enkf_mod

link to source code

Dependency Diagrams:

enkf_mod.svg

Direct Dependency Diagram

enkf_mod_rev.svg

Reverse Dependency Diagram

Description

MODULE enkf_mod (prefix=’enkf’ category=’1. High-level functionality’)

Purpose

Various routines that are useful for implementing an EnKF in MIDAS, including the LETKF.

Quick access

Types

struct_enkfinterpinfo

Routines

enkf_computevertlocation(), enkf_getmodulatedstate(), enkf_interpweights(), enkf_latlonalreadyfound(), enkf_letkfanalyses(), enkf_letkfgetmpiglobaltags(), enkf_letkfsetupmpidistribution(), enkf_modifyamsubobserror(), enkf_rejecthighlatir(), enkf_setupinterpinfo(), enkf_setupmodulationfactor(), getmodulationfactor()

Needed modules

  • mpi (mpi_statuses_ignore())

  • midasmpi_mod: MODULE midasMpi_mod (prefix=’mmpi’ category=’8. Low-level utilities and constants’)

  • utilities_mod: MODULE utilities_mod (prefix=’utl’ category=’8. Low-level utilities and constants’)

  • mathphysconstants_mod: MODULE mathPhysConstants_mod (prefix=’mpc’ category=’8. Low-level utilities and constants’)

  • timecoord_mod: MODULE timeCoord_mod (prefix=’tim’ category=’7. Low-level data objects’)

  • verticalcoord_mod: MODULE verticalCoord_mod (prefix=’vco’ category=’7. Low-level data objects’)

  • horizontalcoord_mod: MODULE horizontalCoord_mod (prefix=’hco’ category=’7. Low-level data objects’)

  • ensemblestatevector_mod: MODULE ensembleStateVector_mod (prefix=’ens’ category=’6. High-level data objects’)

  • gridstatevector_mod: MODULE gridStateVector_mod (prefix=’gsv’ category=’6. High-level data objects’)

  • obsspacedata_mod: MODULE obsSpaceData_mod (prefix=’obs’ category=’6. High-level data objects’)

  • tovsnl_mod: MODULE tovsNL_mod (prefix=’tvs’ category=’5. Observation operators’)

  • ensembleobservations_mod: MODULE ensembleObservations_mod (prefix=’eob’ category=’6. High-level data objects’)

  • gridvariabletransforms_mod: MODULE gridVariableTransforms_mod (prefix=’gvt’ category=’4. Data Object transformations’)

  • localizationfunction_mod: MODULE localizationFunction_mod (prefix=’lfn’ category=’2. B and R matrices’)

  • varnamelist_mod: MODULE varNameList_mod (prefix=’vnl’ category=’7. Low-level data objects’)

  • codeprecision_mod: MODULE codePrecision_mod (prefix=’pre’ category=’8. Low-level utilities and constants’)

  • codtyp_mod: MODULE codtyp_mod (prefix=’codtyp’ category=’8. Low-level utilities and constants’)

  • calcheightandpressure_mod: MODULE calcHeightAndPressure_mod (prefix=’czp’ category=’4. Data Object transformations’)

Types

  • type  enkf_mod/struct_enkfinterpinfo
    Type fields

    • % interpweights (*,*,*) [real ,allocatable]

    • % latindexes (*,*,*) [integer ,allocatable]

    • % latlonstep [integer ]

    • % lonindexes (*,*,*) [integer ,allocatable]

    • % mylatbeg [integer ]

    • % mylatbeghalo [integer ]

    • % mylatend [integer ]

    • % mylatendhalo [integer ]

    • % mylonbeg [integer ]

    • % mylonbeghalo [integer ]

    • % mylonend [integer ]

    • % mylonendhalo [integer ]

    • % numindexes (*,*) [integer ,allocatable]

Variables

Subroutines and functions

subroutine  enkf_mod/enkf_letkfanalyses(algorithm, numsubens, randomshufflesubens, ensembleanl, ensembletrl, ensobs_mpiglobal, ensobsgain_mpiglobal, statevectormeananl, winterpinfo, maxnumlocalobs, hlocalize, hlocalizepressure, vlocalize, mpidistribution, numretainedeigen)
Purpose

Local subroutine containing the code for computing the LETKF analyses for all ensemble members, ensemble mean.

