Changes between Initial Version and Version 1 of ExternalModelLoop


Ignore:
Timestamp:
Apr 29, 2014, 4:28:39 PM (6 years ago)
Author:
lnerger
Comment:

--

Legend:

Unmodified
Added
Removed
Modified
  • ExternalModelLoop

    v1 v1  
     1= Modification of the model code for the 'flexible' ensemble integration =
     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>Modifications for ensemble integration</li>
     11<li><a href="ImplementationofAnalysisStep">Implementation of the analysis step</a></li>
     12<li><a href="AddingMemoryandTimingInformation">Memory and timing information</a></li>
     13</ol>
     14</div>
     15}}}
     16
     17[[PageOutline(2-3,Contents of this page)]]
     18
     19== Overview ==
     20
     21For the flexible implementation variant of the assimilation system, one has to modify the model code so that it is possible to compute several integrations of model states successively. This is possible by adding an additional loop outside of the regular time-stepping loop of the model. This strategy has the potential to the required chances in the model code to the minimum, while allowing for the flexibility. In addition, a routine that simulates model errors might be required to be inserted into the time stepping loop of the model. The required extensions are described below.
     22
     23
     24== External ensemble loop ==
     25
     26The external loop for the ensemble integration has to enclose the time stepping loop of the model. Next to the external loop, a control structure for exiting the external loop as well as two calls to subroutines of PDAF have to be added. These are the calls to `PDAF_get_state` and a filter-specific routine like `PDAF_put_state_seik` for the SEIK filter. Both routines are described in sections below.
     27
     28The extended model code can look like this for the SEIK filter:
     29 {{{
     30  pdaf_modelloop: DO 
     31
     32     CALL PDAF_get_state(nsteps, ..., doexit, ...)
     33
     34     ! Check whether forecast has to be performed
     35     ifcontrol: IF (doexit /= 1) THEN
     36     
     37        IF (nsteps > 0) THEN
     38
     39          ... Time stepping code of the model ...         
     40
     41        END IF
     42
     43        CALL PDAF_put_state_seik(...)
     44
     45     ELSE ifcontrol
     46
     47        ! No more assimilation work; exit loop
     48        EXIT pdaf_modelloop
     49
     50     END IF ifcontrol
     51
     52  END DO pdaf_modelloop
     53}}}
     54In this example, which is taken from the example implementation in `testsuite/src/dummymodel_1D`, we use an unconditional DO loop (while loop). The exit flag `doexit` for this loop is set within `PDAF_get_state`. In addition, the variable `nsteps` is initialized, which defines the number of time steps to be performed during the current forecast phase. Thus, we only execute the time stepping code if `nsteps>0`. (If this has to be implemented using an IF-clause as in the example should be checked for the particular code).
     55
     56== `PDAF_get_state` ==
     57
     58The routine `PDAF_get_state` has the purpose to initialize the information, whether further model integrations have to be computed and how many time steps have to be performed. In addition, the model fields to be propagated are initialized from the array holding the ensemble states.
     59
     60The interface of `PDAF_get_state` is the following:
     61{{{
     62  SUBROUTINE PDAF_get_state(nsteps, timenow, doexit, U_next_observation, U_distribute_state, &
     63                            U_prepoststep, status)
     64}}}
     65with the following arguments:
     66 * `nsteps`: An integer specifying upon exit the number of time steps to be performed
     67 * `timenow`: A real specifying upon exit the current model time. 
     68 * `doexit`: An integer variable defining whether the assimilation process is completed and the program should exit the while loop. For compatibility 1 should be used for exit, 0 for continuing in the loop.
     69 * [#U_next_observationnext_observation.F90 U_next_observation]: The name of a user supplied routine that initializes the variables `nsteps`, `timenow`, and `doexit`
     70 * [#U_distribute_statedistribute_state.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
     71 * [#U_prepoststepprepoststep_seik.F90 U_prepoststep]: The name of a user supplied routine that is called before and after the analysis step. Here the user has the possibility to access the state ensemble and can e.g. compute estimated variances or can write the ensemble states the state estimate into files.
