EvaluationStage#

class rojak.orchestrator.turbulence.EvaluationStage(calibration_result, phases_config, domain, output_dir, plots_dir, name, start_time, image_format)[source]#

Runs the turbulence evaluation stage

For each configured TurbulenceEvaluationPhaseOption, computes the corresponding result (probabilities, EDR, turbulent regions, or diagnostic correlation) from an evaluation dataset, using the thresholds/distribution parameters produced by the calibration stage where needed, and produces the corresponding plot(s). See run_phase().

Parameters:
__init__(calibration_result, phases_config, domain, output_dir, plots_dir, name, start_time, image_format)[source]#
Parameters:

  • calibration_result (Mapping[TurbulenceCalibrationPhaseOption, Result]) – Outcome of CalibrationStage.launch(), providing thresholds/distribution parameters to evaluate with

  • phases_config (TurbulenceEvaluationPhases) – Which evaluation phases to run, and their configuration

  • domain (SpatialDomain) – Spatial (and optionally vertical) domain to restrict evaluation data to

  • output_dir (Path) – Base directory to export zarr results into (a name subdirectory is created)

  • plots_dir (Path) – Base directory to save plots into (a name subdirectory is created)

  • name (RunName) – Name of this run, used to namespace exported files

  • start_time (TimeStr) – Timestamp identifying this run

  • image_format (str) – File format (e.g. "png") to save plots as

Return type:

None

Methods

__init__(calibration_result, phases_config, ...)

create_diagnostic_suite(diagnostics, chunks)

Load the evaluation data, compute diagnostics from it, and attach calibration results

launch(diagnostics, chunks)

Run every configured evaluation phase

run_phase(phase, suite)

Run a single evaluation phase against suite, producing the corresponding plot(s) as a side effect