USER'S GUIDE TO ADD NEW GFL VARIABLES OR NEW GFL
Contents
1. All the checkings which can be done in SUCTRL GFLATTR should go in SUCTRL_GFLATTR adiab cptend F90 adiab cptend_new F90 adiab cptendsm F90 adiab cptend_flex F 90 Introduction of the new GFL if they are true prognostic ones and if they have a diabatic tendency MF physical packages adiab gprcp F90 There are still calls to GPRCP which require individual GFL variables These calls to GPRCP may have to be modified for prognostic GFL which enter the definition of moist air constant x phys_dmn cpchet F90 cf adiab cptend F90 phys_dmn mf_phys F90 The following updates may be required according to the type of physics called e call to INIT_APLPAR APLPAR APLPAR_AROME INITAPLPAR WRITEPHYSIO PROFILECHET take account of the new GFL variable if required compute the diabatic flux for the new GFL variable if required if relevant APLPAR APLPAR_AROME and some of their callees must be updated too e call to CPTEND CPTEND_NEW CPCHET take account of the new GFL variable if required true prognostic GFL with a diabatic tendency e in MF_PHYS pseudo prognostic GFL updated by a physical flux at each time step are currently not treated like true prognostic GFL compute the tendency inside CPTEND or CPTEND_NEW and update GFLT1 in CPUTQY GFLT1 is updated in part 2 9 of MF_PHYS For a new pseudo prognostic GFL containing a time dependent physical flux an update must be added in part 2 9 e in some cases additional operations may2. add variable YFA VGFL Such variables are not used for groups of GFL where the GFL attribute CNAME is directly used x namelist namfa nam h For the new individual GFL VGFL add variable YFA VGFL x setup sufa F90 For the new individual GFL VGFL variable YFA VGFL must be setup in part 1 and printed at the end of part 3 5 Add a new GFL variable additional features when post processing of GFL This is not compulsory to make the new GFL post processable but if desired there is a minimal list of routines which should be updated module parfpos F90 module yomafn F90 namelist namafn nam h setup suafn1 F90 setup suafn2 F90 setup suafn3 F90 fullpos endpos F90 fullpos endpos_prepgfl F90 fullpos fpcordyn F90 pp obs pos F90 pp obs pos_prepgfl F90 fullpos vpos_prep F90 and in some cases fullpos phymfpos F90 x module yomafn F90 and module parfpos F90 Type FULLPOS_TYPE One finds variables TFP_ VGFL for some individual GFL variables for example TFP_SN for snow TFP_O3MX for ozone That means that a new TFP_ VGFL must be added For the new group of GFL GRUP a new variable TFP_ GRUP JPOS GRUP must be added Parameter variable JPOS GRUP must be added in parfpos F 90 Variable JPOS3DF should be updated in parfpos F90 When there are new post processable fields in an already existing group JPOS GRUP should be updated in parfpos F90 x namelist namafn nam h One finds variables TFP_ VGFL for some individual GFL
3. be required to fill arrays ZTENDGFL or ZTENDGFLR x module yomphyder F90 phys_ec ec_phys F90 and phys _ec callpar F90 and probably some routines under callpar F90 Introduction of the new GFL if they are true prognostic ones and if they have a diabatic tendency for the ECMWF physics adiab cpg_pt F90 Introduction of the new GFL if they are true prognostic ones if they have a diabatic tendency at least for one of the physical packages and if they are supposed to be used with a predictor corrector scheme attribute LPT T in this case in this routine currently GFL is treated globally but the diabatic tendencies are treated individually for each GFL Note that this routine is now used only for the interface with ECMWF physics adiab cpg_pt_ulp F90 Modern version of cpg_pt F90 used for the interface of physics called under MF_PHYS This routine uses GFL arrays so there is nothing to do when adding new GFL variables adiab cpg F90 This routine now passes GFL arrays to its callees so there is nothing to do when adding new GFL variables module yomnsv F90 to be modified only if the list of AROME chemicals currently put in the GFL EXT is modified 4 Add a new GFL variable additional features when read write GFL on ARPEGE files There is a minimal list of routines which should be updated e module yomfa F90 e namelist namfa nam h e setup sufa F90 x module yomfa F 90 Type FAD for the new individual GFL VGFL
