Listening for events…

Editorial Development Plan

Living index of every open TerraPulse GitHub issue, with our intended next move for each. Add a row whenever a new issue is filed; update the status and plan as direction changes.

Last updated: 2026-05-26 (initial sweep of 66 open issues).

Status vocabulary (one word per Mike's request):

  • queued — actively planned; we intend to work on it.
  • research — paper idea sitting in the backlog waiting to be picked.
  • blocked — useful, but a prerequisite (data, infra, or upstream paper) must land first.
  • parked — backlog; valuable but not on the near-term path.
  • standing — recurring operational issue (audits, signups, bios), not a one-shot fix.
# Status Plan
#218 blocked Decouple ingestion from API process. The #213 in-memory dedup cache is the patch; this is the proper structural fix. Needs an architecture brainstorm (separate worker process? message queue?) before any code. Monitor for ticker//pulse/recent slowness recurrence; if it reappears, this jumps to queued.
#207 research NWS-alerts climatology PMA — diurnal, day-of-week, seasonal patterns across 77 alert types and 713K records. Tests whether peaks reflect real weather vs. NWS office workflow. Run as a PMA once a current paper finishes.
#206 research North magnetic pole drift acceleration (1590–2024) PMA on the 8,374-position catalog. On the headline-friendly list ("Earth's compass is sprinting"); strong candidate after the fireball paper.
#204 blocked IGRA × SPC tornado lead-time PMA without the WSPR window constraint. Blocked by IGRA history: IGRA in PG only spans 2024-12-31 → 2026-05-24, SPC tornado ends 2023-12-19, zero overlap. Needs the IGRA 1990–2023 backfill (docs/superpowers/plans/2026-05-26-sounding-indices-derive.md) before this can run.
#203 research AQI anomaly characterization across 10 global cities — do >3σ AQI spike days separate into archetypes (PM-dominant wildfire, NOₓ inversion, O₃ photochemical)? Strong PMA candidate; queue after current paper.
#202 research Safecast radiation vs Forbush decreases — V2 of /articles/solar-watch-april-2026. Tests whether ground radiation depresses during cosmic-ray drops. Run after the fireball paper.
#201 queued Bz → Dst predictive lead-time PMA. R0 + R1 shipped today (commits e5ba020, 7b391c2, 63f081a). H1 confirmed, H2 graduated under sensitivity. Issue stays open until Mike does final editor sign-off.
#200 parked WSPR corridor SNR vs continuous Bz — V3 of #132. #132 V1/V2 both returned underpowered nulls; not obvious that a continuous-driver reframe rescues it. Park until a new angle or more storms accumulate.
#199 research M6.6 Sanriku 2026-05-15 foreshock signature + Bath's law test. Timely (recent quake), clean scope, fast PMA. Strong queue candidate.
#196 queued observations autovacuum never completes — 14.66% dead tuples, last_vacuum is NULL. Critical infra hygiene. Plan: tune autovacuum thresholds for this table (autovacuum_vacuum_scale_factor), schedule a one-shot manual VACUUM ANALYZE during low-traffic window, verify completion.
#195 queued 14 GB idx_observations_location index appears orphaned (0 scans, not in \d observations). Plan: confirm it's a leftover from a dropped column or partition swap, then DROP INDEX CONCURRENTLY. Frees substantial disk.
#194 queued castnet-outgoing-data fetcher takes 6–9 h per fetch — completes successfully, just absurdly slow. Plan: profile the loader (likely a giant unpaginated CSV pull), add streaming/pagination, target sub-30 min.
#185 standing Weekly Audit 2026-W20. Closes automatically Friday 23:59 UTC; a fresh weekly-audit issue opens the following Monday. No specific plan — operational standing issue.
#132 parked WSPR corridor SNR superposed-epoch over Kp events. V1/V2 underpowered (N=26). Park until storm sample grows, or pivot to the continuous-driver framing (#200).
#120 research WSPR aircraft detection — prior question for the Godfrey MH370 hypothesis: can WSPR detect commercial aircraft at all against ADS-B ground truth? Clean falsifiable question; PMA candidate.
