Neotoma pollen prerequisite screen¶
The summary and sample decisions preserve the
source-bound screen for the corrected relational snapshot. It reads the
authenticated raw Neotoma/EPD capture and the governed relational materialization.
The source snapshot and build IDs are unchanged; the relational manifest is
6f6777a40b67bd0704eb1a982f19e6d451cad0d6e4a41dfebf5e3f28be33105b.
The Ageröds Mosse candidate sample series
is a source-bound extraction of the first 16 retained samples at Neotoma site
12 (neotoma:sample:355236 through neotoma:sample:355251). Its 409 native
observations, 32 alternative age claims and exact source row locators reconcile
with the whole-snapshot sample decisions. The first four samples have no
sample-owned comparable age; the following 12 have at least one. All 16 remain
refused for numerical LRA because classification, chronology choice, parameter
binding and historical basin geometry have not been approved. The export retains
360 source-pollen-labelled integer count candidates without inventing a pollen
denominator or selecting among age alternatives. It is a bounded source-series
example, not a four-country LRA result.
The screen requires an agreeing pair of source pollen element labels, a finite
integer count and a sample-owned comparable typed age. It finds 10,637 necessary
candidate samples: Denmark 91, Finland 1,797, Norway 3,544, Sweden 5,011 and
unassigned 194. The five partitions sum to 12,388 captured samples, with 1,751
distinct refusals. The complete report retains source row locators and reasoned
sample decisions, including alternatives rather than silently choosing an age.
The chronology projection now follows Neotoma's rule of one default chronology per age type. The previous cross-age-type conflict classification suppressed valid count observations; the corrected materialization retains 32 chronology-control payload conflicts and one genuine same-age-type default conflict. This screen is a prerequisite, not an LRA input or an accepted taxon/PPE/basin/denominator decision. No LRA or measured-vegetation result is claimed.
Reproduce¶
Run the installed inspect_neotoma_lra_pollen_screen consumer against
data/neotoma/raw/neotoma_pollen_dataset_downloads and
data/neotoma/relational. The retained summary binds both exact manifests;
compare those identities and the candidate sample ID digest before reusing a
rerun. The Nordic direct-pollen result
is a separate observed-pollen study with its own age selection and held-out
country procedure.
To regenerate the sample series, call inspect_neotoma_lra_candidate_series
from bijux_pollenomics.analysis.lra.materialization.api with raw root
data/neotoma/raw/neotoma_pollen_dataset_downloads, normalized root
data/neotoma/relational, and the 16 explicit sample IDs above. Write
result.canonical_json without an added newline. Its SHA-256 is
77666e63575f5da0a82842d33d72b0527dc7a592691e8560e595b04d5cd69071.
To reproduce the compact summary from the retained complete sample decisions:
PYTHONDONTWRITEBYTECODE=1 PYTHONPATH=packages/bijux-pollenomics/src:packages/bijux-pollenomics-dev/src \
.venv/bin/python -m bijux_pollenomics_dev.source_completion.neotoma_screen_summary \
--repository . \
--screen docs/report/source-relations/neotoma/pollen-screen/report.json \
--output artifacts/jobs/neotoma-screen-summary/summary.json
Use an absent artifact output and compare its bytes with summary.json.
This exporter checks the candidate digest, sample decisions, country totals and
refusal scope. It preserves the parent report values; it does not revalidate raw
source custody or approve any scientific prerequisite.