4. for the new GFL introduced in particular when the spectral or grid point should be present for some options for example when some physics packages are used e Part 5 add a new item checking that N GRUP is not greater than JP GRUP if a new group of GFL is introduced When relevant additional tests on N GRUP should be coded in this part x setup sugfl2 F90 There are three parts 1 1 1 2 and 1 3 which must be updated Order of the GFL SHOULD be the same in these three parts but the order can be different of the one of part 1 of SUGFL3 e For purely grid point GFL part 1 3 must be updated by a call to ABOR1 if the attribute LSP T This ABOR1 SHOULD be in part 1 3 never in part 1 2 e For some groups of GFL it may exist a constraint like the GFL of the group are either all grid point ones or all spectral ones if a new group of GFL has this constraint a test must be done in part 1 2 to check if when there is at least one of them which is purely a grid point one all of them are grid point ones The corresponding test and CALL ABORI should be done in part 1 2 x setup sugfl3 F 90 e Part 1 Call to SUCSLINT calculation of LLPT LLPC and call of SET GFL_ATTR must be coded for the new GFL added x setup sudyn F 90 e In part 3 15 the only GFL checking currently remaining is about the moisture convergence Only the checkings requiring YOMDYN NAMDYN variables are allowed to be done in SUDYN and they should be in part 3 of SUDYN
5. number of GFL variables in a sub group example JPTRAC TYPE_GFL_COMP typed variables YGFL Y X for each variable X and YGFL YCOMP TYPE_GFL_NAML typed variables read in NAMGFL YGFL Y X _NL for each variable X TYPE_GFLD typed variable YGFL type gflflds F90 additional type definitions used currently in FULL POS yomegfl F90 contains buffers for example GFL GFLT1 TGFL type definition allows to gather all variables declared in yomgfl F90 in one global structure several encapsulated subroutines like ZERO YOMGFL which do operations on GFL buffers yomgfl5 F90 trajectory version of YOMGFL contains buffers for example GFL5 GFL_ DEPART TGFL5 type definition allows to gather all variables declared in yomgfl5 F90 in one global structure e gfl_subs_mod F90 contains additional declarations complementary to yom_ygfl F90 ones several encapsulated setup routines to fill attributes of YGFL Y X NL YGFL Y X YGFL YCOMP allocation and deallocations routines 2 All the set up is now gathered under SUOYOMA gt SUGFL SUGFL calls SUGFL1 SUGFL2 and SUGFL3 SUGFL1 gt default value for some dimensions declared in yom_ygfl F 90 SUDEFO_GFLATTR default values for YGFL Y X _NL attributes often depend on LECMWF or the physics package used reads NAMGFL SUCTRL_GFLATTR does checkings and sometimes resettings of YGFL Y X _NL attributes r
6. routines to update in the GOMS dataflow There is a minimal list of routines which should be updated e module gom_mod F90 references to individual GFL variables appear in several type definitions e obs_preproc sugoms F90 e a subset of the routines using the following variables of module gom_mod F90 GID gfl pointer for array GOM must be checked and in particular those where individual GFL variables are mentioned mostly in directories op_obs and pp obs x List of routines to update in the observation horizontal interpolator Only cobs F90 and cobsad F90 may have to be updated references to individual GFL variables x Updating TL and AD codes Work done on the direct code must be also done for TL and AD codes 7 Adda new GEL variable additional features when applying horizontal diffusion to GFL There is currently no GFL attribute to know if a GFL variable is diffused or not spectral diffusion The choice of diffusing or not a GFL variable is hard coded and can be done only on spectral GFL For the time being specific humidity and ozone are diffused if the attribute LSP is TRUE there are specific keys in NAMDYN to remove diffusion on these variables see documentation IDDH and the other GFL variables are NEVER diffused even if spectral If a new diffusible spectral GFL is introduced the following routines must be updated e module yomdyn F90 e namelist namdyn nam h e setup sudyn F90 e setup suhdir F90 e setup suhdf F