     72 * `status`: The integer status flag. It is zero, if `PDAF_get_state` is exited without errors.
     73
     74PDAF also has a [PdafSimplifiedInterface Simplified Interface] providing the routine `PDAF_get_state_si`. In the simplified interface, the names of all user-supplied call back routines are predefined such that they not appear in the call to `PDAF_get_state_si`. More information on the pre-defined names is provided in the [PdafSimplifiedInterface documentation of PDAF's simplified interface].
     75
     76== `PDAF_put_state_X` ==
     77
     78There is a separate routine `PDAF_put_state_X` for each of the filter algorithms. The name of the routine includes the name of the filter at its end (instead of `X`). The purpose of the `PDAF_put_state_X` routines is to write back the forecast model fields into the array holding the ensemble of model state vectors. In addition, the routine checks if the current forecast phase is completed. If not, the routine is exited and the next cycle of the ensemble loop is performed. If the current forecast phase is completed, the routine executes the analysis step of the chosen filter algorithm. The interface to each put-state routine is specific for each filter algorithm, because the names of several user-supplied routines have to be specified, which are specific for each filter algorithm. However, at the stage of implementing the ensemble integration only the first and last arguments of the routines are relevant.
     79
     80For example, the interface when using the SEIK filter is the following:
     81{{{
     82  SUBROUTINE PDAF_put_state_seik(U_collect_state, U_init_dim_obs, U_obs_op, &
     83                                 U_init_obs, U_prepoststep, U_prodRinvA, U_init_obsvar, status)
     84}}}
     85At this state of the implementation only these arguments are relevant:
     86 * [#U_collect_statecollect_state.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_dist_state` used in `PDAF_get_state`
     87 * `status`: The integer status flag. It is zero, if PDAF_get_state is exited without errors.
     88
     89The other arguments are names of user-supplied subroutines that are only executed if the analysis step is executed (See the section [#Compilationandtesting 'Compilation and testing'] for how to provide these routines for compilation at this stage). These routines are explained in the next section of the implementation guide ([ImplementationofAnalysisStep Implementation of the Analysis step]) separately for each available filter algorithm.
     90
     91PDAF also has a [PdafSimplifiedInterface Simplified Interface] providing the routine `PDAF_out_state_X_si`. In the simplified interface, the names of all user-supplied call back routines are predefined such that they not appear in the call to `PDAF_put_state_X_si`. More information on the pre-defined names is provided in the [PdafSimplifiedInterface documentation of PDAF's simplified interface].
     92
     93== User-supplied routines ==
     94
     95Here, only the user-supplied routines are discussed that are required at this stage of the implementation (that is, the ensemble integration). For testing (see [#Compilationandtesting 'Compilation and testing']), all routines need to exist, but only those described here in detail need to be implemented with functionality.
     96
     97To 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.
     98
     99=== `U_next_observation` (next_observation_pdaf.F90) ===
     100
     101The interface for this routine is
     102{{{
     103SUBROUTINE next_observation(stepnow, nsteps, doexit, timenow)
     104
     105  INTEGER, INTENT(in)  :: stepnow  ! Number of the current time step
     106  INTEGER, INTENT(out) :: nsteps   ! Number of time steps until next obs
     107  INTEGER, INTENT(out) :: doexit   ! Whether to exit forecasting (1 for exit)
     108  REAL, INTENT(out)    :: timenow  ! Current model (physical) time
     109}}}
     110
     111The routine is called once at the beginning of each forecast phase. It is executed by all processes that participate in the model integrations.
     112
     113Based on the information of the current time step, the routine has to define the number of time steps `nsteps` for the next forecast phase. In addition, the flag `doexit` has to be initialized to provide the information if the external ensemble loop can be exited. `timenow` is the current model time. This variable should also be initialized. It is particularly important, if an ensemble task integrates more than one model state. In this case `timenow` can be used to correctly jump back in time.
