API

Initialisation and finalisation

pyPDAF.PDAF.init	This function initialises the PDAF system.
pyPDAF.PDAF.omi_init	Allocating an array of obs_f derived types instances.
pyPDAF.PDAF.deallocate	Finalise the PDAF systems including freeing some of the memory used by PDAF.
pyPDAF.PDAF.print_info	Printing the wallclock time and memory measured by PDAF.

Fully parallel DA algorithms

Sequential DA

diagnoal observation matrix

pyPDAF.PDAF.omi_assimilate_global	Global filters except for 3DVar for a single DA step using diagnoal observation error covariance matrix.
pyPDAF.PDAF.omi_assimilate_lenkf	Covariance localised stochastic EnKF for a single DA step using diagnoal observation error covariance matrix.
pyPDAF.PDAF.localomi_assimilate	Domain local filters for a single DA step using diagnoal observation error covariance matrix.

non-diagnoal observation matrix

pyPDAF.PDAF.omi_assimilate_global_nondiagR	Global filters except for 3DVar and stochastic EnKF for a single DA step using non-diagnoal observation error covariance matrix.
pyPDAF.PDAF.omi_assimilate_enkf_nondiagR	Stochastic EnKF for a single DA step using non-diagnoal observation error covariance matrix.
pyPDAF.PDAF.omi_assimilate_lenkf_nondiagR	Covariance localised stochastic EnKF for a single DA step using non-diagnoal observation error covariance matrix.
pyPDAF.PDAF.localomi_assimilate_nondiagR	Domain local filters for a single DA step using non-diagnoal observation error covariance matrix.
pyPDAF.PDAF.omi_assimilate_nonlin_nondiagR	Global nonlinear filters for a single DA step using non-diagnoal observation error covariance matrix.
pyPDAF.PDAF.localomi_assimilate_lnetf_nondiagR	LNETF for a single DA step using non-diagnoal observation error covariance matrix.
pyPDAF.PDAF.localomi_assimilate_lknetf_nondiagR	LKNETF for a single DA step using non-diagnoal observation error covariance matrix.

Variational DA

diagnoal observation matrix

pyPDAF.PDAF.omi_assimilate_3dvar	3DVar DA for a single DA step using diagnoal observation error covariance matrix.
pyPDAF.PDAF.omi_assimilate_en3dvar_estkf	3DEnVar for a single DA step using diagnoal observation error covariance matrix.
pyPDAF.PDAF.omi_assimilate_hyb3dvar_estkf	Hybrid 3DEnVar for a single DA step using diagnoal observation error covariance matrix.
pyPDAF.PDAF.localomi_assimilate_en3dvar_lestkf	3DEnVar for a single DA step where the ensemble anomaly is generated by LESTKF using diagnoal observation error covariance matrix.
pyPDAF.PDAF.localomi_assimilate_hyb3dvar_lestkf	Hybrid 3DEnVar for a single DA step using diagnoal observation error covariance matrix.

non-diagnoal observation matrix

pyPDAF.PDAF.omi_assimilate_3dvar_nondiagR	3DVar DA for a single DA step using non-diagnoal observation error covariance matrix.
pyPDAF.PDAF.omi_assimilate_en3dvar_estkf_nondiagR	3DEnVar for a single DA step using non-diagnoal observation error covariance matrix.
pyPDAF.PDAF.omi_assimilate_hyb3dvar_estkf_nondiagR	Hybrid 3DEnVar for a single DA step using non-diagnoal observation error covariance matrix.
pyPDAF.PDAF.localomi_assimilate_en3dvar_lestkf_nondiagR	3DEnVar for a single DA step where the ensemble anomaly is generated by LESTKF using non-diagnoal observation error covariance matrix.
pyPDAF.PDAF.localomi_assimilate_hyb3dvar_lestkf_nondiagR	Hybrid 3DEnVar for a single DA step using diagnoal observation error covariance matrix.

Flexible DA algorithms

Sequential DA

diagnoal observation matrix

pyPDAF.PDAF.omi_put_state_global	Global filters except for 3DVar for a single DA step using diagnoal observation error covariance matrix without post-processing, distributing analysis, and setting next observation step.
pyPDAF.PDAF.omi_put_state_lenkf	Stochastic EnKF (ensemble Kalman filter) with covariance localisation using diagnoal observation error covariance matrix without post-processing, distributing analysis, and setting next observation step.
pyPDAF.PDAF.localomi_put_state	Domain local filters for a single DA step using diagnoal observation error covariance matrix without post-processing, distributing analysis, and setting next observation step.