7. 0 The new attribute must be added in FALSIFY_GFLC module gfl_subs_mod F90 If the new attribute should be reset to F when there is no advection it must be added in NOADVECT _GFLC module gfl_subs_mod F90 The new attribute must be added in COPY_GFL COMP module gfl_subs mod F90 x Add a new attribute in TYPE_GFLD This attribute must be added in the definition of TYPE_ GFLD in module yom_ygfl F90 Calculation of this new attribute must be added in DEFINE_ GFL COMP module gfl_subs mod F90 In some specific cases the value of this new attribute may be modified in some other routines of module gfl_subs mod F90 SET_GFL_ATTR PRINT_GFL if the value of this attribute should be printed DEACT_CLOUD GFL and REACT_CLOUD GPL if the value of this attribute should be modified when deallocating reallocating cloud GFL variables x Remarks Future rewriting is expected in order 11 to avoid redundant attributes and to allow simplification of some set up routines to put in NAMGFL only attributes which have to be modified via namelist normally attributes like LT9 LT1 must not be in NAMGFL to allow a better separation between attributes which must pass in DEFINE_GFL_COMP and those which must pass in SET_GFL_ATTR References IDGOMU Geer A 2013 GOM user guide interpolation to observation space Internal note 3pp IDGFL Tudor M 2004 A short description of physics dynamics interface in the new data flow Internal
8. 90 setup suhdf_ec F90 and aladin setup suehdf F90 8 Adda new GFL variable additional features when applying nudging to GFL Nudging is used for climatic simulations There is currently no GFL attribute to know if a GFL variable can be nudged or not The choice of nudging or not a GFL variable is hard coded and can be done only on humidity For the time being only specific humidity is nudged in spectral space if the attribute LSP is TRUE and there is some code to nudge humidity in grid point space too Other GFL variables are NEVER nudged If a new nudgeable spectral GFL is introduced with nudging in spectral space the following routines must be updated e module yomnud F90 e namelist namnud nam h e setup sunud F90 e adiab spchor F90 If grid point nudging must be applied to a new GFL one should look at all routines using variables LNUDQG and XNUDQG 9 Addanew GEL variable additional features for LAM models coupling and spectral nudging x Coupling Coupling is controlled by attribute NCOUPLING Coupling is automatically updated for new prognostic GFL variables simply specify the right default for attribute NCOUPLING no modification is expected in pieces of code called in coupling x Spectral nudging There is currently no GFL attribute to know if a GFL variable can be nudged or not The choice of nudging or not a GFL variable is hard coded and can be done only on humidity For the time being only specific humidit
9. RIB codes for GFL some additional consistency checks can be added when required case NOT LARPEGEF e Part 3 reset the GFL attributes LSP and LGP for configurations where there is no GFL at all reset the attributes LGP and LSP for the new GFL introduced e Part 4 1 check for all GFL that LGP and LSP are not both TRUE do the checking for the new GFL introduced e Part 4 2 check for all grid point GFL required in the AROME physics that when the AROME physics is switched on the attribute LGP is set to T for all these GFL if a new grid point GFL required in the AROME physics is added the test should be added for this new GFL e Part 4 3 some GFL should be present when the ALARO prognostic convection scheme is switched on if a new GFL required in the ALARO prognostic convection scheme is added the test should be added for this new GFL e Part 4 3b some convective precipitation GFL should be present when option LGPCMT is T if a new GFL required for option LGPCMT is added the test must be updated for this new GFL e Part 4 4 check that attribute LTDIABLIN is T only for advected GFL LTDIABLIN T has a sense only in a semi Lagrangian job e Part 4 5 check that attribute LHORTURB is T only for advected GFL and when L3DTURB T LHORTURBS T has a sense only in a semi Lagrangian job e Part 4 6 check that attribute LINTLIN is T only for advected GFL LINTLIN T has a sense only in a semi Lagrangian job e Part 4 8 add additional checkings