     114
     115Some hints:
     116 * If the time interval between successive observations is known, `nsteps` can be simply initialized by dividing the time interval by the size of the time step
     117 * `doexit` should be 0 to continue the assimilation process. In most cases `doexit` is set to 1, when `PDAF_get_state` is called after the last analysis for which observations are available.
     118 * At the first call to `U_next_obs` the variable `timenow` should be initialized with the current model time. At the next call a forecast phase has been completed. Thus, the new value of `timenow` follows from the timer interval for the previous forecast phase.
     119
     120=== `U_distribute_state` (distribute_state_pdaf.F90) ===
     121
     122The interface for this routine is
     123{{{
     124SUBROUTINE distribute_state(dim_p, state_p)
     125
     126  INTEGER, INTENT(in) :: dim_p           ! State dimension for PE-local model sub-domain
     127  REAL, INTENT(inout) :: state_p(dim_p)  ! State vector for PE-local model sub-domain
     128}}}
     129
     130This routine is called during the forecast phase as many times as there are states to be integrated by a model task. Again, the routine is executed by all processes that belong to model tasks.
     131
     132When the routine is called a state vector `state_p` and its size `dim_p` are provided. As the user has defined how the model fields are stored in the state vector, one can initialize the model fields from this information. If the model is not parallelized, `state_p` will contain a full state vector. If the model is parallelized using domain decomposition, `state_p` will contain the part of the state vector that corresponds to the model sub-domain for the calling process.
     133
     134Some hints:
     135 * If the state vector does not include all model fields, it can be useful to keep a separate array to store those additional fields. This array has to be kept separate from PDAF, but can be defined using a module like `mod_assimilation`.
     136
     137
     138=== `U_prepoststep` (prepoststep_ens_pdaf.F90) ===
     139
     140The 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. Here, we exemplify the interface on the example of the SEIK filter.
     141
     142The interface for this routine is
     143{{{
     144SUBROUTINE prepoststep(step, dim_p, dim_ens, dim_ens_p, dim_obs_p, &
     145                       state_p, Uinv, ens_p, flag)
     146
     147  INTEGER, INTENT(in) :: step        ! Current time step
     148                         ! (When the routine is called before the analysis -step is provided.)
     149  INTEGER, INTENT(in) :: dim_p       ! PE-local state dimension
     150  INTEGER, INTENT(in) :: dim_ens     ! Size of state ensemble
     151  INTEGER, INTENT(in) :: dim_ens_p   ! PE-local size of ensemble
     152  INTEGER, INTENT(in) :: dim_obs_p   ! PE-local dimension of observation vector
     153  REAL, INTENT(inout) :: state_p(dim_p) ! PE-local forecast/analysis state
     154                                     ! The array 'state_p' is not generally not initialized in the case of SEIK/EnKF/ETKF.
     155                                     ! It can be used freely in this routine.
     156  REAL, INTENT(inout) :: Uinv(dim_ens-1, dim_ens-1) ! Inverse of matrix U
     157  REAL, INTENT(inout) :: ens_p(dim_p, dim_ens)      ! PE-local state ensemble
     158  INTEGER, INTENT(in) :: flag        ! PDAF status flag
     159}}}
     160
     161The 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`).
     162
     163The 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. If the smoother is used, also the smoothed ensembles can be analyzed. In addition, the estimates can be written to disk.
     164
     165Hint:
     166 * If a user considers to perform adjustments to the estimates (e.g. for balances), this routine is the right place for it.
     167 * 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`.
     168 * 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])
     169
     170
     171=== `U_collect_state` (collect_state_pdaf.F90) ===
     172
     173The interface for this routine is
     174{{{
     175SUBROUTINE collect_state(dim_p, state_p)
     176
     177  INTEGER, INTENT(in) :: dim_p           ! State dimension for PE-local model sub-domain
     178  REAL, INTENT(inout) :: state_p(dim_p)  ! State vector for PE-local model sub-domain
     179}}}
     180
     181This routine is called during the forecast phase as many times as there are states to be integrated by a model task. It is called at the end of the integration of a member state of the ensemble. The routine is executed by all processes that belong to model tasks.
