Goal
Extend multi-sector RHIME to linked primary-species/tracer inversions using shared sector states and
tracer-specific sensitivities.
Tasks
- Prepare canonical inputs with explicit species/tracer and sector structure.
- Create each sector latent once when the science requires shared degrees of freedom.
- Map sector latents into tracer-specific
mu contributions through sensitivity matrices and
conversion factors.
- Add tracer-specific total modelled observations.
- Add tracer-specific likelihoods, errors, offsets, and boundary-condition handling where needed.
Acceptance criteria
- A synthetic CO2/O2 fixture samples successfully.
- One sector latent appears once but contributes to both CO2 and O2 modelled-observation vectors.
- Latent dimensionality is lower than independent CO2 and O2 sector inversions.
Goal
Extend multi-sector RHIME to linked primary-species/tracer inversions using shared sector states and
tracer-specific sensitivities.
Tasks
mucontributions through sensitivity matrices andconversion factors.
Acceptance criteria