#111 parked Realtime alert classifier — auto-flag interesting events (M6+ quakes, X-class flares, Kp≥6) from the live stream and broadcast via WebSocket. Product feature, not a paper. Defer behind paper queue.
#110 parked Migrate to NASA FIRMS active-fire data (resolves #48 drought-fire-AQI null). Useful but requires NASA EarthData API key signup and a new fetcher. Defer behind paper queue.
#108 parked Auto-generate workspace cross-citations from results.json. Quality-of-life infra; current manual citations work.
#107 parked Knowledge graph 2.0 — domain clusters, time slider, search, path finder on /garden. Big lift; current graph works.
#106 parked /lab page redesign — thumbnails grid, filter chips, sort options. Site UX improvement; current cards-with-Load-More works.
#105 research Starlink reentry watch — track decay events from CelesTrak TLEs. PMA candidate; clean scope, timely.
#103 queued Automated Forbush-decrease catalog from NMDB neutron monitors. Builds on the April-2026 solar-watch detection. Infra: detection script + new forbush_decrease metric. Enables #202 directly.
#102 research WSPR + LWA cross-validation at overlapping HF frequencies. Independent monitors → strong methodology. PMA candidate.
#99 standing Brad needs to add his bio/photo to /team. Operational ping; no plan beyond reminder.
#96 parked LWA radio telescope integration — fetcher is deployed but the analysis pipeline isn't built. Unblocks #102.
#94 standing Sign up for Space-Track.org + AMS API keys. Action-item issue; closes when credentials are added to .env. AMS key is now particularly relevant for the fireball paper.
#91 blocked WSPR transient deep-dive — characterize the 66 unexplained events from #90's detector. Blocked on the denoising pipeline (#86) and the trend re-eval (#87) so events are scored against clean data.
#89 blocked WSPR FFT re-analysis on denoised data — does the 27-day solar rotation emerge once noise is corrected? Blocked on #86 denoising pipeline.
#88 blocked WSPR storm detection — separate propagation drop from noise rise during Kp≥5. Blocked on #86 denoising pipeline.
#87 blocked WSPR secular trend re-evaluation — is the −0.1 dB/year decline real or a network-composition artifact? Blocked on #86 denoising pipeline.
#86 queued WSPR denoising pipeline. Foundational infra that unblocks #87, #88, #89, #91. Build scripts/wspr_denoise.py joining wsprdaemon noise data with corridor + daily aggregates; produce three correction levels.
#85 parked Real-time propagation dashboard at /propagation. Product feature; defer behind paper queue.
#84 parked NWS severe weather dashboard at /severe-weather. Product feature; defer behind paper queue.
#81 parked WSPR corridor pipeline — schedule hourly + backfill ±3 days around 788 DONKI storms. Useful but #132 superposed-epoch nulls suggest more events alone won't rescue the analysis.
#78 research Sentinel-1 SAR scene-catalog coverage patterns — do acquisitions track natural disasters? 8,803 scenes available. PMA candidate.
#76 research DONKI catalog cascade analysis — radiation belt, storm Kp, high-speed streams as cascade triggers. 788 + 544 + 477 events. PMA candidate, overlaps with #68.
#75 research SPC convective outlook verification — Day 1/2/3 forecasts vs observed NWS severe weather. Sample is small (152 records per Day) but grows daily. PMA candidate.
#71 research São Paulo AQI 150–180+ multi-domain attribution. Drought × inversion × emissions. Compelling case-study PMA.
#69 research LIGO/Virgo duty cycle vs space weather — do geomagnetic storms degrade GW-detector sensitivity? Speculative but very headline-friendly. PMA candidate.
#68 research Full Sun-to-ground cascade timing PMA — CME → L1 (DSCOVR) → Kp → ionosphere (WSPR) propagation delays. Overlaps with #76 and #201. PMA candidate.
#60 parked SBDB WATCH_LIST config improvements — move from hardcoded list to admin UI / auto-add objects approaching < 0.5 LD. Quality-of-life infra.
#59 parked "Heavy Elements" — onboard bespoke EPA/data.gov long-duration datasets (GLENDA, Narragansett 182-year benthic, etc.) as ground-truth catalysts. Big strategic value but slow ROI; surface when one becomes specifically needed by a paper.