diagnoal observation matrix

pyPDAF.PDAF.omi_put_state_global_nondiagR	Global filters except for 3DVar and stochastic EnKF for a single DA step using non-diagnoal observation error covariance matrix without post-processing, distributing analysis, and setting next observation step.
pyPDAF.PDAF.omi_put_state_enkf_nondiagR	Stochastic EnKF for a single DA step using non-diagnoal observation error covariance matrix without post-processing, distributing analysis, and setting next observation step.
pyPDAF.PDAF.omi_put_state_lenkf_nondiagR	Covariance localised stochastic EnKF for a single DA step using non-diagnoal observation error covariance matrix without post-processing, distributing analysis, and setting next observation step.
pyPDAF.PDAF.localomi_put_state_nondiagR	Domain local filters for a single DA step using non-diagnoal observation error covariance matrix without post-processing, distributing analysis, and setting next observation step.
pyPDAF.PDAF.omi_put_state_nonlin_nondiagR	Global nonlinear filters for a single DA step using non-diagnoal observation error covariance matrix without post-processing, distributing analysis, and setting next observation step.
pyPDAF.PDAF.localomi_put_state_lnetf_nondiagR	LNETF for a single DA step using non-diagnoal observation error covariance matrix without post-processing, distributing analysis, and setting next observation step.
pyPDAF.PDAF.localomi_put_state_lknetf_nondiagR	LKNETF for a single DA step using non-diagnoal observation error covariance matrix without post-processing, distributing analysis, and setting next observation step.

Variational DA

diagnoal observation matrix

pyPDAF.PDAF.omi_put_state_3dvar	3DVar DA for a single DA step using diagnoal observation error covariance matrix without post-processing, distributing analysis, and setting next observation step.
pyPDAF.PDAF.omi_put_state_en3dvar_estkf	3DEnVar for a single DA step using diagnoal observation error covariance matrix without post-processing, distributing analysis, and setting next observation step.
pyPDAF.PDAF.omi_put_state_hyb3dvar_estkf	Hybrid 3DEnVar for a single DA step using diagnoal observation error covariance matrix without post-processing, distributing analysis, and setting next observation step.
pyPDAF.PDAF.localomi_put_state_en3dvar_lestkf	3DEnVar for a single DA step where the ensemble anomaly is generated by LESTKF using diagnoal observation error covariance matrix without post-processing, distributing analysis, and setting next observation step.
pyPDAF.PDAF.localomi_put_state_hyb3dvar_lestkf	Hybrid 3DEnVar for a single DA step using diagnoal observation error covariance matrix without post-processing, distributing analysis, and setting next observation step.

non-diagnoal observation matrix

pyPDAF.PDAF.omi_put_state_3dvar_nondiagR	3DVar DA for a single DA step using non-diagnoal observation error covariance matrix without post-processing, distributing analysis, and setting next observation step.
pyPDAF.PDAF.omi_put_state_en3dvar_estkf_nondiagR	3DEnVar for a single DA step using non-diagnoal observation error covariance matrix without post-processing, distributing analysis, and setting next observation step.
pyPDAF.PDAF.omi_put_state_hyb3dvar_estkf_nondiagR	Hybrid 3DEnVar for a single DA step using non-diagnoal observation error covariance matrix without post-processing, distributing analysis, and setting next observation step.
pyPDAF.PDAF.localomi_put_state_en3dvar_lestkf_nondiagR	3DEnVar for a single DA step without post-processing, distributing analysis, and setting next observation step.
pyPDAF.PDAF.localomi_put_state_hyb3dvar_lestkf_nondiagR	Hybrid 3DEnVar for a single DA step using non-diagnoal observation error covariance matrix without post-processing, distributing analysis, and setting next observation step.

OMI functions

obs_f setter functions

pyPDAF.PDAF.omi_set_doassim	Setting the doassim attribute of obs_f.
pyPDAF.PDAF.omi_set_disttype	Setting the disttype attribute of obs_f.
pyPDAF.PDAF.omi_set_ncoord	Setting the ncoord attribute of obs_f.
pyPDAF.PDAF.omi_set_id_obs_p	Setting the id_obs_p attribute of obs_f.
pyPDAF.PDAF.omi_set_icoeff_p	This function sets the icoeff_p attribute of obs_f typically used in user-supplied function py__init_dim_obs_pdaf.
pyPDAF.PDAF.omi_set_domainsize	This function sets the domainsize attribute of obs_f typically used in user-supplied function py__init_dim_obs_pdaf.
pyPDAF.PDAF.omi_set_obs_err_type	This function sets the obs_err_type attribute of obs_f typically used in user-supplied function py__init_dim_obs_pdaf.
pyPDAF.PDAF.omi_set_use_global_obs	This function sets the use_global_obs attribute of obs_f typically used in user-supplied function py__init_dim_obs_pdaf.
pyPDAF.PDAF.omi_set_inno_omit	This function sets the inno_omit attribute of obs_f typically used in user-supplied function py__init_dim_obs_pdaf.
pyPDAF.PDAF.omi_set_inno_omit_ivar	This function sets the inno_omit_ivar attribute of obs_f typically used in user-supplied function py__init_dim_obs_pdaf.

