EBS pollock stock-assessment repository

This repository develops an assessment-year structure for eastern Bering Sea pollock stock assessments. The initial 2024 worked example shows how the nmfs-ost/region_spp_stock_assessment folder concept can be extended for an Alaska groundfish assessment while keeping annual data, runs, projections, review decisions, and deliverables traceable.

It describes the 2024 eastern Bering Sea pollock assessment using the committed state of the public assessment repository at commit 41dc244. Only the committed source snapshot was used to establish provenance; no uncommitted working-tree content from the source assessment was incorporated.

Repository status

This is currently a local development repository and has not been published. The 2024 content is a provenance and organization example, not an authoritative copy or independently reproduced assessment. Entries marked TO_CONFIRM, missing, or incomplete identify information that still requires an authoritative source.

The 2026 dashboard is generated directly from the annual CSV records and summarizes schedule, data readiness, bridge progress, open decisions, and deliverable status.

For local data entry, run:

Rscript apps/assessment-tracker/run.R

The local editor updates those CSV records, validates them before saving, and remains separate from the read-only GitHub Pages site.

What this example demonstrates

• an assessment-year directory rather than a year in the repository name;
• explicit data, model-run, review, and deliverable tracking records;
• separation of a bridge run, updated model, data-withholding runs, diagnostics, and projections;
• fixed links to the source repository and assessment commit;
• a validation script that checks the tracking records without requiring restricted data;
• a compact Quarto summary suitable for onboarding and annual release notes.

This is a provenance and organization example, not a redistributed assessment. Model inputs, executables, detailed results, and the full SAFE remain in the authoritative assessment repository.

Layout

R/                         reusable assessment functions belong here
data-raw/                  scripts and instructions that construct inputs
data/                      shareable data only; restricted data stay external
assessment-years/             conventions shared by all annual cycles
assessment-years/2024/
  assessment.yml           annual identity, source, software, and entry points
  records/
    data-sources.csv        provenance and access status for annual inputs
    model-runs.csv          role and status of each model configuration
    review-decisions.csv    author, Plan Team, and SSC decisions
    deliverables.csv        final documents and machine-readable products
index.qmd                  website homepage and compact annual summary
scripts/                   validation and workflow entry points
apps/assessment-tracker/   local or hosted editor for the 2026 tracking records
schemas/                   common columns and allowed status values

Quick check

Rscript scripts/validate_records.R
quarto render

The first command uses base R only. The report is deliberately static so the example can render without assessment packages or confidential inputs.

On main, GitHub Actions runs the same validation, renders the Quarto project, and deploys _site/ to GitHub Pages.

Reproducing the actual assessment

1. Clone the authoritative assessment repository.
2. Check out commit 41dc244ade54692282b1ac530b76757d4687fd37.
3. Restore the documented R and ADMB environment.
4. Verify access to the annual input products listed in assessment-years/2024/records/data-sources.csv.
5. Run or load Model 23 and its comparisons using R/pm24.R in the source repository.
6. Validate convergence, projections, tables, and figures before rendering the SAFE.

The source assessment reports a maximum gradient below 1e-3, with MCMC diagnostics of R-hat below 1.01 and effective sample size above 400 for the key parameters. These claims are recorded here as reported evidence, not rerun verification.