#54 blocked Scaling Laws Phase 3 — spatial clustering + fractal dimension. Blocked on Phase 1 (magnitude) and Phase 2 (#53 inter-event times) being stable.
#53 parked Scaling Laws Phase 2 — inter-event time distributions, Omori-law aftershock decay, Weibull fits. Required for the "Probability Engine." Park until Phase 1 results justify the next phase.
#52 parked Cross-domain transfer entropy — nonlinear Granger. Method upgrade for prior null results. Worth doing once we have a specific pair worth re-testing.
#51 parked Nightly anomaly-detection engine — z-score all metrics, flag 3+ metric >2σ days, classify compound event type, post to /lab/anomalies. Product feature; defer behind papers.
#50 parked Global heatmap dashboard at /map — multi-layer real-time visualization (Leaflet/Mapbox + WebSocket). Product feature; defer behind papers.
#43 research Tidal forcing on volcanic eruptions — Schuster test on 5,000 Smithsonian eruptions. Clean scope, classic question, fast PMA. Candidate.
#42 research Methane × earthquakes — 42-yr monthly methane vs earthquake catalog. PMA candidate.
#39 parked Add FAA OPSNET / BTS flight-delay data. Natural-experiment dataset; not blocking anything specific. Onboard when a paper requires it.
#38 parked Add power-grid frequency data. Onboard when a paper requires it (e.g., a 1989-Quebec-style transformer paper).
#37 parked Add eBird + NOAA marine-mammal stranding data — biological signals. Onboard when a paper requires it.
#36 parked Add Google Trends API — crowd-sourced event detection. Useful as a validation layer; onboard when a paper requires it.
#35 parked Add NASA TROPOMI/OMI satellite SO₂ — volcanic-emission proxy. Onboard when a volcanic-emissions paper requires it.
#33 parked Add Schumann resonance monitoring (7.83 Hz cavity). Connects lightning, solar storms, possibly earthquakes. Onboard when a paper requires it.
#31 parked Add NOAA OISST sea-surface temperature. Enables ENSO/MHW/hurricane-fuel papers. Onboard when one of those papers gets selected.
#27 parked Index and search NWS alert types (46+ categories). Largely subsumed by #207's climatology paper, which will produce the taxonomy as a side effect. Close once #207 ships.
#26 queued Backfill solar flux F10.7 from NCEI / Penticton archives — current data is 7-day rolling window only. Small infra job; useful for any solar paper.
#25 queued Backfill USGS earthquakes to early 1900s for multi-cycle solar overlap. Enables real-power solar-cycle × earthquake tests. Infra: USGS FDSN historical pull + idempotent loader.
#24 parked Searchable compound-event catalog (3+ metrics simultaneously z>2). Overlaps with #51 anomaly engine.
#23 research Volcano-earthquake precursor — do swarms predict eruptions? Classic volcanology Q with our dataset. PMA candidate; pairs naturally with the Campi Flegrei caldera coverage.
#17 research Radiation anomaly mapping — extend the Safecast baseline workspace into a global anomaly map keyed to nuclear facilities + geology. PMA candidate.
#16 research Hydrology deep-dive — USGS streamflow vs Open-Meteo precip lag analysis. Watershed response times mapped spatially. PMA candidate.
#11 parked Knowledge-gap analysis on the embedding index. Meta-research tool; useful but not on the current path.
#8 parked Paper-Data Alignment — replicate specific claims from indexed arXiv papers against our data. Meta-research workflow; defer behind primary papers.
#4 parked Data visualizations + normalized query API + NL→query interface. Major product feature; defer until site/monetization milestone forces it.

Maintenance protocol

  1. New issue filed — add a row at the top of the table (or in natural numerical order) with a status and a plan.
  2. Status change — update in place; don't churn the row.
  3. Issue closed — remove the row from this file (closed issues live in GitHub).
  4. Plan column is the editorial take — what we'd do and why — not just the issue title. Keep it to a few sentences.
  5. Re-sweep the table quarterly to catch drift.
Live Feed