Observation operators

pyPDAF.PDAF.omi_obs_op_gridpoint	A (partial) identity observation operator
pyPDAF.PDAF.omi_obs_op_gridavg	Observation operator that average values on given model grid points.
pyPDAF.PDAF.omi_obs_op_interp_lin	Observation operator that linearly interpolates model grid values to observation location.
pyPDAF.PDAF.omi_obs_op_adj_gridavg	The adjoint observation operator of pyPDAF.PDAF.omi_obs_op_gridavg().
pyPDAF.PDAF.omi_obs_op_adj_gridpoint	The adjoint observation operator of pyPDAF.PDAF.omi_obs_op_gridpoint().
pyPDAF.PDAF.omi_obs_op_adj_interp_lin	The adjoint observation operator of pyPDAF.PDAF.omi_obs_op_interp_lin().

Interpolations

pyPDAF.PDAF.omi_get_interp_coeff_tri	The coefficient for linear interpolation in 2D on unstructure triangular grid.
pyPDAF.PDAF.omi_get_interp_coeff_lin1D	The coefficient for linear interpolation in 1D.
pyPDAF.PDAF.omi_get_interp_coeff_lin	The coefficient for linear interpolation up to 3D.

Localisation

pyPDAF.PDAF.omi_obsstats_l	This function is called in the update routine of local filters and write statistics on locally used and excluded observations.
pyPDAF.PDAF.omi_weights_l	This function computes a weight vector according to the distances of observations from the local analysis domain with a vector of localisation radius.
pyPDAF.PDAF.omi_weights_l_sgnl	This function computes a weight vector according to the distances of observations from the local analysis domain with given localisation radius.
pyPDAF.PDAF.omi_init_dim_obs_l_iso	The function has to be called in init_dim_obs_l_OBTYPE in each observation module if a domain-localized filter (LESTKF/LETKF/LNETF/LSEIK)is used.
pyPDAF.PDAF.omi_init_dim_obs_l_noniso	The function has to be called in init_dim_obs_l_OBTYPE in each observation module if a domain-localized filter (LESTKF/LETKF/LNETF/LSEIK)is used.
pyPDAF.PDAF.omi_init_dim_obs_l_noniso_locweights	The function has to be called in init_dim_obs_l_OBTYPE in each observation module if a domain-localized filter (LESTKF/LETKF/LNETF/LSEIK)is used.
pyPDAF.PDAF.omi_localize_covar_iso	The function has to be called in localize_covar_OBTYPE in each observation module.
pyPDAF.PDAF.omi_localize_covar_noniso	The function has to be called in localize_covar_OBTYPE in each observation module.
pyPDAF.PDAF.omi_localize_covar_noniso_locweights	The function has to be called in localize_covar_OBTYPE in each observation module.

Others

pyPDAF.PDAF.omi_set_domain_limits	This is used to set the domain limits for the use of pyPDAF.PDAF.omi_set_use_global_obs.Currently, it only supports 2D limitations.
pyPDAF.PDAF.omi_set_debug_flag	This sets the debug flag for OMI.
pyPDAF.PDAF.omi_gather_obs	Gather the dimension of a given type of observation across multiple local domains/filter processors.
pyPDAF.PDAF.omi_gather_obsstate	This function is used to implement custom observation operators.
pyPDAF.PDAF.omi_gather_obsdims	This function gathers the information about the full dimension of each observation type in each process-local subdomain.
pyPDAF.PDAF.omi_obsstats	The function is called in the update routine of global filters and writes statistics on used and excluded observations.
pyPDAF.PDAF.omit_obs_omi	This function computes innovation and omit corresponding observations in assimilation if the innovation is too large.
pyPDAF.PDAF.omi_deallocate_obs	Deallocate OMI-internal obsrevation arrays
pyPDAF.PDAF.omi_check_error	This function returns the value of the PDAF-OMI internal error flag.

