The storm, frozen before the swell exists
Hurricane Polo dissipated without a Southern California chapter worth more than its archive. The storm that matters now is Rachel: 80 kt at 19.2 N / 108.5 W as of 1 October 03Z, with the National Hurricane Center forecasting 105 kt — Category 3 — by 2–3 October on a west-northwest track along 20 N, from about 109 W to 115 W through 5 October. That track keeps a swell-generating fetch aimed at Southern California for roughly 4 days.
Everything this article grades was frozen the night it was written: the NHC advisories (hashed, SHA-256), and the GFS-Wave model bulletins of cycle 2026-09-30 18Z at five Southern California buoys. The lock is dated by version control; editing it after the outcome is the one forbidden move.
One trap defused first: which way does the model say the swell points?
The wave-model bulletins do not document their direction convention, and the Macaronis grading error earlier this programme was exactly a direction mistake. So the convention was resolved against an instrument rather than assumed: buoy 46222 measured the existing 14-second swell arriving from 204 degrees, while the model bulletin wrote the same train as 3 degrees. The bulletin's directions are therefore "toward" — the reciprocal — and a bulletin direction near 0 means swell from the south.
The model's read at San Pedro, day by day
- 5 October — first arrivals: 0.34 m at 12.5 s (deep-water southerly swell partition)
- 6 October — building: 0.71 m at 12.1 s
- 7 October — peak: 0.97 m at 11.9 s, around 7 AM local
- 8 October — holding: 0.85 m at 11.5 s
- 9 October — fading: 0.63 m at 10.9 s
- 10–15 October — a separate, longer 15–16.5 s pulse from farther south. That one is NOT Rachel and must not be credited to Rachel.
The locked call
Four claims, graded as the week unfolds. The grading rule is the Macaronis lesson, now binding: the swell partition is scored against the buoy's swell partition, never against the total sea. Ground truth is CDIP buoys 045, 067, 092 and 191 plus NDBC 46222 and 46224, all frozen daily by the capture runner.
- Arrival of a distinct S swell, period 11 s or longerNot scored
- Called
- Monday 5 October, offshore sentinels first
- Measured
- Pending — graded as the week unfolds
- Peak day and sizeNot scored
- Called
- Wednesday 7 October morning: 0.8–1.2 m at 11–13 s, deep-water partition
- Measured
- Pending
- FadeNot scored
- Called
- Below 0.5 m by Friday 9 October evening
- Measured
- Pending
- A second, separate swellNot scored
- Called
- 15–16.5 s from farther south around 10–11 October — not Rachel
- Measured
- Pending
The sea it lands on is already a foot too high
For three straight days before this lock — 28 to 30 September, 72 hourly readings per station — observed water level ran about one foot above the tide predictions at two gauges 150 km apart: Los Angeles Outer Harbor averaged +0.99 ft above the tables (peak reading 7.40 ft against a predicted 6.22), and La Jolla averaged +0.97 ft, with the same shape and the same peak hour. King-tide water, delivered without a king-tide prediction.
Attribution of that residual stays notEstablished — it bundles warm-water expansion, any coastally trapped signal, air pressure and wind setup, and two gauges over 72 hours cannot separate them. What it does mean mechanically: Rachel's runup arrives on an elevated baseline, so midweek high-tide flooding risk in low-lying spots is higher than the wave heights alone imply. The full measurement lives with the Kelvin-wave verification, where the sea-level call of 14 September remains locked and unedited.
What would make this wrong
These are single-cycle numbers from one model chain. A track wobble south or west, faster decay over cooler water, or the Baja peninsula shadowing the southern edge of the fetch all shrink the result; Rachel holding Category 3 longer than the cycle assumed pushes the peak above the called band. At lock time the National Weather Service surf-zone forecasts for Los Angeles and San Diego carried no Rachel statement yet — their first high-surf statement, when and if it comes, is independent corroboration. Estimates are models, not guarantees; this page exists so the model is graded by instruments, not by memory.
This is research. Estimates are models, not guarantees; a model is never an observation, and a machine estimate is never “confirmed” — only a human reviewing it in the field makes it verified. Nothing on this page is an operational-capability claim, a safety promise, or a prediction of a specific event at a specific place.
References & data
- 1.NOAA National Hurricane Center — Hurricane Rachel (EP3) — Forecast/advisory of 1 October 03Z and CurrentStorms.json, frozen with SHA-256 at lock time.
- 2.NOAA/NCEP GFS-Wave point bulletin, buoy 46222 — Cycle 2026-09-30 18Z, the frozen model read this article grades; direction convention verified against the buoy.
- 3.NDBC buoy 46222 — San Pedro — Measured swell used to resolve the direction convention (14 s from 204 degrees) and part of the grading ground truth.
- 4.CDIP buoys 045, 067, 092, 191 — Directional wave ground truth for grading, frozen daily by the capture runner.
- 5.NOAA CO-OPS gauges 9410660 Los Angeles and 9410230 La Jolla — Observed-minus-predicted water level, 28–30 September: +0.99 ft and +0.97 ft mean residuals.
- 6.AYON Coastal Intelligence frozen captures — Event directory rachel-2026 and the Event 001 capture runner; commits 11aacad and 34974bb.