Changes between Initial Version and Version 1 of ImplementAnalysisGlobal


Ignore:
Timestamp:
Nov 16, 2020, 12:43:56 PM (2 years ago)
Author:
lnerger
Comment:

--

Legend:

Unmodified
Added
Removed
Modified
  • ImplementAnalysisGlobal

    v1 v1  
     1= Implementation of the Analysis step for the global filters =
     2
     3{{{
     4#!html
     5<div class="wiki-toc">
     6<h4>Implementation Guide</h4>
     7<ol><li><a href="ImplementationGuide">Main page</a></li>
     8<li><a href="AdaptParallelization">Adaptation of the parallelization</a></li>
     9<li><a href="InitPdaf">Initialization of PDAF</a></li>
     10<li><a href="ModifyModelforEnsembleIntegration">Modifications for ensemble integration</a></li>
     11<li><a href="ImplementationofAnalysisStep">Implementation of the analysis step</a></li>
     12<ol>
     13<li>Implementation for global filters</li>
     14<li><a href="ImplementAnalysisLocal">Implementation for local filters</a></li>
     15<li><a href="ImplementAnalysisestkf">Implementation for ESTKF</a></li>
     16<li><a href="ImplementAnalysislestkf">Implementation for LESTKF</a></li>
     17<li><a href="ImplementAnalysisetkf">Implementation for ETKF</a></li>
     18<li><a href="ImplementAnalysisletkf">Implementation for LETKF</a></li>
     19<li><a href="ImplementAnalysisseik">Implementation for SEIK</a></li>
     20<li><a href="ImplementAnalysislseik">Implementation for LSEIK</a></li>
     21<li><a href="ImplementAnalysisseek">Implementation for SEEK</a></li>
     22<li><a href="ImplementAnalysisenkf">Implementation for EnKF</a></li>
     23<li><a href="ImplementAnalysislenkf">Implementation for LEnKF</a></li>
     24<li><a href="ImplementAnalysisnetf">Implementation for NETF</a></li>
     25<li><a href="ImplementAnalysislnetf">Implementation for LNETF</a></li>
     26<li><a href="ImplementAnalysispf">Implementation for PF</a></li>
     27</ol>
     28<li><a href="AddingMemoryandTimingInformation">Memory and timing information</a></li>
     29<li><a href="EnsembleGeneration">Ensemble Generation</a></li>
     30<li><a href="DataAssimilationDiagnostics">Diagnostics</a></li>
     31</ol>
     32</div>
     33}}}
     34
     35[[PageOutline(2-3,Contents of this page)]]
     36
     37== Overview ==
     38
     39With Version 1.8 of PDAF, the ESTKF [Error Subspace Transform Kalman Filter] algorithm has been introduced. The user-supplied routines required for the ESTKF are identical to those required for the SEIK filter and amost identical to those required for the ETKF method.
     40
     41For the analysis step of the ESTKF different operations related to the observations are needed. These operations are requested by PDAF by call-back routines supplied by the user. Intentionally, the operations are split into separate routines in order to keep the operations rather elementary and efficient. This procedure should simplify the implementation. The names of the required routines are specified in the call to the routine `PDAF_assimilate_estkf` in the fully-parallel implementation (or `PDAF_put_state_estkf` for the 'flexible' implementation) that was discussed before. With regard to the parallelization, all these routines are executed by the filter processes (`filterpe=.true.`) only.
     42
     43For completeness we discuss here all user-supplied routines that are specified in the interface to PDAF_assimilate_estkf. Thus, some of the user-supplied routines that are explained on the page describing the modification of the model code for the ensemble integration are repeated here.
     44
     45The ESTKF and the ETKF (Ensemble Transform Kalman Filter) are very similar. For this reason, the interface to the user-supplied routines is almost identical. Depending on the implementation it can be possible to use identical routines for the ESTKF and the ETKF. Differences are marked in the text below.
     46
     47== `PDAF_assimilate_estkf` ==
     48
     49The general aspects of the filter specific routines `PDAF_assimilate_*` have been described on the page [ModifyModelforEnsembleIntegration Modification of the model code for the ensemble integration] and its sub-page on [InsertAnalysisStep inserting the analysis step]. The routine is used in the fully-parallel implementation variant of the data assimilation system. When the 'flexible' implementation variant is used, the routines `PDAF_put_state_*' is used as described further below. Here, we list once more the full interface of the routine. Subsequently, the full set of user-supplied routines specified in the call to `PDAF_assimilate_estkf` is explained.