Arguments

  • algorithm [character ,in]

  • numsubens [integer ,in]

  • randomshufflesubens [logical ,in]

  • ensembleanl [struct_ens ,inout]

  • ensembletrl [struct_ens ,inout,pointer]

  • ensobs_mpiglobal [struct_eob ,in,target]

  • ensobsgain_mpiglobal [struct_eob ,in]

  • statevectormeananl [struct_gsv ,in]

  • winterpinfo [struct_enkfinterpinfo ,in]

  • maxnumlocalobs [integer ,in]

  • hlocalize (*) [real ,in]

  • hlocalizepressure (*) [real ,in]

  • vlocalize [real ,in]

  • mpidistribution [character ,in]

  • numretainedeigen [integer ,in]

Called from

midas_letkf

Call to

utl_tmg_start(), enkf_letkfsetupmpidistribution(), ens_getnummembers(), ens_getnumlev(), ens_gethco(), ens_getvco(), gsv_allocate(), tim_getdatestamp(), gsv_zero(), ens_computemean(), ens_copyensmean(), utl_abort(), utl_randomorderint(), lfn_setup(), enkf_computevertlocation(), gsv_getvarnamefromk(), vnl_varlevelfromvarname(), gsv_getlevfromk(), vnl_varkindfromvarname(), utl_tmg_stop(), enkf_letkfgetmpiglobaltags(), eob_getlocalbodyindices(), lfn_response(), utl_matinverse(), utl_eigendecomp(), enkf_interpweights(), ens_getonelev_r4(), getmodulationfactor(), gsv_deallocate()

subroutine  enkf_mod/enkf_computevertlocation(vertlocation_r4, statevectormeantrl)
Purpose

Compute extract global 3D vertical location field from supplied stateVector. Can be either logPressure or depth levels.

Arguments

  • vertlocation_r4 (*,*,*) [real ,inout,allocatable]

  • statevectormeantrl [struct_gsv ,inout]

Called from

enkf_letkfanalyses()

Call to

gsv_getnumlev(), utl_abort(), gsv_allocate(), tim_getdatestamp(), gsv_zero(), gsv_copy(), gsv_transposetilestostep(), gsv_deallocate()

subroutine  enkf_mod/enkf_letkfsetupmpidistribution(mynumlatlonrecv, mynumlatlonsend, mylatindexesrecv, mylonindexesrecv, mylatindexessend, mylonindexessend, myprocindexesrecv, myprocindexessend, mynumprocindexessend, mpidistribution, winterpinfo)
Purpose

Setup for distribution of grid points over mpi tasks.

Arguments

  • mynumlatlonrecv [integer ,out]

  • mynumlatlonsend [integer ,out]

  • mylatindexesrecv (*) [integer ,out,allocatable]

  • mylonindexesrecv (*) [integer ,out,allocatable]

  • mylatindexessend (*) [integer ,out,allocatable]

  • mylonindexessend (*) [integer ,out,allocatable]

  • myprocindexesrecv (*) [integer ,out,allocatable]

  • myprocindexessend (*,*) [integer ,out,allocatable]

  • mynumprocindexessend (*) [integer ,out,allocatable]

  • mpidistribution [character ,in]

  • winterpinfo [struct_enkfinterpinfo ,in]

Called from

enkf_letkfanalyses()

Call to

enkf_latlonalreadyfound(), utl_abort()

subroutine  enkf_mod/enkf_letkfgetmpiglobaltags(latlontagmpiglobal, mylatindexesrecv, mylonindexesrecv)
Arguments

  • latlontagmpiglobal (*,*) [integer ,out]

  • mylatindexesrecv (*) [integer ,in]

  • mylonindexesrecv (*) [integer ,in]

Called from

enkf_letkfanalyses()

function  enkf_mod/enkf_latlonalreadyfound(alllatindexesrecv, alllonindexesrecv, latlonindex, procindex)
Arguments

  • alllatindexesrecv (*,*) [integer ,in]

  • alllonindexesrecv (*,*) [integer ,in]

  • latlonindex [integer ,in]

  • procindex [integer ,in]

Return

found [logical ]

Called from

enkf_letkfsetupmpidistribution()

subroutine  enkf_mod/enkf_setupinterpinfo(winterpinfo, hco, weightlatlonstep, mylonbeg, mylonend, mylatbeg, mylatend)
Purpose

Setup the weights and lat/lon indices needed to bilinearly interpolate the LETKF weights from a coarse grid to the full resolution grid. The coarseness of the grid is specified by the weightLatLonStep argument.

