without specification of individual variables 3 Add a new surface variable or group additional features when read write this surface variable on ARPEGE files 4 Add a new surface variable additional features when post processing it 5 Add a new surface variable additional features when assimilating it 6 Add a new surface variable additional features when applying nudging on it 7 Add a new surface attribute 8 Add a new surface field or group in climatological files configuration 923 E923 METEO FRANCE purpose only 9 References 5 Or Ww 1 Introduction The new surface structure has been introduced by ECMWF Mats Hamrud in CY31R2 and now the surface buffers are split into a SP one prognostic surface variables and a SD one diagnostic surface variables The word surface can be sometimes misleading because we find not only surface or internal soil quantities but also upper air quantities at a specified level temperature at 10 meters height of the top of t
8. Y NG Remarks cf remarks given in part 2 3 phys_dmn mf_phys F90 If the new group contains pseudo historic SD field the same thing as for a new variable in a already existing class must be done x other routines Other routines may have to be modified in particular those requiring dummy arguments PSP_ NG or PSD_ NG 3 Adda new surface variable or group additional features when read write this surface variable on ARPEGE files This topic will not be described in detail Additionally some of the involved routines are currently under deep rewriting It is recommended to check the occurrences of use of YRSURF in the sugrid or iogrid named routines and their callees 4 Add a new surface variable additional features when post processing it Modifications will touch at least the following pieces of code x module yomafn F90 and namelist namafn nam h Variable GFP_ NEWVAR must be added it is desirable that the name into NEWVAR will be identical to the one used in the surface variables x module parfpos F90 If necessary value of JPOSSGP must be increased x other routines Adding a new variable GFP_ NEWVAR may generate modifications in routines using GFP in particular e module type_fpdsphys F90 e setup suafn1 F90 setup suafn2 F90 and setup suafn3 F90 e setup sucfu F90 and setup suxfu F90 e some arpifs fullpos and aladin fullpos routines check those using GFP_ requiring YDSURF PSP_ NG or PS
9. YPE_SFL_ NGRUP YSD_ NG Definition of type TYPE_SURF must be updated If the new group enters SP routines GPPOPER and GPOPER should be updated if the new group enters SD routine GPOPER should be updated GPOPER_2 must be called for 2D fields GPOPER_3 for 3D fields The new group must be added in most of encapsulated routines the name of which ends by SURF setup su_surf_flds F 90 In SU_SURF_FLDS we add the corresponding code for a new group in section 1 if this is a prognostic variable in section 2 if this is a diagnostic variable That provides a piece of code which is divided into 4 parts and looks like the new variable is assumed here to be a 2D one entering SD Group NG NGRUP compute number of fields allocated in this group INGRUP call to INI_SFLP2 allocation IF LL_FIRST_CALL NULLIFY YSD_ NG Y NG IF ASSOCIATED YSD_ NG Y NG DEALLOCATE YSD_ NG Y NG ALLOCATE YSD_ NG 4Y NG JPMAXSFLDS CALL INI_SFLP2 YSD_ NG D YSD_ NG Y NG INGRUP FALSE NGRUP SD_ NG gt YSD_ NG Y NEWVAR gt YSD_ NG Y NG JPMAXSFLDS set up attributes for active fields IF LECMWF THEN IF test THEN YSD_ NG Y NEWVAR gt YSD_ NG Y NG YSD_ NG DZIPTR CALL SETUP_SFLP2 YSD_ NG D YSD_ NG Y NEWVAR amp amp some optional arguments ENDIF ELSE IF test THEN YSD_ NG Y NEWVAR gt YSD_ NG Y NG YSD_ NG DZIPTR CALL SETUP_SFLP2 YSD_
10. he PBL total cloudiness SP buffers are split into several groups SD buffers also They replace the old buffer GPPBUF which has been removed The total number of fields reaches a total close to 200 and new fields are regularly added Adding new surface fields is not so easy that one could believe even if the number of routines to be modified is rather small and the purpose of the present paper is to provide a user s guide for people wanting to introduce new surface variables or new surface attributes This is not a comprehensive documentation of all the surface features Some other internal notes describe the new surface data flow Three types of SD surface variables can be found independently of the different groups e The constants for example the land sea mask which do not evolve and which are always input data in the physics e The diagnostics for example the temperature at 10 meters which are always output data in the physics they are not conserved from one timestep to the following one e The pseudo prognostic fields like the Charnock constant which are input of the physics modified by the physics and the new value is conserved for the input of physics at the following timestep Additionally to that some of them are read on a file some of them can be post processed some of them can be assimilated Note that there are some other groups of surface fields which have not been put in the SP a
11. ic code name NEWVAR that the new group will have a generic code name NG long version of the name NGRUP and that there will be one dimension INGRUP to increment x List of routines to update For the time being we assume that we do not add any new attribute There is a minimal list of routines which should be updated e module surface_fields_mix F90 e setup su _surf_fids F90 e in some cases phys dmn mf_phys F90 x The first questions to answer The first question to answer is to know if this is a prognostic variable which must enter SP or a diagnostic one which must enter SD If it must enter SD is it a constant a diagnostic or a pseudo prognostic field Is it a 3D field several layers or a 2D field one layer x module surface_fields_mix F90 Now we assume that we have answered the previous questions The new type TYPE_SFL_ NGRUP must be defined paragraph Group specific type definitions for example the new variable is assumed here to be a 2D one entering SD Group NG NGRUP provide comments TYPE TYPE_SFL_ NGRUP TYPE TYPE_SURF_MTL_2D POINTER Y NEWVAR provide comments TYPE TYPE_SURF_MTL_2D POINTER Y NG END TYPE TYPE_SFL_ NGRUP In the part Data structures Prognostic multi time level fields if SP or one time level fields if SD add NGRUP REAL KIND JPRB ALLOCATABLE SD_ NG TYPE TYPE_SURF_GEN YSD_ NG D TYPE T