10. USER S GUIDE TO ADD NEW GFL VARIABLES OR NEW GFL ATTRIBUTES IN ARPEGE IFS ALADIN AROME CYCLE 41 YESSAD K METEO FRANCE CNRM GMAP ALGO August 6 2014 Contents 1 Introduction 2 2 Basics about GFL dataflow 2 3 Add anew GFL variable main features 3 4 Add anew GFL variable additional features when read write GFL on ARPEGE files 6 5 Add anew GFL variable additional features when post processing of GFL 7 6 Add anew GFL variable additional features when assimilating GFL 8 7 Add anew GFL variable additional features when applying horizontal diffusion to GFL 9 8 Add anew GFL variable additional features when applying nudging to GFL 9 9 Add a new GFL variable additional features for LAM models coupling and spectral nudging 9 10 Add a new GFL attribute 10 11 References 11 1 Introduction The GFL structure has been introduced since several years for conservative prognostic variables and also pseudo historical variables and more and more GFL variables have been recently introduced Some recent contributions show that people involved in the introduction of new GFL variables or attributes have not always a comprehensive knowledge of the GFL data flow the consequence being misplaced pieces of code or forgotten pieces of code able to generate hidden bugs This is due to the fact that some parts of the code deal with GFL individually and not in a global way Adding GFL is not so easy that one could believe and the
11. X ICOR for input of FPHOR12 The same thing must probably be done for groups of GFL but no example is currently provided in FPCORDYN to know how to code it pp_obs pos F90 e Part 1 1 1 for new individual GFL VGFL add line CALL SUPTRPPGFL_POS 1 IPPGFL YLPPIC I VGFL e Additional pieces of code may be required for GFL in part 1 2 2 for example if horizontal derivatives are needed pp_obs pos_prepgfl F 90 e Part 1 define LD GFL YDPPIC I VGFL for new individual GFL VGFL e Part 2 define LDAP_GFL YDPPIC I VGFL for new individual GFL VGFL e Part 3 define KGFLCOD YDPPIC I VGFL for new individual GFL VGFL e Part 4 define KDIN_ GFL YDPPIC I VGFL for new individual GFL VGFL e Part 5 define LDIN GFL YDPPIC I VGFL for new individual GFL VGFL x fullpos vpos_prep F 90 e Part 1 define YDIN I VGFL for new individual GFL VGFL and also derivatives YDIN I VGFL L and YDIN I VGFL M if relevant e Part 2 define YDGFL LL VGFL for new individual GFL VGFL and also derivatives YDGFL LL VGFL L and YDGFL LL VGFLIM if relevant x Additional modifications Additional modifications may be required for example if new GFL variables become new input dummy arguments to some GP routines example of routines to check PHYMFPOS 6 Add a new GFL variable additional features when assimilating GFL There are several aspects to be considered at least the GOMS dataflow and the observation interpolator x List of
12. at three locations declaration of Y VGFL _SAVE in routine DEACT CLOUD_GFL and in routine REACT_CLOUD _GFL module type_gfiflds F 90 New GFL variables must be added in the type definitions present in this module x module yom_yegfl F90 e Value of JPNAMED _GFL must be updated this is the number of individual GFL e If an existing group generic name GRUP has to include some new GFL variables JP GRUP must be updated e If a new group generic name GRUP has to be created the new variable JP GRUP must be created e Type TYPE_GFLD For a new individual GFL generic name VGFL TYPE GFL COMP typed attribute Y VGFL must be added For a new group of GFL generic name GRUP TYPE GFL_COMP typed attribute Y GRUP must be added For a new group of GFL generic name GRUP variable N GRUP or NGFL_ GRUP must be added Sometimes a new logical variable must be declared For a new individual GFL generic name VGFL TYPE_GFL_NAML typed attribute Y VGFL _NL must be added For a new group of GFL generic name GRUP TYPE_GFL_NAML typed attribute Y GRUP _NL JP GRUP must be added Dimension of YEXT_NL when a new group GRUP is added JP GRUP should be subtracted to the original dimension x namelist namgfl nam h e For a new individual GFL generic name VGFL variable Y VGFL _NL must be added e For a new group of GFL generic name GRUP variable Y GRUP NL must be added x setup sugfl1 F 90 e Par
13. dded preferably in routine DEFINE GFL_COMP input dummy argument and internal code to update YDGFLC YGFL and YPTRC or and in routine SET GFL_ATTR input dummy argument and internal code to update YDGFLC and YGFL if necessary If the dummy arguments of DEFINE GFL_COMP are modified SUGFL2 must be updated If the dummy arguments of SET_GFL_ATTR are modified SUGFL3 must be updated This attribute SHOULD get a default value in parts la and 1b of SUDEFO_GFLATTR for all GFL variables Additional checkings can be added if required in SUCTRL_GFLATTR after reading NAMGFL Additional checkings can be added if required in part 3 of SUDYN after reading NAMDYN if these checkings use NAMDYN variables but in this case it would be desirable to move the involved NAMDYN YOMDYN variables into NAMDYNA YOMDYNA and to do the checkings in SUCTRL_GFLATTR x Add a new attribute in TYPE_GFL_COMP This attribute must be added in the definition of TYPE GFL COMP in module yom_ygfl F90 This attribute must be added preferably in routine DEFINE GFL_COMP input dummy argument and internal code to update YDGFLC YGFL and YPTRC or and in routine SET GFL_ATTR input dummy argument and internal code to update YDGFLC and YGFL if necessary If the dummy arguments of DEFINE GFL_COMP are modified SUGFL2 must be updated If the dummy arguments of SET_GFL_ATTR are modified SUGFL3 must be updated The new attribute must be printed in PRINT_GFL module gfl_subs mod F90 1