     182
     183When the routine is called, a state vector `state_p` and its size `dim_p` are provided. The operation to be performed in this routine is inverse to that of the routine `U_distribute_state`. That is, the state vector `state_p` has to be initialized from the model fields. If the model is not parallelized, `state_p` will contain a full state vector. If the model is parallelized using domain decomposition, `state_p` will contain the part of the state vector that corresponds to the model sub-domain for the calling process.
     184
     185Some hints:
     186 * If the state vector does not include all model fields, it can be useful to keep a separate array to store those additional fields. This array has to be kept separate from PDAF, but can be defined using a module like `mod_assimilation`.
     187
     188== Simulating model errors ==
     189
     190The implementation of the filter algorithms does not support the specification of a model error covariance matrix. This was left out, because in the SEEK and SEIK filter, the handling can be extremely costly, as the model error covariance matrix has to be projected onto the ensemble space. Instead PDAF support the simulation of model errors by disturbing fields during the model integration. For this, some routine will be required that is inserted into the time stepping loop of the model. As this procedure is specific to each model, the is no routine provided by PDAF for this.
     191
     192== Compilation and testing ==
     193
     194To compile the extended model code with PDAF, one has to extend the Makefile for the model by adding the additional user-supplied routines. While all of the user-supplied routines need to exist not all of them need to be fully implemented at this time if the following procedure is used. The routines that will not be called are `U_init_dim_obs`, `U_obs_op`, `U_init_obs`, `U_prodRinvA`, `U_init_obsvar`. A simple way to provide them for the compilation could be to copy the corresponding files (i.e. named without `U_`) from the template directory `templates/` and to include these files in the compilation and linking. These templates are simple stubs without any functionality.
     195
     196At this implementation stage one can use the preprocessor definition `PDAF_NO_UPDATE` (available from Version 1.6.1). With this, the actual analysis step of the chosen filter algorithm is not executed. Accordingly, only the user-supplied routines used in `PDAF_get_state` as well as the routine `U_collect_state` need to be implemented with functionality. The other routines will not be executed, because they are only called during the analysis step. Generally with `PDAF_NO_UPDATE` the program performs just an ensemble integration. That is, PDAF is initialized by `PDAF_init`. Then a forecast is computed by using `PDAF_get_state` and the chosen `PDAF_put_state_*` routine. At the initial time `U_prepoststep` is executed by `PDAF_get_state`. `U_next_obs` will provide the number of time steps to be computed by the model and `U_distributed_state` will initialize the model fields. Subsequently the ensemble integration is performed and the forecast fields are written back to the ensemble array by `U_collect_state`. Upon completion of the forecast phase, the routine `U_prepoststep` is executed twice. The first time is the regular call before the analysis is executed. Thus, it allows to access the forecast ensemble. If the analysis would not be deactivated, the second call to `U_prepoststep` would be after the analysis allowing access to the ensemble directly after the analysis. As the analysis is deactivated here, the ensemble will be the same as in the first call.
     197
     198This test allows to check the following:
     199 * Is `U_prepoststep` working correctly?
     200 * Does `U_next_observation` work correctly and is the information from this routine used correctly for the model integration
     201 * Are `U_distribute_state` and `U_collect_state` work correctly?
     202One could also comment out the actual time stepping part of the model. This would allow to only test the interfacing between PDAF and the model.
     203
     204It is important to ensure that the ensemble integration performs correctly. The simplest case should be a parallel configuration in which the number of model tasks equals the ensemble size as here the model tasks always compute forward in time. If the number of model tasks is smaller than the ensemble size, some model tasks will have to integrate multiple states of the ensemble. If a model task has to integrate two states, the model will have to jump back in time for the integration of the second state. It might be that some arrays of the model need to be re-initialized to ensure that the second integration is consistent. Also, one might need to check if the initialization of forcing fields (e.g. wind stress over the ocean) performs correctly for the second integration. (Sometimes model are implemented with the constraint that the model time always increases, which is the normal case for pure model simulations without assimilation.) A useful test is to initialize an ensemble in which all states are equal. If this ensemble is integrated the forecast states of the ensemble should, of course, still be equal.