     50
     51The interface when using the ESTKF is the following:
     52{{{
     53  SUBROUTINE PDAF_assimilate_estkf(U_collect_state, U_distribute_state, U_init_dim_obs, &
     54                                 U_obs_op, U_init_obs, U_prepoststep, U_prodRinvA, &
     55                                 U_init_obsvar, U_next_observation, status)
     56}}}
     57with the following arguments:
     58 * [#U_collect_statecollect_state_pdaf.F90 U_collect_state]: The name of the user-supplied routine that initializes a state vector from the array holding the ensemble of model states from the model fields. This is basically the inverse operation to `U_distribute_state` used in `PDAF_get_state` as well as here.
     59 * [#U_distribute_statedistribute_state_pdaf.F90 U_distribute_state]:  The name of a user supplied routine that initializes the model fields from the array holding the ensemble of model state vectors.
     60 * [#U_init_dim_obsinit_dim_obs_pdaf.F90 U_init_dim_obs]: The name of the user-supplied routine that provides the size of observation vector
     61 * [#U_obs_opobs_op_pdaf.F90 U_obs_op]: The name of the user-supplied routine that acts as the observation operator on some state vector
     62 * [#U_init_obsinit_obs_pdaf.F90 U_init_obs]: The name of the user-supplied routine that initializes the vector of observations
     63 * [#U_prepoststepprepoststep_ens_pdaf.F90 U_prepoststep]: The name of the pre/poststep routine as in `PDAF_get_state`
     64 * [#U_prodRinvAprodrinva_pdaf.F90 U_prodRinvA]: The name of the user-supplied routine that computes the product of the inverse of the observation error covariance matrix with some matrix provided to the routine by PDAF. This operation occurs during the analysis step of the SEIK, ETKF, and ESTKF algorithms.
     65 * [#U_init_obsvarinit_obsvar_pdaf.F90 U_init_obsvar]: The name of the user-supplied routine that provides a mean observation error variance to PDAF (This routine will only be executed, if an adaptive forgetting factor is used)
     66 * [#U_next_observationnext_observation.F90 U_next_observation]: The name of a user supplied routine that initializes the variables `nsteps`, `timenow`, and `doexit`. The same routine is also used in `PDAF_get_state`.
     67 * `status`: The integer status flag. It is zero, if `PDAF_assimilate_estkf` is exited without errors.
     68
     69
     70== `PDAF_put_state_estkf` ==
     71
     72When the 'flexible' implementation variant is chosen for the assimilation system, the routine `PDAF_put_state_estkf` has to be used instead of `PDAF_assimilate_estkf`. The general aspects of the filter specific routines `PDAF_put_state_*` have been described on the page [ModifyModelforEnsembleIntegration Modification of the model code for the ensemble integration]. The interface of the routine is identical with that of `PDAF_assimilate_estkf` with the exception the specification of the user-supplied routines `U_distribute_state` and `U_next_observation` are missing.
     73
     74The interface when using the ESTKF is the following:
     75{{{
     76  SUBROUTINE PDAF_put_state_estkf(U_collect_state, U_init_dim_obs, U_obs_op, &
     77                                 U_init_obs, U_prepoststep, U_prodRinvA, U_init_obsvar, status)
     78}}}
     79
     80== User-supplied routines ==
     81
     82Here all user-supplied routines are described that are required in the call to `PDAF_assimilate_estkf`. For some of the generic routines, we link to the page on [ModifyModelforEnsembleIntegration modifying the model code for the ensemble integration].
     83
     84To indicate user-supplied routines we use the prefix `U_`. In the template directory `templates/` as well as in the example implementation in `testsuite/src/dummymodel_1D` these routines exist without the prefix, but with the extension `_pdaf.F90`. In the section titles below we provide the name of the template file in parentheses.
     85
     86In the subroutine interfaces some variables appear with the suffix `_p`. This suffix indicates that the variable is particular to a model sub-domain, if a domain decomposed model is used. Thus, the value(s) in the variable will be different for different model sub-domains.
     87
     88
     89=== `U_collect_state` (collect_state_pdaf.F90) ===
     90
     91This routine is independent of the filter algorithm used.