14. eads NAMCHEM if there are chemistry fields activated NCHEM gt 0 SUGFL2 partially fills YGFL Y X and YGFL YCOMP calls DEFINE_GFL COMP encapsulated in gfl_subs_mod F90 for each GFL variable for that SUGFL3 ends to fill YGFL Y X and YGFL YCOMP calls SET_GFL ATTR encapsulated in gfl_subs_mod F90 for each GFL variable for that 3 Add anew GEL variable main features 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 module gfl_subs mod F90 module type_gfifids F90 module yom_ygfl F90 namelist namgfl nam h setup sugfll F90 setup sudefo_gflattr F90 setup suctrl_gflattr F90 setup sugfl2 F90 setup sugfl3 F90 setup sudyn F90 adiab cptend F90 adiab cptend_new F 90 adiab cptendsm F90 adiab cptend_flex F90 adiab gprcp F90 phys dmn cpchet F90 phys dmn mf_phys F90 adiab cpg_pt F90 module yomphyder F90 phys_ec ec_phys F90 phys_ec callpar F90 and routines under phys_ec callpar F90 for ECMWF applications only module gfl_subs_mod F90 Some GFL are linked to clouds for the time being liquid water ice cloud fraction rain snow graupels hail convective precipitation flux and in some applications they need to be deallocated routine DEACT CLOUD _GFL or reallocated routine REACT_CLOUD _GFL If a new GFL of generic name VGFL enters this series of cloud variables some code must be added in module gfl_subs_mod F90
15. f the final value of LVSPLIP if the LVSPLIP semi Lagrangian interpolator is allowed for the new GFL a test must be added to check that the other attributes linked to semi Lagrangian interpolators are F see what is coded for ozone 1 2 Attribute LHV if the LHV semi Lagrangian interpolator is not coded for the new GFL the test generating an ABOR1 must be updated if the LHV semi Lagrangian interpolator is allowed for the new GFL a test must be added to check that the attribute LSLHD is F see what is coded for ozone 1 3 Attribute LSLHD it is a priori assumed that the LSLHD semi Lagrangian interpolator is coded for the new GFL add the new GFL in the definition of the final value of LLSLHD 1 4 Attribute LGP It should be T for some GFL variables no spectral representation For some grid point GFL we should ensure that the Eulerian horizontal advections are not called check that if NSTOP gt 0 the semi Lagrangian scheme is switched on 1 5 Attribute LADV For some of the GFL variables LADV can be set to T only if LSLAG T because they are grid point GFL 1 6 Attribute LCOMAD it is a priori assumed that the LCOMAD semi Lagrangian interpolator is coded for the new GFL add the new GFL in the definition of the final value of LLCOMAD add the new GFL in the testing saying that SLHD and COMAD interpolations are not both switched on e Part 2 some consistency checks for G
16. note 21pp IDDH Yessad K 2014 Horizontal diffusion in the cycle 41 of ARPEGE IFS Internal note available on http www cnrm meteo fr gmapdoc 11
17. purpose of the present paper is to provide a user s guide for people wanting to introduce new GFL variables or new GFL attributes This is not a comprehensive documentation of all the GFL features A partial description of the GFL data flow can be found for example in Tudor 2004 GFL variables can be divided into two main classes the genuine prognostic variables like specific humidity and the pseudo historical variables Some of them are simply diagnosed in the code some other ones have an initial state which is read on a file Some of them can be post processed Some of them can be assimilated Some of them can be diffused in spectral space There are individual GFL for example specific humidity and groups of GFL for example aerosols AERO We first start to list the common features to be added when introducing GFL 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 GFL on files e how to post process GFL e how to assimilate GFL e how to apply horizontal diffusion on GFL e how to apply nudging on GFL e how to add new attributes 2 Basics about GFL dataflow Structures are defined in the following modules e yom_ygfl F90 contains TYPE_ GFL_COMP for attributes TYPE_GFL_NAML subset of attributes which are in namelist NAMGFL TYPE_GFLD for dimensions and variables typed by TYPE GFL_COMP and TYPE_GFL_NAML maximal
