# Closing interpretation — what the year of ocean actually established

Answers to the seven closure questions, from the completed 365-day integrations at 6 h and 3 h, 481 objects, daily GLORYS12V1 surface currents plus 1.2% of the monthly-mean NCEP wind, physics frozen throughout.

## 1. Did the southern cartwheel occur?

**The motion occurred routinely. The transport it was supposed to produce did not.**

In the 35.5–38.7° S band, 57 of 240 objects at 6 h and 68 at 3 h travelled south of their release latitude and then ran west; 24–29 sustained a westward run of at least two months and five degrees of longitude, the largest covering 36° over 172 days. Forty per cent revisited a position they had held a month or more earlier, which is genuine eddy residence rather than mere wandering.

Four objects at 6 h and six at 3 h reached any western gate.

The previous report's claim that the motion "occurred zero times in 240" was an endpoint flag misread as a chronological test, and it was wrong. The corrected statement is narrower and holds under refinement: **southward excursion followed by westward motion is a normal behaviour of this ocean; it reverses, and the eastward mean flow reasserts itself.**

The sharpest detail is in the opposite direction from the hypothesis. Of the thirteen objects in the full ensemble that crossed 80° E at 6 h, nine never went 0.5° south of their release latitude, and **all four of the band's gate-crossers had a southward excursion of exactly zero.** Among the objects that reached the west, going south is not merely unhelpful — it is absent.

## 2. Was westward access genuinely unusual?

Yes, and it survives refinement. Thirteen objects of 481 crossed 80° E at 6 h; eleven at 3 h. Beyond that: 4 and 5 at 70° E, 2 and 3 at 60° E, 1 and 2 at 50° E. Against 451 and 452 objects carried east past 100° E, at a median of 102 days.

Every western crosser in both runs originated at 33° S or 35.5° S, except a single 37.5° S object at 6 h that did not reproduce. **No object released at or south of 36.5° S reached a western gate in either integration.**

The rate is a supportable quantity. The identities are not: only four of the thirteen cross again at 3 h, and first-passage dates shift by up to 65 days. The named list of thirteen in the earlier report should be read as one realisation of a rate.

## 3. Did the ocean favour a northern route west?

Yes, and this is the most robust positive finding in the experiment. Every western crosser moved equatorward before turning west, finishing between 10.8° S and 29.3° S — the northern limb of the South Indian Ocean subtropical gyre, where the South Equatorial Current runs west. The result holds at both timesteps and is reinforced at 3 h, where the one southerly exception disappears.

## 4. Did the southern objects predominantly travel east?

Yes, and the censoring makes the statement stronger rather than weaker. Every object released at 38.7° S and south reached 100° E. Measured stirring falls monotonically with latitude — tortuosity from 4.84 at 33° S to 1.23 at 45° S, recurrence from 0.12 to 0.00 — and the fraction terminating against the eastern side of the domain rises from 1.7% to 100%.

The southern bands did not merely travel east; they left the experiment doing so. Their 0% westward access is **0% observed before censoring**, and for 43–45° S the year is largely a measurement of how quickly objects reach a boundary that was drawn by a data subset.

## 5. Did we identify a transport boundary?

**No.** What the data supports is a broad accessibility gradient with latitude, smooth across all eight release bands. Westward access falls 15.0% → 5.0% → 0% → 1.7% → 0% → 0% → 0% → 0% from 33° S to 45° S, at 60 objects per band. A zero sitting between a 5.0% and a 1.7% at that sample size is not a discontinuity, and the release grid was arbitrary sampling of the flow rather than a designed boundary experiment.

No separatrix, no event horizon, no permanent barrier is claimed or supported. The three monotone quantities — westward access, stirring, eastern termination — describe a regime change from an eddy-rich subtropical interior to a comparatively direct eastward conveyor on the circumpolar flank. That is a gradient in a year of one ocean.