Local module functions

pyPDAF.PDAF.local_set_indices	Set index vector to map local state vector to global state vectors.
pyPDAF.PDAF.local_set_increment_weights	This function initialises a PDAF_internal local array of increment weights.
pyPDAF.PDAF.local_clear_increment_weights	This function deallocates the local increment weight vector in pyPDAF.PDAF.local_set_increment_weights if it is allocated

Utilities

Synthetic experiments

pyPDAF.PDAF.omi_generate_obs	Generation of synthetic observations based on given error statistics and observation operator for diagonal observation error covariance matrix.
pyPDAF.PDAF.generate_obs	Generation of synthetic observations based on given error statistics and observation operator.
pyPDAF.PDAF.omi_put_state_generate_obs	Generation of synthetic observations based on given error statistics and observation operator for diagonal observation error covariance matrix without post-processing, distributing analysis, and setting next observation step.
pyPDAF.PDAF.put_state_generate_obs	Generation of synthetic observations based on given error statistics and observation operator without post-processing, distributing analysis, and setting next observation step.

Statistical diagnostics

pyPDAF.PDAF.diag_effsample	Calculating the effective sample size of a particle filter.
pyPDAF.PDAF.diag_ensstats	Computing the skewness and kurtosis of the ensemble of a given element of the state vector.
pyPDAF.PDAF.diag_histogram	Computing the rank histogram of an ensemble.
pyPDAF.PDAF.diag_CRPS	Obtain a continuous rank probability score for an ensemble.
pyPDAF.PDAF.diag_CRPS_nompi	A continuous rank probability score for an ensemble without using MPI parallelisation.

Ensemble generation

pyPDAF.PDAF.eofcovar	EOF analysis of an ensemble of state vectors by singular value decomposition.
pyPDAF.PDAF.SampleEns	This function generates an ensemble from singular values and their vectors (EOF modes) centred on given mean state.

Diagnostics

pyPDAF.PDAF.get_assim_flag	This function returns the flag that indicates if the DA is performed in the last time step.
pyPDAF.PDAF.get_ensstats	This is a diagnotics function for LKNETF which returns the skewness and kutosis used there.
pyPDAF.PDAF.get_localfilter	This function returns whether a local filter is used.
pyPDAF.PDAF.get_memberid	This function returns the ensemble member id on the current process.
pyPDAF.PDAF.get_obsmemberid	This function returns the ensemble member id when observation operator is being applied.
pyPDAF.PDAF.get_smootherens	This function returns the smoothed ensemble in earlier time steps.
pyPDAF.PDAF.set_debug_flag	This function activates the debug output of the PDAF.
pyPDAF.PDAF.print_local_obsstats	This function print the observation statistics of local domain on screen.
pyPDAF.PDAF.print_domain_stats	This function make screen output of statistics of the local domains on current process.

Advanced manipulation

pyPDAF.PDAF.set_ens_pointer	This function returns the ensemble array in a numpy array where the internal array data has the same memoery address as PDAF ensemble array.
pyPDAF.PDAF.set_smootherens	This function can be used in the offline implementation when a smoother is used.
pyPDAF.PDAF.set_memberid	This function sets the ensemble member index to given value.
pyPDAF.PDAF.set_comm_pdaf	Setting the MPI communicator used by PDAF.
pyPDAF.PDAF.set_offline_mode	This function activates offline mode.
pyPDAF.PDAF.reset_forget	This function allows a user to reset the forgetting factor manually during the assimilation process.
pyPDAF.PDAF.init_local_obsstats	This function initialise the observation statistics of local domain.
pyPDAF.PDAF.incr_local_obsstats	This function update the observation statistics of local domain.
pyPDAF.PDAF.force_analysis	This function overwrite member index of the ensemble state by local_dim_ens (number of ensembles for current process, in full parallel setup, this is 1.) and the counter cnt_steps by nsteps-1.

