Add NOAA flare catalogue ingestion and ΔΦ(t) precursor-window evaluation

[Copilot]

Extends the GOES-18 analysis pipeline with a flare-prediction evaluation layer: ingests a NOAA flare catalogue (or auto-detects events from XRS peaks), evaluates the ΔΦ(t) instability operator in a 6–24 h precursor window, and computes ROC/AUC, lead-time distribution, and false-alarm/missed-event rates.

New functions

• shared/data_loader.load_noaa_flare_catalogue(path, *, start, end) — CSV/JSON ingestion with UTC normalisation, onset_time → time_begin fallback, and interval filtering.
• shared/event_evaluation.align_flare_onsets(flare_df, delta_phi_df) — nearest-sample alignment of flare onset times onto the ΔΦ(t) grid; returns aligned_time + delta_phi_at_onset.
• analysis/precursor_evaluation.evaluate_precursor_window(...) — per-flare extraction of [onset − 24 h, onset − 6 h) window, boolean threshold indicators, window stats, and full ROC/AUC/lead-time output via existing helpers.

New experiment script

experiments/eval_flare_catalogue.py — self-contained pipeline:

1. Reads noaa_goes18_xrs_1m.csv.zip; strips fill values (flux ≤ 0).
2. Computes ΔΦ(t) from log₁₀-normalised XRS long-wave flux (backward diff, Δt = 1 h).
3. Loads an external catalogue via --catalogue, or auto-detects C-class events from smoothed XRS local maxima with a 2 h minimum separation.
4. Runs evaluate_precursor_window over θ ∈ [0.01, 0.90] (50 points).
5. Saves results/flare_catalogue_eval.json, flare_catalogue_roc.png, flare_catalogue_lead_times.png, flare_catalogue_delta_phi.png.

# with auto-detected flares
python experiments/eval_flare_catalogue.py --months 1

# with a real NOAA catalogue
python experiments/eval_flare_catalogue.py --catalogue /path/to/goes_flares.csv --months 3

Results (January 2024, 67 auto-detected events)

Metric	Value
AUC	0.7875
TPR at θ = 0.19	0.881
FPR at θ = 0.19	0.138
FAR at θ = 0.46 (mid)	0.009
MER at θ = 0.46 (mid)	0.836
Median lead time	11.34 h

Documentation

Section 12 of ANALYSIS_AND_VALIDATION.md (§12.1–12.5) populated with catalogue ingestion description, precursor-window definition, ΔΦ(t) signal statistics, ROC/AUC table, lead-time distribution, and known limitations (circular evaluation, single channel, short interval).

Warning

Firewall rules blocked me from connecting to one or more addresses (expand for details)

I tried to connect to the following addresses, but was blocked by firewall rules:

• services.swpc.noaa.gov
  • Triggering command: /usr/bin/python python -m pytest test/ --ignore=test/runtests.jl --ignore=test/test_math_utils.jl --ignore=test/test_spiral_time.jl --ignore=test/test_energy_transfer.jl --ignore=test/test_topology.jl --ignore=test/test_release_events.jl -x --tb=short -q (dns block)

If you need me to access, download, or install something from one of these locations, you can either:

• Configure Actions setup steps to set up my environment, which run before the firewall is enabled
• Add the appropriate URLs or hosts to the custom allowlist in this repository's Copilot coding agent settings (admins only)

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.

[Copilot]

Pull request overview

This PR adds a flare-prediction evaluation layer to the GOES-18 analysis pipeline. It introduces NOAA flare catalogue ingestion, flare onset alignment onto the ΔΦ(t) grid, and a precursor-window evaluation function that computes ROC/AUC, lead-time distributions, and false-alarm/missed-event rates. A self-contained experiment script ties the pipeline together and produces JSON metrics and plots.

Changes:

• New shared utilities: load_noaa_flare_catalogue (CSV/JSON ingestion with UTC normalization and fallback column resolution) and align_flare_onsets (nearest-sample alignment of flare onsets onto the ΔΦ(t) timeline).
• New evaluate_precursor_window function in the analysis layer that extracts per-flare [onset−24h, onset−6h) windows, computes boolean threshold indicators, window stats, and delegates to existing ROC/AUC/lead-time helpers.
• New experiment script eval_flare_catalogue.py and expanded documentation in §12 of ANALYSIS_AND_VALIDATION.md.

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 6 comments.

Show a summary per file

File	Description
shared/data_loader.py	Adds load_noaa_flare_catalogue for CSV/JSON ingestion with UTC normalization, column fallback, and interval filtering.
shared/event_evaluation.py	Adds align_flare_onsets for nearest-sample alignment of flare onset times onto the ΔΦ(t) grid.
analysis/precursor_evaluation.py	Adds evaluate_precursor_window for per-flare precursor-window extraction and evaluation with threshold sweep.
experiments/eval_flare_catalogue.py	New self-contained pipeline script: loads XRS data, computes ΔΦ(t), detects/loads flares, evaluates precursor windows, and saves metrics/plots.
ANALYSIS_AND_VALIDATION.md	Populates §12.1–12.5 with detailed methodology, results tables, and limitations.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

You can also share your feedback on Copilot code review. Take the survey.