Analysis (rojak.turbulence.analysis)

Post-processing analyses for turbulence diagnostics

This module implements PostProcessor subclasses that turn raw turbulence diagnostic values (as computed by rojak.turbulence.diagnostic) into forecasts and skill statistics, namely:

• Threshold-based turbulence detection: computing severity thresholds from percentiles of a calibration dataset (TurbulenceIntensityThresholds), applying those thresholds to obtain boolean turbulent regions (TurbulentRegionFromThreshold, TurbulentRegionsBySeverity), and the resulting probability of encountering each severity (TurbulenceProbabilityBySeverity).

• Mapping raw diagnostic values onto the Eddy Dissipation Rate (EDR) scale, using the log-normal distribution of the diagnostic on a calibration dataset (DiagnosticHistogramDistribution, ComputeDistributionParametersForEDR, TransformToEDR).

• Correlation between turbulence diagnostics, either globally (CorrelationBetweenDiagnostics, MatthewsCorrelationOnDataset, MatthewsCorrelationOnThresholdedDiagnostics) or stratified by hemisphere and latitudinal region (LatitudinalCorrelationBetweenDiagnostics).

• Association measures (RelationshipBetween and its subclasses) between an arbitrary binary feature and turbulence diagnostics, dispatched via RelationshipBetweenFactory and applied across every diagnostic in a dataset via RelationshipBetweenXAndTurbulence.

Classes

ComputeDistributionParametersForEDR(...)	Computes the mean and variance of the log-normal distribution of a turbulence diagnostic's values
CorrelationBetweenDiagnostics(...)	Computes the correlation between turbulence diagnostics
DiagnosticHistogramDistribution(...[, ...])	Computes histogram bins for log-normal distribution of turbulence diagnostics.
Hemisphere(*values)	Hemisphere to restrict a latitudinal region to, used by LatitudinalCorrelationBetweenDiagnostics
HistogramData(hist_values, bins, mean, variance)	Histogram of the log-normal distribution of a turbulence diagnostic's values
JaccardIndex(this_feature, other_feature[, ...])	Jaccard index (see rojak.turbulence.metrics.jaccard_index_multidim()) between two boolean features
LatitudinalCorrelationBetweenDiagnostics(...)	Computes the correlation between turbulence diagnostics by latitudinal region
LatitudinalRegion(*values)	Latitudinal band to restrict data to, used by LatitudinalCorrelationBetweenDiagnostics
MatthewsCorrelation(this_feature, other_feature)	Matthew's Correlation Coefficient (see rojak.turbulence.metrics.matthews_corr_coeff_multidim())
MatthewsCorrelationOnDataset(is_dataset, ...)	Computes the Matthew's Correlation Coefficient between every pair of boolean DataArrays within a Dataset
MatthewsCorrelationOnThresholdedDiagnostics(...)	Computes the Matthew's Correlation Coefficient between diagnostics, thresholded at each turbulence severity
ProbabilityThisGivenNotOther(this_feature, ...)	Conditional probability \(P(\text{this\_feature} | \lnot \text{other\_feature})\)
ProbabilityThisGivenOther(this_feature, ...)	Conditional probability \(P(\text{this\_feature} | \text{other\_feature})\)
RelationshipBetween(this_feature, other_feature)	Abstract base class for computing an association measure between two boolean features
RelationshipBetweenAlphaVelAndTurbulence(...)	RelationshipBetweenXAndTurbulence specialised to use jet stream regions identified from an AlphaVelField as the other_feature
RelationshipBetweenFactory(this_feature, ...)	Factory for constructing the RelationshipBetween subclass corresponding to a RelationshipBetweenTypes
RelationshipBetweenXAndTurbulence(...[, ...])	Computes the association between a feature and every turbulence diagnostic in a Dataset
RelativeRisk(this_feature, other_feature[, ...])	Relative risk (see rojak.turbulence.metrics.relative_risk()) between two boolean features
SampleOddsRatio(this_feature, other_feature)	Sample odds ratio (see rojak.turbulence.metrics.sample_odds_ratio()) between two boolean features
TransformToEDR(computed_diagnostic[, mean, ...])	Transforms turbulence diagnostic values into EDR
TurbulenceIntensityThresholds(...)	Computes threshold diagnostic value for each turbulence intensity using percentiles
TurbulenceProbabilityBySeverity(...)	Computes probability of encountering turbulence of each severity for a given turbulence diagnostic
TurbulentRegionFromThreshold(...)	Computes a boolean mask of where a turbulence diagnostic falls within a given severity's threshold bounds
TurbulentRegionsBySeverity(...[, has_parent])	Computes turbulent regions by severity for a given turbulence diagnostic