Qa Thresholds¶

Module: spectralbridge.qa.thresholds.

Configurable QA thresholds.

Defaults are intentionally conservative and provisional until they are tuned against a pinned set of real NEON, MicaSense, and Landsat comparison data.

API glossary · Workflow / scientific context

Signatures and documentation below are collected from the current source. Advanced objects are not newly declared stable by inclusion here.

Audience: public. Also exported as spectralbridge.qa.QAThresholds.

QAThresholds `dataclass` ¶

QAThresholds(valid_fraction_warn: float = 0.9, valid_fraction_fail: float = 0.7, negative_fraction_warn: float = 0.01, negative_fraction_fail: float = 0.05, overbright_fraction_warn: float = 0.01, overbright_fraction_fail: float = 0.05, correction_abs_q99_warn: float = 0.2, correction_abs_q99_fail: float = 0.5, seam_score_warn: float = 1.5, seam_score_fail: float = 2.5, chunk_difference_tolerance: float = 1e-06, srf_coverage_warn: float = 0.98, srf_coverage_fail: float = 0.9, provisional: bool = True)

Provisional standard-mode thresholds expressed as fractions.

Attributes:

Name	Type	Description
`valid_fraction_warn`	`float, default=0.9`	Fractional support/range review threshold in [0, 1]; warn/fail interpretation follows the corresponding metric, not a universal scientific validity rule.
`valid_fraction_fail`	`float, default=0.7`	Fractional support/range review threshold in [0, 1]; warn/fail interpretation follows the corresponding metric, not a universal scientific validity rule.
`negative_fraction_warn`	`float, default=0.01`	Fractional support/range review threshold in [0, 1]; warn/fail interpretation follows the corresponding metric, not a universal scientific validity rule.
`negative_fraction_fail`	`float, default=0.05`	Fractional support/range review threshold in [0, 1]; warn/fail interpretation follows the corresponding metric, not a universal scientific validity rule.
`overbright_fraction_warn`	`float, default=0.01`	Fractional support/range review threshold in [0, 1]; warn/fail interpretation follows the corresponding metric, not a universal scientific validity rule.
`overbright_fraction_fail`	`float, default=0.05`	Fractional support/range review threshold in [0, 1]; warn/fail interpretation follows the corresponding metric, not a universal scientific validity rule.
`correction_abs_q99_warn`	`float, default=0.2`	Review threshold for the 99th percentile of absolute reflectance correction, in the diagnostic reflectance units.
`correction_abs_q99_fail`	`float, default=0.5`	Review threshold for the 99th percentile of absolute reflectance correction, in the diagnostic reflectance units.
`seam_score_warn`	`float, default=1.5`	Dimensionless seam/interior gradient-ratio review threshold; landscape structure can also exceed it.
`seam_score_fail`	`float, default=2.5`	Dimensionless seam/interior gradient-ratio review threshold; landscape structure can also exceed it.
`chunk_difference_tolerance`	`float, default=1e-06`	Absolute same-unit tolerance for chunk-invariance comparisons.
`srf_coverage_warn`	`float, default=0.98`	Fractional support/range review threshold in [0, 1]; warn/fail interpretation follows the corresponding metric, not a universal scientific validity rule.
`srf_coverage_fail`	`float, default=0.9`	Fractional support/range review threshold in [0, 1]; warn/fail interpretation follows the corresponding metric, not a universal scientific validity rule.
`provisional`	`bool, default=True`	Whether these configured review bounds are explicitly provisional rather than empirically calibrated acceptance criteria.

to_dict ¶

to_dict() -> dict[str, float | bool]

Serialize this record to its established machine-readable mapping.

Returns:

Type	Description
`dict[str, float \| bool]`	Mapping of record fields and workflow-derived summary entries in the established serialization schema.

Audience: advanced.

classify_high_bad ¶

classify_high_bad(check_id: str, value: float | None, *, warn: float, fail: float, units: str | None = None, interpretation: str, provisional: bool = True, reason: str | None = None) -> QACheck

Classify a metric against inclusive upper warn/fail bounds; None yields NOT EVALUATED.

Audience: advanced.

classify_low_bad ¶

classify_low_bad(check_id: str, value: float | None, *, warn: float, fail: float, units: str | None = None, interpretation: str, provisional: bool = True, reason: str | None = None) -> QACheck

Classify a metric against inclusive lower warn/fail bounds; None yields NOT EVALUATED.

Explicitly exported constants and type aliases¶

These values describe package metadata, schemas or configuration; they are not observations from a production dataset.