BayesianBackgroundCheck: check observation closeness to background, accounting for probability of gross error, i.e. that observation is bad. More...

#include <BayesianBackgroundCheck.h>

Inheritance diagram for ufo::BayesianBackgroundCheck:

[legend]

Collaboration diagram for ufo::BayesianBackgroundCheck:

[legend]

Public Types
typedef BayesianBackgroundCheckParameters	Parameters_

Public Member Functions
	BayesianBackgroundCheck (ioda::ObsSpace &obsdb, const Parameters_ &parameters, std::shared_ptr< ioda::ObsDataVector< int > > flags, std::shared_ptr< ioda::ObsDataVector< float > > obserr)

	~BayesianBackgroundCheck ()

Public Member Functions inherited from ufo::FilterBase
	FilterBase (ioda::ObsSpace &, const FilterParametersBaseWithAbstractAction &parameters, std::shared_ptr< ioda::ObsDataVector< int > >, std::shared_ptr< ioda::ObsDataVector< float > >)

	FilterBase (ioda::ObsSpace &, const eckit::Configuration &, std::shared_ptr< ioda::ObsDataVector< int > >, std::shared_ptr< ioda::ObsDataVector< float > >)

	~FilterBase ()

Public Member Functions inherited from ufo::ObsProcessorBase
	ObsProcessorBase (ioda::ObsSpace &, bool deferToPost, std::shared_ptr< ioda::ObsDataVector< int > >, std::shared_ptr< ioda::ObsDataVector< float > >)

	~ObsProcessorBase ()

void	preProcess () override

void	priorFilter (const GeoVaLs &) override

void	postFilter (const ioda::ObsVector &, const ioda::ObsVector &, const ObsDiagnostics &) override

oops::Variables	requiredVars () const override

oops::Variables	requiredHdiagnostics () const override

Static Public Member Functions
static const std::string	classname ()

Private Member Functions
void	print (std::ostream &) const override

void	applyFilter (const std::vector< bool > &apply, const Variables &filtervars, std::vector< std::vector< bool >> &flagged) const override
	Apply Bayesian background check filter. Return flagged=true for rejected obs. More...

int	qcFlag () const override
	Return bayesianQC flag for observations rejected by Bayesian BG check. More...

Variable	backgrErrVariable (const Variable &filterVariable) const
	Return the name of the variable containing the background error estimate of the specified filter variable. More...

template<class T >
std::vector< T >	reduceVector (const std::vector< T > &vector_full, const std::vector< size_t > &j_reduced) const
	Reduce a vector to only the elements for which j_reduced=true. More...

template<class T >
void	unreduceVector (const std::vector< T > &vector_reduced, std::vector< T > &vector_full, const std::vector< size_t > &j_reduced) const
	Copy a reduced vector back into the correct indices of the full vector. More...

Private Attributes
Parameters_	parameters_

Additional Inherited Members
Protected Attributes inherited from ufo::FilterBase
const eckit::LocalConfiguration	config_

ufo::Variables	filtervars_

Protected Attributes inherited from ufo::ObsProcessorBase
ioda::ObsSpace &	obsdb_

std::shared_ptr< ioda::ObsDataVector< int > >	flags_

std::shared_ptr< ioda::ObsDataVector< float > >	obserr_

ufo::Variables	allvars_

ObsFilterData	data_

Detailed Description

BayesianBackgroundCheck: check observation closeness to background, accounting for probability of gross error, i.e. that observation is bad.

Flag observations whose values differ from their model equivalents by too much. "Too much" depends on the uncertainties of both observation and model background, and the prior probability of the observation being "bad", or in gross error. Can apply to scalar or vector observation types on single-level. (Cannot handle profiles.) Calls function ufo::BayesianPGEUpdate.

Requires the following be specified in .yaml, under

obs filters:

filter: Bayesian Background Check
- prob density bad obs: uniform probability density that observation is "bad", i.e. in gross error;
- initial prob gross error: initial PGE value of the uniform probability distribution, before adjustment depending on obs-BG closeness;

May also specify the following optional parameters (defaults in ufo/utils/ProbabilityOfGrossErrorParameters.h):

PGE threshold: fail if after-check PGE exceeds this value;
obs minus BG threshold: fail if \(([y-H(x)]/{\sigma})^2\) exceeds this;
max exponent: maximum allowed value of the exponent in the 'good' probability distribution;
obs error multiplier: weight of observation error in combined error variance;
BG error multiplier: weight of background error in combined error variance.

Definition at line 75 of file BayesianBackgroundCheck.h.

