# ESSD Manuscript Scripts

This folder contains the scripts used to regenerate the ESSD manuscript
numbers, diagnostics, and figures from the released SeismicX-Cont files.

## Main Entry Points

- `reproduce_manuscript_outputs.sh` regenerates the ESSD inventory, independent
  coverage audit, consistency report, and core figures from released files.
- `audit_manuscript_numbers.py` regenerates the key inventory, annotation,
  and exact NSLC point-coverage values, including finite-sample checks for
  floating-point HDF5 arrays. Automatic-output diagnostics are excluded unless
  `--include-example-outputs` is supplied.
- `build_reference_arrivals.py` exports an independent SQLite arrival table with
  C0 point coverage and configuration-specific C1--C3 window, component,
  sampling-rate, gap, and response fields. Selection is deterministic,
  components are not mixed across locations or families, response intervals are
  half-open, and response-qualified C3 requires exactly one epoch match for each
  selected component.
- `audit_waveform_quality.py` checks every SQLite segment row for required
  fields, timing consistency, duplicate keys, exact-NSLC gaps, and overlaps,
  then compares index metadata with a deterministic stratified sample of HDF5
  datasets and short waveform windows.
- `check_manuscript_consistency.py` checks the regenerated inventory,
  annotation, and coverage values against the current manuscript.
- `plot_dataset_overview.py` regenerates `figures/dataset_overview.pdf`.
- `plot_monitoring_regime_characterization.py` regenerates the monitoring
  regime characterization figure and writes JSON/text summaries under
  `essd_scripts/outputs/`.
- `plot_essd_qc_figures.py` regenerates the workflow, station-day coverage, and
  independently computed label-coverage figures. Legacy example-output figures
  require `--include-example-output-figures`.
- `evaluate_picks.py` matches automatic picker JSONL outputs to reference labels
  one-to-one after P/Pg/Pn and S/Sg/Sn normalization. Its deterministic
  time-ordered assignment maximizes TP-tolerance matches before diagnostic-window
  matches and total absolute residual. `--reference-db` reuses the versioned
  C0--C3 flags without repeatedly reopening the HDF5 archive. Optional
  reference-window reporting intersects merged manual-reference windows with a
  selected output's station-time eligibility domain and reports both raw and
  effective processed-window duration; it is not conventional precision.
- `regenerate_multimodel_phase_matching.sh` regenerates the optional
  `data/validation/multimodel_phase_matching/` records and summaries from the
  stored source JSONL files. These outputs are automated agreement and file-
  interoperability diagnostics, not a standardized workflow comparison.
- `compare_associated_events.py` evaluates associated-event JSONL outputs
  against the reference catalog.

## Quick Checks

```bash
./essd_scripts/reproduce_manuscript_outputs.sh
python essd_scripts/audit_manuscript_numbers.py --format text
python essd_scripts/check_manuscript_consistency.py
python essd_scripts/audit_waveform_quality.py
python essd_scripts/build_reference_arrivals.py \
  --window-before-s 30 --window-after-s 30 \
  --required-components Z H1 H2 --require-response
python essd_scripts/plot_dataset_overview.py
python essd_scripts/plot_monitoring_regime_characterization.py
python essd_scripts/plot_essd_qc_figures.py
./essd_scripts/regenerate_multimodel_phase_matching.sh
```

The one-command script runs the manuscript-number audit first, exports JSON,
checks the current manuscript against regenerated numbers, then rebuilds the overview,
monitoring-regime characterization, and QC figures. Outputs are written to
`essd_scripts/outputs/` and `figures/`.

The plotting scripts write manuscript figures under `figures/`. The audit script
is read-only unless its output is redirected by the caller.

## Manuscript Coverage

The scripts cover the manuscript outputs as follows:

| Manuscript output | Reproduction source |
|---|---|
| Fig. 1, data-product workflow | `plot_essd_qc_figures.py`, using the released HDF5 directory, SQLite index, and annotation JSON for the displayed counts |
| Fig. 2, inventory and monitoring-regime overview | `plot_dataset_overview.py`, using the annotation JSON, SQLite index, and HDF5 waveform files |
| Fig. 3, waveform coverage matrix | `plot_essd_qc_figures.py`, using the SQLite waveform index |
| Fig. 4, monitoring-regime characterization | `plot_monitoring_regime_characterization.py`, using the annotation JSON and SQLite waveform index |
| Fig. 5, label coverage and provenance | `plot_essd_qc_figures.py`, using the annotation JSON, exact NSLC intervals in the SQLite index, and finite-sample checks in the HDF5 arrays |
| Phase-reference composition and C0--C3 eligibility table | `audit_manuscript_numbers.py --format latex-reference` and `check_manuscript_consistency.py` |
| Index timing, exact-NSLC gap/overlap, metadata agreement, and sampled waveform diagnostics | `audit_waveform_quality.py` |
| Optional multi-output phase-agreement and JSONL interoperability records | `regenerate_multimodel_phase_matching.sh` and `evaluate_picks.py` |
| Support-stratified and eligibility-normalized C0 manual-reference agreement | `scripts/validate_multi_output_candidates.py` |

Non-standardized automatic-output files are optional software-interoperability
fixtures and are not inputs to the manuscript's coverage or data-quality claims.