12. ids F90 e New attribute NEWATTR must be added in the definition of TYPE SURF_MTL _2D resp TYPE_SURF_MTL _3D e Update INI_SFLP2 and SETUP_SFLP2 for 2D variables resp INI SFLP3 and SETUP_SFLP3 for 3D variables e In some cases this new attribute may become a new optional dummy argument of SETUP_SFLP2 2D or SETUP_SFLP3 3D Update the calls to SETUP_SFLP2 and SETUP_SFLP3 when required in SU SURF FLDS 8 Add a new surface field or group in climatological files configuration 923 E923 METEO FRANCE purpose only To simplify we assume that the new variable will have a generic code name NEWVAR If a new group must be created for that the new group will have a generic code name NG long version of the name NGRUP x List of routines to update if the new variable enters an existing group Once known the group where the new variable must enter the proper INCLI routine INCLI2 to INCLI10 must be updated Reading of the new surface field on a database interpolating it and writing it on an ARPEGE file must be added in the proper INCLI x List of routines to update if the new variable enters a new group It must be questioned if a new INCLI routine must be created or if the code relative to the new variable can enter an existing INCLI In the first case a new INCLI routine must be written with the same structure as in INCLI2 to INCLI10 reading of the new surface field on a database interpolating it and writing it on an ARP
13. must enter SD If it must enter SD is it a constant a diagnostic or a pseudo prognostic field e The second question is to determine in what group it must enter For example if this is a climatological constant or a geometric constant it must enter SD in the group VARSF e Is it a 3D field several layers or a 2D field one layer x module surface_fields_mix F90 Now we assume that we have answered the previous questions In SURFACE_FIELDS_MIX we go in the topics Group specific type definitions and e We add the attribute Y NEWVAR in the adequate type definition TYPE_SFL_ group The attribute is itself a derivated type variable of type TYPE_LSURF_MTL_2D or TYPE_SURF_MTL_S3D according to the fact that this is a 2D variable or a 3D variable A comment should always be provided at this place to know what is the variable NEWVAR setup su_surf_flds F90 Go in the right group For each group the code is divided into 4 parts e A part which counts the number of useful and allocated fields e A part which does some allocations and memory equivalences and calls INILSFLP3 or INI_SFLP2 e A part which set ups attributes for useful allocated fields separate LECMWF and NOT LECMWF codes This part calls SETUP_SFLP3 or SETUP_SFLP2 This is the only part where some complicated tests are allowed e A part calling FINALISE_SFLP3 or FINALISE_SFLP2 In particular these routines set up attributes YSD_ group _Y variable _LSET or YSP_ g
14. nd SD buffers for example some AROME surface fields which are still stored in GPARBUF some VCLIU VRADF VTILE fields which are stored in specific buffers Such quantities are out of the scope of this paper We first start to list the common features to be added when introducing surface fields read or not on a file post processed or not assimilated or not Some details will be then given to describe e how to read write new surface fields on files e how to post process new surface fields e how to assimilate new surface fields e how to apply nudging on new surface fields e how to add new attributes e how to add a new group in SP or SD 2 Add anew surface variable main features 2 1 No new group creation For the time being the new variable is assumed to enter an existing group To simplify we assume that the new variable will have a generic code name NEWVAR x List of routines to update For the time being we assume that we do not add any new attribute There is a minimal list of routines which should be updated e module surface_fields_mix F90 e setup su_surf_fids F90 e in some cases adiab cpg F90 e in some cases adiab cpwts F90 e in some cases module traj_surface_mod F90 e in some cases module gom_mod F90 e in some cases phys dmn mf_phys F90 x The first questions to answer e The first question to answer is to know if this is a prognostic variable which must enter SP or a diagnostic one which
15. roup Y variable _LSET saying if surface fields are active or not Individual fields must appear in the same order in these 4 parts When a new field is introduced e In part 1 number of allocated fields must be properly updated e In part 2 a line must be added YS P or D _ group Y NEWVAR gt YS P or D _ group Y group JPMAXSFLDS e In part 3 two statements must be added example given with SETUP_SFLP2 but that must be SETUP_SFLP3 for some groups IF test THEN YS P or D _ group Y NEWVAR gt YS P or D _ group Y group YS P or D _ group D IPTR CALL SETUP_SFLP2 YS P or D _ group D YS P or D _ group Y NEWVAR optional arguments ENDIF The optional arguments to be given vary according to the field for fields read on files the GRIB name or the ARPEGE file identifier appears if KREQIN 1 that means that the field is read on a file x phys_dmn mf_phys F90 Some specific cautions should be taken for pseudo historic SD fields for purely diagnostic calls of MF_PHYS this is the case for LDCONFX sort of timestep zero in the XFU diagnostics For such calls of MF_PHYS the initial content of surface variables should be saved in a local array and restored at the end of MF_PHYS I do not know if this question is relevant also for EC_PHYS 2 2 New group creation with specification of individual variables The new variable is now assumed to enter a new group To simplify we assume that the new variable will have a gener
Download Pdf Manuals
Related Search