Member Typedef Documentation

◆ Parameters_

typedef BayesianBackgroundCheckParameters ufo::BayesianBackgroundCheck::Parameters_

The type of parameters accepted by the constructor of this filter. This typedef is used by the FilterFactory.

Definition at line 80 of file BayesianBackgroundCheck.h.

Constructor & Destructor Documentation

◆ BayesianBackgroundCheck()

ufo::BayesianBackgroundCheck::BayesianBackgroundCheck	(	ioda::ObsSpace &	obsdb,
		const Parameters_ &	parameters,
		std::shared_ptr< ioda::ObsDataVector< int > >	flags,
		std::shared_ptr< ioda::ObsDataVector< float > >	obserr
	)

BayesianBackgroundCheck: check observation closeness to background, accounting for probability of gross error, i.e. that observation is bad.

Definition at line 34 of file BayesianBackgroundCheck.cc.

Here is the call graph for this function:

◆ ~BayesianBackgroundCheck()

ufo::BayesianBackgroundCheck::~BayesianBackgroundCheck ( )

Definition at line 51 of file BayesianBackgroundCheck.cc.

Member Function Documentation

◆ applyFilter()

void ufo::BayesianBackgroundCheck::applyFilter	(	const std::vector< bool > &	apply,
		const Variables &	filtervars,
		std::vector< std::vector< bool >> &	flagged
	)		const

overrideprivatevirtual

Apply Bayesian background check filter. Return flagged=true for rejected obs.

Apply the Bayesian background check filter.

Implements ufo::FilterBase.

Definition at line 93 of file BayesianBackgroundCheck.cc.

Here is the call graph for this function:

◆ backgrErrVariable()

Variable ufo::BayesianBackgroundCheck::backgrErrVariable ( const Variable & filterVariable ) const

private

Return the name of the variable containing the background error estimate of the specified filter variable.

Definition at line 59 of file BayesianBackgroundCheck.cc.

Here is the call graph for this function:

Here is the caller graph for this function:

◆ classname()

static const std::string ufo::BayesianBackgroundCheck::classname ( )

inlinestatic

Definition at line 82 of file BayesianBackgroundCheck.h.

◆ print()

void ufo::BayesianBackgroundCheck::print ( std::ostream & os ) const

overrideprivatevirtual

Implements ufo::FilterBase.

Definition at line 266 of file BayesianBackgroundCheck.cc.

◆ qcFlag()

int ufo::BayesianBackgroundCheck::qcFlag ( ) const

inlineoverrideprivatevirtual

Return bayesianQC flag for observations rejected by Bayesian BG check.

Implements ufo::FilterBase.

Definition at line 97 of file BayesianBackgroundCheck.h.

◆ reduceVector()

template<class T >

std::vector< T > ufo::BayesianBackgroundCheck::reduceVector	(	const std::vector< T > &	vector_full,
		const std::vector< size_t > &	j_reduced
	)		const

private

Reduce a vector to only the elements for which j_reduced=true.

Parameters

[in]	vector_full	full vector,
[in]	j_reduced	indices of vector_full to be copied into...

Returns: vector_reduced.

Definition at line 67 of file BayesianBackgroundCheck.cc.

Here is the caller graph for this function:

◆ unreduceVector()

template<class T >

void ufo::BayesianBackgroundCheck::unreduceVector	(	const std::vector< T > &	vector_reduced,
		std::vector< T > &	vector_full,
		const std::vector< size_t > &	j_reduced
	)		const

private

Copy a reduced vector back into the correct indices of the full vector.

Parameters

[in]	vector_reduced	reduced vector to restore to...
[in,out]	vector_full	full vector,
[in]	j_reduced	indices of vector_full to be populated.

Definition at line 81 of file BayesianBackgroundCheck.cc.

Here is the caller graph for this function:

Member Data Documentation

◆ parameters_

Parameters_ ufo::BayesianBackgroundCheck::parameters_

private

Definition at line 119 of file BayesianBackgroundCheck.h.

The documentation for this class was generated from the following files:

Public Types

Public Member Functions

Static Public Member Functions

Private Member Functions

Private Attributes

Additional Inherited Members

Detailed Description

Member Typedef Documentation

◆ Parameters_

Constructor & Destructor Documentation

◆ BayesianBackgroundCheck()

◆ ~BayesianBackgroundCheck()

Member Function Documentation

◆ applyFilter()

◆ backgrErrVariable()

◆ classname()

◆ print()

◆ qcFlag()

◆ reduceVector()

◆ unreduceVector()

Member Data Documentation

◆ parameters_