weatherbench2.config.Eval
- class weatherbench2.config.Eval(metrics, regions=None, evaluate_persistence=False, evaluate_climatology=False, evaluate_probabilistic_climatology=False, probabilistic_climatology_start_year=None, probabilistic_climatology_end_year=None, probabilistic_climatology_hour_interval=None, against_analysis=False, derived_variables=<factory>, temporal_mean=True, output_format='netcdf')
Evaluation configuration class.
- Parameters:
metrics (Dict[str, Metric]) –
regions (Optional[Dict[str, Union[Region, ExtraTropicalRegion, SliceRegion, LandRegion]]]) –
evaluate_persistence (Optional[bool]) –
evaluate_climatology (Optional[bool]) –
evaluate_probabilistic_climatology (Optional[bool]) –
probabilistic_climatology_start_year (Optional[int]) –
probabilistic_climatology_end_year (Optional[int]) –
probabilistic_climatology_hour_interval (Optional[int]) –
against_analysis (Optional[bool]) –
derived_variables (Dict[str, DerivedVariable]) –
temporal_mean (Optional[bool]) –
output_format (str) –
- metrics
Dictionary of Metric instances.
- Type:
Dict[str, weatherbench2.metrics.Metric]
- regions
Optional dictionary of Region instances.
- Type:
Optional[Dict[str, Union[weatherbench2.regions.Region, weatherbench2.regions.ExtraTropicalRegion, weatherbench2.regions.SliceRegion, weatherbench2.regions.LandRegion]]]
- evaluate_persistence
Evaluate persistence forecast, i.e. forecast at t=0.
- Type:
Optional[bool]
- evaluate_climatology
Evaluate climatology forecast.
- Type:
Optional[bool]
- evaluate_probabilistic_climatology
Evaluate probabilistic climatology, derived from using each year of the ground-truth dataset as a member.
- Type:
Optional[bool]
- probabilistic_climatology_start_year
First year of ground-truth to use for probabilistic climatology.
- Type:
Optional[int]
- probabilistic_climatology_end_year
Last year of ground-truth to use for probabilistic climatology.
- Type:
Optional[int]
- probabilistic_climatology_hour_interval
Hour interval to compute probabilistic climatology.
- Type:
Optional[int]
- against_analysis
Use forecast at t=0 as ground-truth. Warning: only for by-valid convention. For by-init, specify analysis dataset as obs.
- Type:
Optional[bool]
- derived_variables
dict of DerivedVariable instances to compute on the fly.
- Type:
- temporal_mean
Compute temporal mean (over time/init_time) for metrics.
- Type:
Optional[bool]
- output_format
whether to save to ‘netcdf’ or ‘zarr’.
- Type:
str
- __init__(metrics, regions=None, evaluate_persistence=False, evaluate_climatology=False, evaluate_probabilistic_climatology=False, probabilistic_climatology_start_year=None, probabilistic_climatology_end_year=None, probabilistic_climatology_hour_interval=None, against_analysis=False, derived_variables=<factory>, temporal_mean=True, output_format='netcdf')
- Parameters:
metrics (Dict[str, Metric]) –
regions (Optional[Dict[str, Union[Region, ExtraTropicalRegion, SliceRegion, LandRegion]]]) –
evaluate_persistence (Optional[bool]) –
evaluate_climatology (Optional[bool]) –
evaluate_probabilistic_climatology (Optional[bool]) –
probabilistic_climatology_start_year (Optional[int]) –
probabilistic_climatology_end_year (Optional[int]) –
probabilistic_climatology_hour_interval (Optional[int]) –
against_analysis (Optional[bool]) –
derived_variables (Dict[str, DerivedVariable]) –
temporal_mean (Optional[bool]) –
output_format (str) –
- Return type:
None
Methods
__init__(metrics[, regions, ...])Attributes