     92See the page on [InsertAnalysisStep#U_collect_statecollect_state_pdaf.F90 inserting the analysis step] for the description of this routine.
     93
     94
     95=== `U_distribute_state` (distribute_state_pdaf.F90) ===
     96
     97This routine is independent of the filter algorithm used.
     98See the page on [InsertAnalysisStep#U_distribute_statedistribute_state_pdaf.F90 inserting the analysis step] for the description of this routine.
     99
     100
     101=== `U_init_dim_obs` (init_dim_obs_pdaf.F90) ===
     102
     103This routine is used by all global filter algorithms (SEEK, SEIK, EnKF, ETKF, ESTKF).
     104
     105The interface for this routine is:
     106{{{
     107SUBROUTINE init_dim_obs(step, dim_obs_p)
     108
     109  INTEGER, INTENT(in)  :: step       ! Current time step
     110  INTEGER, INTENT(out) :: dim_obs_p  ! Dimension of observation vector
     111}}}
     112
     113The routine is called at the beginning of each analysis step.  It has to initialize the size `dim_obs_p` of the observation vector according to the current time step. Without parallelization `dim_obs_p` will be the size for the full model domain. When a domain-decomposed model is used, `dim_obs_p` will be the size of the observation vector for the sub-domain of the calling process.
     114
     115Some hints:
     116 * It can be useful to not only determine the size of the observation vector at this point. One can also already gather information about the locations of the observations, which will be used later, e.g. to implement the observation operator. An array for the locations can be defined in a module like `mod_assimilation` of the example implementation.
     117
     118
     119=== `U_obs_op` (obs_op_pdaf.F90) ===
     120
     121This routine is used by all global filter algorithms (SEEK, SEIK, EnKF, ETKF, ESTKF).
     122
     123The interface for this routine is:
     124{{{
     125SUBROUTINE obs_op(step, dim_p, dim_obs_p, state_p, m_state_p)
     126
     127  INTEGER, INTENT(in) :: step               ! Currrent time step
     128  INTEGER, INTENT(in) :: dim_p              ! PE-local dimension of state
     129  INTEGER, INTENT(in) :: dim_obs_p          ! Dimension of observed state
     130  REAL, INTENT(in)    :: state_p(dim_p)     ! PE-local model state
     131  REAL, INTENT(out) :: m_state_p(dim_obs_p) ! PE-local observed state
     132}}}
     133
     134The routine is called during the analysis step. It has to perform the operation of the observation operator acting on a state vector that is provided as `state_p`. The observed state has to be returned in `m_state_p`.
     135
     136For a model using domain decomposition, the operation is on the PE-local sub-domain of the model and has to provide the observed sub-state for the PE-local domain.
     137
     138Hint:
     139 * If the observation operator involves a global operation, e.g. some global integration, while using domain-decomposition one has to gather the information from the other model domains using MPI communication.
     140
     141
     142=== `U_init_obs` (init_obs_pdaf.F90) ===
     143
     144This routine is used by all global filter algorithms (SEEK, SEIK, EnKF, ETKF, ESTKF).
     145
     146The interface for this routine is:
     147{{{
     148SUBROUTINE init_obs(step, dim_obs_p, observation_p)
     149
     150  INTEGER, INTENT(in) :: step             ! Current time step
     151  INTEGER, INTENT(in) :: dim_obs_p        ! PE-local dimension of obs. vector
     152  REAL, INTENT(out)   :: observation_p(dim_obs_p) ! PE-local observation vector
     153}}}
     154
     155The routine is called during the analysis step.
     156It has to provide the vector of observations in `observation_p` for the current time step.
     157
     158For a model using domain decomposition, the vector of observations that exist on the model sub-domain for the calling process has to be initialized.
     159
     160
     161=== `U_prepoststep` (prepoststep_ens_pdaf.F90) ===
     162
     163The routine has already been described on the [ModifyModelforEnsembleIntegration#U_prepoststepprepoststep_ens_pdaf.F90 page on modifying the model code for the ensemble integration]. For completeness, the description is repeated:
     164
     165The interface of the routine is identical for all filters. However, the particular operations that are performed in the routine can be specific for each filter algorithm.
     166
     167The interface for this routine is
     168{{{
     169SUBROUTINE prepoststep(step, dim_p, dim_ens, dim_ens_p, dim_obs_p, &
     170                       state_p, Uinv, ens_p, flag)
     171
     172  INTEGER, INTENT(in) :: step        ! Current time step
     173                         ! (When the routine is called before the analysis -step is provided.)