Others

pyPDAF.PDAF.get_state	Distributing analysis state vector to an array.
pyPDAF.PDAF.assimilate_prepost	It is used to preprocess and postprocess of the ensemble.
pyPDAF.PDAF.prepost	This function does not perform any DA.
pyPDAF.PDAF.put_state_prepost	It is used to preprocess and postprocess of the ensemble.
pyPDAF.PDAF.gather_dim_obs_f	Gathers the dimension of observation vector across multiple local domains/filter processors.
pyPDAF.PDAF.gather_obs_f	In the local filters (LESKTF, LETKF, LSEIK, LNETF) this function returns the total observation vector from process-local observations.
pyPDAF.PDAF.gather_obs_f2	In the local filters (LESKTF, LETKF, LSEIK, LNETF) this function returns the full observation coordinates from process-local observation coordinates.
pyPDAF.PDAF.incremental	This is a helper function to apply analysis increment to model state in model forecast phase.
pyPDAF.PDAF.add_increment	This function directly adds analysis increment to given state vector without the need for user-supplied functions.
pyPDAF.PDAF.local_weight	The function is used for localisation in the analysis step of a filter and computes a weight according to the specified distance and the settings for the localising function.
pyPDAF.PDAF.local_weights	This function returns a vector of the localisation weights based on distance and localisation functions and radii.
pyPDAF.PDAF.gather_obs_f2_flex	In the local filters (LESKTF, LETKF, LSEIK, LNETF) this function returns the full observation coordinates from process-local observation coordinates.
pyPDAF.PDAF.gather_obs_f_flex	In the local filters (LESKTF, LETKF, LSEIK, LNETF) this function returns the total observation vector from process-local observations.

Internal matrix operations

pyPDAF.PDAF.seik_TtimesA	This is an internal function in PDAF where it perform matrix calculation of B = TA.
pyPDAF.PDAF.etkf_Tleft	This is an internal function in PDAF where it perform matrix calculation of B = TA.
pyPDAF.PDAF.estkf_OmegaA	This function is an internal function in PDAF.
pyPDAF.PDAF.enkf_omega	Generation of a random matrix with orthogonal basis following SEEK approach for EnKF with given properties.
pyPDAF.PDAF.seik_omega	Generation of a random matrix with orthogonal basis following SEIK approach.

Internal callback functions

pyPDAF.PDAF.omi_init_obs_f_cb	This function is an internal PDAF-OMI function that is used as a call-back function to initialise the observation vector.
pyPDAF.PDAF.omi_init_obsvar_cb	This function is an internal PDAF function that is used as a call-back function to initialise the observation error variance.
pyPDAF.PDAF.omi_g2l_obs_cb	This function is an internal PDAF-OMI function that is used as a call-back function to convert between global and local observation vectors in domain localisation.
pyPDAF.PDAF.omi_init_obs_l_cb	This function is an internal PDAF-OMI function that is used as a call-back function to initialise local observation vector in domain localisation.
pyPDAF.PDAF.omi_init_obsvar_l_cb	This function is an internal PDAF-OMI function that is used as a call-back function to initialise local observation vector in domain localisation.
pyPDAF.PDAF.omi_prodRinvA_l_cb	This function is an internal PDAF-OMI function that is used as a call-back function to perform the matrix multiplication inverse of local observation error covariance and a matrix A in domain localisation.
pyPDAF.PDAF.omi_likelihood_l_cb	This is an internal PDAF-OMI function that is used as a call-back function to compute the likelihood of the observation for a given ensemble member according to the observations used for the local analysis in the localized LNETF.
pyPDAF.PDAF.omi_prodRinvA_cb	This function is an internal PDAF-OMI function that is used as a call-back function to perform the matrix multiplication inverse of observation errro covariance and a matrix A.
pyPDAF.PDAF.omi_likelihood_cb	This is an internal PDAF-OMI function that is used as a call-back function to compute the likelihood of the observation for a given ensemble member according to the observations used for the local analysis for NETF or particle filter.
pyPDAF.PDAF.omi_add_obs_error_cb	This is an internal PDAF-OMI function that is used as a call-back function to add random observation error to stochastic EnKF.
pyPDAF.PDAF.omi_init_obscovar_cb	This is an internal PDAF-OMI function that is used as a call-back function to construct a full observation error covariance matrix used only in stochastic EnKF.
pyPDAF.PDAF.omi_init_obserr_f_cb	This is an internal PDAF-OMI function that is used as a call-back function to construct a full observation error covariance matrix used only in stochastic EnKF.
pyPDAF.PDAF.omi_prodRinvA_hyb_l_cb	This function is an internal PDAF-OMI function that is used as a call-back function to perform the matrix multiplication inverse of local observation error covariance and a matrix A in LKNETF.
pyPDAF.PDAF.omi_likelihood_hyb_l_cb	This is an internal PDAF-OMI function that is used as a call-back function to compute the likelihood of the observation for a given ensemble member according to the observations used for the local analysis in LKNETF.
pyPDAF.PDAF.omi_omit_by_inno_l_cb	The function is called during the analysis step on each local analysis domain.
pyPDAF.PDAF.omi_omit_by_inno_cb	The function is called during the analysis step of a global filter.
pyPDAF.PDAF.local_g2l_cb	Project a global to a local state vector for the localized filters.
pyPDAF.PDAF.local_l2g_cb	Initialise elements of a global state vector from a local state vector.