openghg_inversions.postprocessing.sigma#

Reconstruct observation-aligned sigma values from inversion output.

openghg_inversions.postprocessing.sigma.reconstruct_sigma_aligned(trace: InferenceData, *, model_data: Dataset | None = None) → DataArray#

Reconstruct posterior sigma values on the observation dimension.

Parameters:

  • trace – Inference data containing posterior sigma and, unless model_data is supplied, registered constant model data.

  • model_data – Optional canonical sigma alignment data.

Returns:

Posterior sigma values indexed onto nmeasure.

Raises:

  • AttributeError – If required inference-data groups are absent.

  • KeyError – If posterior sigma or registered alignment data are absent.

  • ValueError – If registered alignment data are invalid.