18. t 1 1 for a new group of GFL generic name GRUP a default value must be given to the new variable N GRUP e Part 1 2 in some cases this part may have to be modified e Parts 5 6 7 in some cases these parts may have to be modified e Part 8 update the value of NGEMS when relevant x setup sudefo_gflattr F 90 e Part la for a new individual GFL generic name VGFL a default value must be given to all the attributes of Y VGFL _NL Rules of coding are the following ones for each attribute and each case ALL the individual GFL appear and ALWAYS in the same order Attribute LT1 must be T for allocated prognostic GFL F otherwise e Part 1b for a new group of GFL generic name GRUP a default value must be given to all the at tributes of Y GRUP _NL for all values of JGFL in 1 to JP VGFL Attributs SHOULD always appear in the same order this is the only way to easily check that there is no attribute forgotten or repeated sev eral times The order currently used is LGP LSP NREQIN REFVALI LREQOUT LGPINGP LTRAJIO LDIAG LADV LADV5 LCDERS LT9 LT1 LT5 LPT LPC CNAME IGRBCODE NCOUPLING REFVALC LTDIABLIN LHORTURB LSLHD LCOMAD LVSPLIP LQM LQMH LHV LINTLIN LPHY x setup suctrl_gflattr F 90 e Part 1 some consistency checks for GFL attributes 1 1 Attribute LVSPLIP if the LVSPLIP semi Lagrangian interpolator is not coded for the new GFL the test generating an ABORI must be updated add the new GFL in the definition o
19. variables for example TFP_SN for snow TFP_O3MX for ozone That means that a new TFP_ VGFL must be added x setup suafn1 F90 suafn2 F90 suafn3 F90 In SUAFN1 default value for the new TFP_ VGFL or TFP_ GRUP topic GFL part 0 bis must be added calculation of the attribute ILMOD of the new TFP end of SUAFN1 must be added In SUAFN2 routine CTRL_TFP must be called for the new GFL variables part 1 In SUAFNS routine PRINT_TFP must be called for the new GFL variables part 1 The ordering of GFL and if possible of all the post processable variables must be the same in these three routines x fullpos endpos F 90 Part 1 1 1 for new individual GFL VGFL add line CALL SUPTRPPGFL_ENDPOS 1 IPPGFL YLPPIC I VGFL x fullpos endpos_prepgfl F 90 e Part 1 define LD GFL YDPPIC I VGFL for new individual GFL VGFL e Part 2 define LDS1_GFL YDPPIC I VGFL for new individual GFL VGFL e Part 3 define LDS2 GFL YDPPIC I VGFL for new individual GFL VGFL e Part 4 define KGFLCOD YDPPIC I VGFL for new individual GFL VGFL e Part 5 define KDIN GFL YDPPIC I VGFL for new individual GFL VGFL e Part 6 define LDIN GFL YDPPIC I VGFL for new individual GFL VGFL fullpos fpcordyn F 90 In the JFLD loop add an adequate piece of code for a new individual GFL VGFL if it should be corrected for example if it should be bounded by a lower threshold or an upper threshold and provide the proper value of ZMIN ZMA
20. y is nudged in spectral space if the attribute LSP is TRUE Other GFL variables are NEVER nudged If a new nudgeable spectral GFL is introduced with nudging in spectral space the following routines must be updated Later module elbcOa_mod F90 add variable SPNUD X ald_inc namelist nemelbc0a nam h add variable SPNUD X coupling external espcpl F90 add nudging on new GFL variable Side effects are expected on coupling external espsc2r F90 and aladin control espcm F 90 adding a new attribute SPNUD working like NCOUPLING may be studied no nudging if 0 nudging if above 0 externalisable part of spectral nudging must be rewritten in order to do spectral nudging on a subset of spectral GFL fields without explicitly naming them 10 Add a new GEL attribute x List of routines to update There is a minimal list of routines which should be updated module gfl_subs mod F90 module yom_ygfl F90 setup sudefo_gflattr F90 setup suctrl_gflattr F90 setup sugfl2 F90 setup sugfl3 F90 setup sudyn F90 setup sucslint F90 x Add a new attribute in TYPE_GFL_NAML This attribute must be added in the definition of TYPE GFL_NAML in module yom_ygfl F90 If the attribute is linked to a semi Lagrangian interpolator for example add a new interpolator routine SUCSLINT must be modified and must take account of this new attribute to compute the variable CLSLINT of SUGFL3 update the calls of CLSLINT in SUGFL3 This attribute must be a
Download Pdf Manuals
Related Search