     174  INTEGER, INTENT(in) :: dim_p       ! PE-local state dimension
     175  INTEGER, INTENT(in) :: dim_ens     ! Size of state ensemble
     176  INTEGER, INTENT(in) :: dim_ens_p   ! PE-local size of ensemble
     177  INTEGER, INTENT(in) :: dim_obs_p   ! PE-local dimension of observation vector
     178  REAL, INTENT(inout) :: state_p(dim_p) ! PE-local forecast/analysis state
     179                                     ! The array 'state_p' is not generally not initialized in the case of SEIK/EnKF/ETKF/ESTKF.
     180                                     ! It can be used freely in this routine.
     181  REAL, INTENT(inout) :: Uinv(dim_ens-1, dim_ens-1) ! Inverse of matrix U
     182  REAL, INTENT(inout) :: ens_p(dim_p, dim_ens)      ! PE-local state ensemble
     183  INTEGER, INTENT(in) :: flag        ! PDAF status flag
     184}}}
     185
     186The routine `U_prepoststep` is called once at the beginning of the assimilation process. In addition, it is called during the assimilation cycles before the analysis step and after the ensemble transformation. The routine is called by all filter processes (that is `filterpe=1`).
     187
     188The routine provides for the user the full access to the ensemble of model states. Thus, user-controlled pre- and post-step operations can be performed.  For example the forecast and the analysis states and ensemble covariance matrix can be analyzed, e.g. by computing the estimated variances. In addition, the estimates can be written to disk.
     189
     190Hint:
     191 * If a user considers to perform adjustments to the estimates (e.g. for balances), this routine is the right place for it.
     192 * Only for the SEEK filter the state vector (`state_p`) is initialized. For all other filters, the array is allocated, but it can be used freely during the execution of `U_prepoststep`.
     193 * The interface has a difference for ETKF and ESTKF: For the ETKF, the array `Uinv` has size `dim_ens` x `dim_ens`. In contrast it has size `dim_ens-1` x `dim_ens-1` for the ESTKF. (For most cases, this will be irrelevant, because most usually the ensemble array `ens_p` is used for computations, rather than `Uinv`. Only for the SEIK filter with fixed covariance matrix, `Uinv` is required to compute the estimate analysis error. The fixed covariance matrix mode is not available for the ETKF or ESTKF.)
     194 * The interface through which `U_prepoststep` is called does not include the array of smoothed ensembles. In order to access the smoother ensemble array one has to set a pointer to it using a call to the routine `PDAF_get_smootherens` (see page on [AuxiliaryRoutines auxiliary routines])
     195
     196
     197=== `U_prodRinvA` (prodrinva_pdaf.F90) ===
     198
     199This routine is used by all filter algorithms that use the inverse of the observation error covariance matrix (SEEK, SEIK, ETKF, ESTKF).
     200
     201The interface for this routine is:
     202{{{
     203SUBROUTINE prodRinvA(step, dim_obs_p, rank, obs_p, A_p, C_p)
     204
     205  INTEGER, INTENT(in) :: step                ! Current time step
     206  INTEGER, INTENT(in) :: dim_obs_p           ! PE-local dimension of obs. vector
     207  INTEGER, INTENT(in) :: rank                ! Rank of initial covariance matrix
     208  REAL, INTENT(in)    :: obs_p(dim_obs_p)    ! PE-local vector of observations
     209  REAL, INTENT(in)    :: A_p(dim_obs_p,rank) ! Input matrix from analysis routine
     210  REAL, INTENT(out)   :: C_p(dim_obs_p,rank) ! Output matrix
     211}}}
     212
     213The routine is called during the analysis step. In the algorithms the product of the inverse of the observation error covariance matrix with some matrix has to be computed. For the ESTKF this matrix holds the observed part of the ensemble perturbations. The matrix is provided as `A_p`. The product has to be given as `C_p`.
     214
     215For a model with domain decomposition, `A_p` contains the part of the matrix that resides on the model sub-domain of the calling process. The product has to be computed for this sub-domain, too.
     216
     217Hints:
     218 * The routine does not require that the product is implemented as a real matrix-matrix product. Rather, the product can be implemented in its most efficient form. For example, if the observation error covariance matrix is diagonal, only the multiplication of the diagonal with matrix `A_p` has to be implemented.
