Satcom, flight models and drift, drawn from your own files. Where a source is missing or provisional, the page says so. The satellite and flight chapter closed on 6 October 2026 and was partly reopened on 7 October: the arc north of 27.5°S turns out to lie outside the model’s weather data, so northern endpoints were never testable. The drift branch is still open.
5,029 BTO/BFO rows from the Inmarsat log, the canonical 19-row event table and the published satellite states.
Average Day flights re-flown from your package, the High-Fidelity Average Day Solver (TASK_021), the TASK_022 terminal experiment, 97 Phase 5A candidates, BTO-only survivors.
Arc, candidates, search areas and the reverse-drift density on one switchable map.
17-month currents map, reverse-drift density, currents-only forward test, Tasks 7 to 9 (Task 9 complete).
What the satcom and flight branch established, what it rests on, and an appendix of readings that were wrong.
Task 10 results, drifter hindcast, forward release grid, permutation null, the B1 stop, the provenance audit, and why no source region can be determined.
| Model | Status | What this atlas shows | What is missing here |
|---|---|---|---|
| Satellite BTO/BFO | frozen | Raw burst log; June manifest facts (bias, calibration residual) | Frozen v06 scorer outputs are not in the folders I could read |
| Average Day flight model | frozen | 3 reference flights re-flown from your 3 Oct package (verify.py passes), TASK_021 solver, TASK_022 terminal experiment, TASK_026 convention sweep, TASK_029 reachability | Score stops at 00:11; no endpoint by design. Chapter closed 6 Oct — see the closing report |
| Other endpoint work | provisional | TASK_021, TASK_022, the TASK_026 convention sweep and the TASK_030 arc-1 sweep, on the Flight models page | No final coordinate exists in any source. Northern arc-1 starts score ten times better than C120's own start under a grid 145× too coarse to represent C120 — a positive indication that cannot be tested. The arc north of 27.5°S is outside the ERA5 domain and outside every search box in the manifest |
| Ocean currents | data | GLORYS12V1 daily surface currents, 17 months | Not a drift result by itself |
| Debris drift | assumption-dependent | Baseline density; Tasks 7 and 8 results | Tasks 9–16: no source region is determinable. The method recovers known synthetic sources, but on the real finds the answer moves 1,000–1,900 km when an unmeasured assumption changes. B1 found no supported bias correction; the 1.2% slip floor was fitted to forcing products B0 does not use |
| Data | File | Snapshot |
|---|---|---|
| Canonical events, satellite state, Average Day flights | MH370_AverageDay_algorithm_2026-10-03.zip (data/, examples/, re-flown here) | 3 Oct 2026 |
| Forward currents-only test | reverse_drift_forward_summary.csv | Mar 2014 test |
| Burst log | satcom_observations.csv (WS-004 canon) | 13 Jul 2026 recovery |
| Event table, survivors | satcom_events.csv, candidate_survivors_top.csv (WS-002) | 11 Jun 2026 restart |
| Candidates | MH370_PHASE5A_CANDIDATE_STATUS.csv (WS-008) | 2 Aug 2026 |
| 7th arc | reference_arc.json (Codex 2026-10-03), 40,000 ft reference | 3 Oct 2026 |
| Radar, search areas | radar_track_points.csv, search_history_manifest.csv (WS-004) | 13 Jul 2026 |
| Drift density, Tasks 7 to 9 (Task 9 complete) | Codex outputs, diagnostics_v2 | 3–5 Oct 2026 |
| Currents | GLORYS12V1 via Copernicus Marine; NCEP/NCAR winds | Mar 2014 – Jul 2015 |
archive/source_snapshots.