Arguments

  • winterpinfo [struct_enkfinterpinfo ,out]

  • hco [struct_hco ,in]

  • weightlatlonstep [integer ,in]

  • mylonbeg [integer ,in]

  • mylonend [integer ,in]

  • mylatbeg [integer ,in]

  • mylatend [integer ,in]

Called from

midas_letkf

subroutine  enkf_mod/enkf_interpweights(winterpinfo, weights)
Purpose

Perform the bilinear interpolation of the weights using the precalculated interpolation info.

Arguments

  • winterpinfo [struct_enkfinterpinfo ,in]

  • weights (,,-(winterpinfo%mylonbeghalo)+1,-(winterpinfo%mylatbeghalo)+1) [real ,out]

Called from

enkf_letkfanalyses()

subroutine  enkf_mod/enkf_modifyamsubobserror(obsspacedata)
Arguments

obsspacedata [struct_obs ,inout,target]

Called from

midas_letkf, midas_prepcma

Call to

obs_numheader(), obs_headelem_r(), obs_headelem_i(), codtyp_get_codtyp()

subroutine  enkf_mod/enkf_rejecthighlatir(obsspacedata)
Arguments

obsspacedata [struct_obs ,inout,target]

Called from

midas_letkf, midas_prepcma

Call to

obs_numheader(), obs_headelem_r(), obs_headelem_i(), tvs_isidburphyperspectral(), obs_bodyset_i(), obs_bodyelem_i()

subroutine  enkf_mod/enkf_getmodulatedstate(statevector_in, statevectormeantrl, vlocalizelengthscale, numretainedeigen, nens, eigenvectorcolumnindex, statevector_out, besilent)
Purpose

Compute vertical localization matrix, and the corresponding eigenvectors/eigenvalues, to obtain modulated stateVector.

Arguments

  • statevector_in [struct_gsv ,in]

  • statevectormeantrl [struct_gsv ,in]

  • vlocalizelengthscale [real ,in]

  • numretainedeigen [integer ,in]

  • nens [integer ,in]

  • eigenvectorcolumnindex [integer ,in]

  • statevector_out [struct_gsv ,inout]

  • besilent [logical ,in]

Called from

midas_ensembleh, midas_letkf

Call to

utl_tmg_start(), utl_abort(), gsv_copy(), gsv_add(), gsv_varexist(), getmodulationfactor(), utl_tmg_stop()

subroutine  enkf_mod/enkf_setupmodulationfactor(vco, numretainedeigen, nens, vlocalizelengthscale, besilent)
Purpose

setup modulationFactorArray by calling getModulationFactor for first time.

Arguments

  • vco [struct_vco ,in,pointer]

  • numretainedeigen [integer ,in]

  • nens [integer ,in]

  • vlocalizelengthscale [real ,in]

  • besilent [logical ,in]

Called from

midas_letkf

Call to

getmodulationfactor()

subroutine  enkf_mod/getmodulationfactor(vco, eigenvectorlevelindex, eigenvectorcolumnindex, numretainedeigen, nens, vlocalizelengthscale, modulationfactor_r4[, besilent_opt])
Purpose

compute modulation factor needed to multiply ensemble perturbation to get the modulated perturbation: (Nens*nLambda/(Nens - 1))^1/2 * Lambda^1/2

Arguments

  • vco [struct_vco ,in,pointer]

  • eigenvectorlevelindex [integer ,in]

  • eigenvectorcolumnindex [integer ,in]

  • numretainedeigen [integer ,in]

  • nens [integer ,in]

  • vlocalizelengthscale [real ,in]

  • modulationfactor_r4 [real ,out]

Options

besilent_opt [logical ,in,]

Called from

enkf_letkfanalyses(), enkf_getmodulatedstate(), enkf_setupmodulationfactor()

Call to

lfn_setup(), lfn_response(), utl_eigendecomp(), utl_abort()