     219 * The observation vector `obs_p` is provided through the interface for cases where the observation error variance is relative to the actual value of the observations.
     220 * The interface has a difference for ESTKF and ETKF: For ETKF the third argument is the ensemble size (`dim_ens`), while for the ESTKF it is the rank (`rank`) of the covariance matrix (usually ensemble size minus one). In addition, the second dimension of `A_p` and `C_p` has size `dim_ens` for ETKF, while it is `rank` for the ESTKF.  (Practically, one can usually ignore this difference as the fourth argument of the interface can be named arbitrarily in the routine.)
     221
     222=== `U_init_obsvar` (init_obsvar_pdaf.F90) ===
     223
     224This routine is used by the global filter algorithms SEIK, ETKF, and ESTKF as well as the local filters LSEIK, LETKF, ad LESTKF. The routine is only called if the adaptive forgetting factor is used (`type_forget=1` in the example impementation).
     225
     226The interface for this routine is:
     227{{{
     228SUBROUTINE init_obsvar(step, dim_obs_p, obs_p, meanvar)
     229
     230  INTEGER, INTENT(in) :: step          ! Current time step
     231  INTEGER, INTENT(in) :: dim_obs_p     ! PE-local dimension of observation vector
     232  REAL, INTENT(in) :: obs_p(dim_obs_p) ! PE-local observation vector
     233  REAL, INTENT(out)   :: meanvar       ! Mean observation error variance
     234}}}
     235
     236The routine is called in the global filters during the analysis or
     237by the routine that computes an adaptive forgetting factor (PDAF_set_forget).
     238The routine has to initialize the mean observation error variance. 
     239For the global filters this should be the global mean.
     240
     241Hints:
     242 * For a model with domain-decomposition one might use the mean variance for the model sub-domain of the calling process. Alternatively one can compute a mean variance for the full model domain using MPI communication (e.g. the function `MPI_allreduce`).
     243 * The observation vector `obs_p` is provided to the routine for the case that the observation error variance is relative to the value of the observations.
     244
     245
     246=== `U_next_observation` (next_observation_pdaf.F90) ===
     247
     248This routine is independent of the filter algorithm used.
     249See the page on [InsertAnalysisStep#U_next_observationnext_observation_pdaf.F90 inserting the analysis step] for the description of this routine.
     250
     251
     252== Execution order of user-supplied routines ==
     253
     254For the ESTKF, the user-supplied routines are essentially executed in the order they are listed in the interface to `PDAF_assimilate_estkf`. The order can be important as some routines can perform preparatory work for later routines. For example, `U_init_dim_obs` can prepare an index array that provides the information for executing the observation operator in `U_obs_op`.
     255
     256Before the analysis step is called the following routine is executed:
     257 1. [#U_collect_statecollect_state_pdaf.F90 U_collect_state]
     258
     259The analysis step is executed when the ensemble integration of the forecast is completed. During the analysis step the following routines are executed in the given order:
     260 1. [#U_prepoststepprepoststep_ens_pdaf.F90 U_prepoststep] (Call to act on the forecast ensemble, called with negative value of the time step)
     261 1. [#U_init_dim_obsinit_dim_obs_pdaf.F90 U_init_dim_obs]
     262 1. [#U_obs_opobs_op_pdaf.F90 U_obs_op] (A single call to operate on the ensemble mean state)
     263 1. [#U_init_obsinit_obs_pdaf.F90 U_init_obs]
     264 1. [#U_obs_opobs_op_pdaf.F90 U_obs_op] (`dim_ens` calls: one call for each ensemble member)
     265 1. [#U_init_obsvarinit_obsvar_pdaf.F90 U_init_obsvar] (Only executed, if the adaptive forgetting factor is used (`type_forget=1` in the example implemention))
     266 1. [#U_prodRinvAprodrinva_pdaf.F90 U_prodRinvA]
     267 1. [#U_prepoststepprepoststep_ens_pdaf.F90 U_prepoststep] (Call to act on the analysis ensemble, called with (positive) value of the time step)
     268
     269In case of the routine `PDAF_assimilate_estkf`, the following routines are executed after the analysis step:
     270 1. [#U_distribute_statedistribute_state_pdaf.F90 U_distribute_state]
     271 1. [#U_next_observationnext_observation_pdaf.F90 U_next_observation]