LandClim II author and published parameter correspondence

The 31-row comparison pairs the exact author REVEALS profile with the PANGAEA LandClim II PPE/FSP source table. Each author target has exactly one published row with the same numerical fall speed and an RPP mean within 0.00050001 of the displayed mean. This pairing also resolves display-name differences such as author Picea versus published Picea abies, and author Quercus deciduous versus published Quercus deciduous t.. It compares source versions; it does not silently rename native pollen observations.

The published RPP standard deviations agree with the square roots of the author covariance diagonal to within 0.00050001 for 28 of 31 targets. The other three exceed what three-decimal display rounding alone explains:

Target	Author SD from variance	Published SD	Author minus published
Cerealia-t	0.380949	0.380	+0.000949
Cyperaceae	0.050621	0.050	+0.000621
Phillyrea	0.075000	0.076	−0.001000

These small differences may reflect source revisions or another display rule; the retained sources do not establish which. The author covariance values remain unchanged for reproducibility, and the published SD values remain unchanged as comparison targets. The report retains every exact source string, all 31 differences, footnoted published fall-speed markers and input hashes. It admits zero Swedish LRA samples: source parameter correspondence does not resolve the native denominator or historical basin geometry.

Replay from the repository root:

PYTHONDONTWRITEBYTECODE=1 .venv/bin/python -m \
  bijux_pollenomics_dev.source_completion.landclim_parameter_correspondence \
  --repository . \
  --output artifacts/jobs/landclim-parameter-correspondence/replay

Compare the replayed report.json byte-for-byte with this directory.