## 6. What remains unresolved?

**The censored population, which is about half the ensemble.** 235 objects at 6 h and 222 at 3 h terminated near the eastern side of the study domain. Of these, 196 and 192 left beyond 119.5° E — the edge of the downloaded subset, where behaviour afterwards is simply unknown. A further 39 and 28 terminated between 113° E and 119.5° E, inside the domain, consistent with the Australian shelf but not demonstrated as land contact without a coastline product. Two objects at 3 h terminated west of 113° E and are the only candidates for a genuine data defect.

Nothing is extrapolated beyond that boundary or beyond the year. Whether the eastern population recirculates, beaches on Australia, or passes into the Pacific is outside what was computed.

Also unresolved: path length and tortuosity are resolution-limited rather than converged — both rose under refinement, so they bound stirring from below rather than measuring it. Individual trajectories are not resolved at all; at day 365 corresponding objects differ by a median of 846 km and up to 4,230 km. And only one ocean reconstruction exists, so no ensemble over current-field uncertainty was possible and no result is "robust" in that sense.

## 7. What does this mean for MH370 drift interpretation?

Carefully, and less than it might appear.

**Conditional transport accessibility.** In 2014–15, surface material released between 33° S and 45° S in 85–98° E was carried east, quickly and overwhelmingly. Westward transport out of that region existed but was rare, late, one-way, and available only from the northernmost release latitudes. Material released at and south of 36.5° S did not reach the western Indian Ocean within a year in either integration — subject to the censoring above, which removes the southern bands from observation before the year is out.

**Discovery-selection bias, which cuts against over-reading all of this.** The western debris record consists of objects that reached coastlines where people look and were then recognised and reported. Material that stayed at sea, sank, travelled east, or came ashore where no one was looking leaves no trace in that record. The forward experiment is useful precisely because it never consults the finds — but what it shows is that the eastern half of the transport story is, by construction, nearly unobservable by beachcombing. A western recovery is weak evidence about origin when the collecting coastline draws from a wide catchment and the alternative outcomes are invisible.

**What must not be concluded.** The 481 objects are a circulation survey, not a probability distribution over crash sites, and they are not convertible into one. No crash location follows from this work. Nothing follows from G0 beyond the behaviour of that coordinate in this circulation — it went south, never ran west, and left the eastern boundary. Nothing follows from satellite-image objects, whose identities are unverified. And a reverse-drift density maximum is a property of where recovered debris entered the record, not a crash coordinate.

The honest summary for an outside reader: **this experiment constrains the ocean, not the aircraft.** Its most useful contribution is a clean negative — the proposed southern-to-western pathway does not operate in this year's circulation, and the westward route that does exist runs through the tropics from latitudes north of the search area.

---

## Completed and uncompleted checks

| check | status |
|---|---|
| Chronological analysis of saved 6 h trajectories (Part A) | **complete** |
| Same analysis repeated on the 3 h year | **complete** |
| Checkpoint/resume implementation | **complete** |
| Restart equivalence, continuous versus resumed | **complete** — bit-identical |
| Resumed run verified against archived 182-day 3 h run | **complete** — 0.000 m, 0 status mismatches |
| Full 365-day 3 h run | **complete** |
| Duplicate-row and repeated-event integrity | **complete** — none |
| Full-year 3 h versus 6 h comparison (Part C) | **complete** |
| Eastern-termination categories separated | **complete** |
| Recurrence-based eddy residence, artifact corrected | **complete** |
| Convergence beyond one refinement | **not run** — a sensitivity test is reported as such; no convergence claim is made |
| 1.5 h check on the 120 northern objects | **not run** — identified as the smallest useful follow-up, not performed |
| Ensemble over current-field uncertainty | **not run and not possible** — one reconstruction exists |
| Behaviour of the censored eastern population | **not run** — would require a wider domain |
| Coastline product to separate land contact from masked cells | **not obtained** |
