to connect half the time
(−750 to −550 m)
(−950 to −900 m)
A long point does not break all at once. From 241 clear Sentinel-2 images taken since 2015, each 50 m stretch of La Saladita can be scored as breaking or not, and matched to the swell offshore at that moment. That gives, for every stretch, the swell at which it breaks half the time. The result is a profile of the ride: sections that work on almost any swell, and gaps between them that need more. This is an early, exploratory result, but it points at the number that would matter most if the wave changed.
Two sections that almost always break, and two gaps that need bigger swell.
The ride has two gaps: one just past the point (−950 to −900 m) and one in the middle (−750 to −550 m). The whole ride connects half the time at about 1.65 m of offshore swell, more than either gap needs on its own.
The point tip (−1,000 m) breaks on about 0.6 m. The outer section (−850 to −800 m) and the inner section (−500 to −400 m) break on about 0.7–0.9 m. The middle gap needs about 1.15 m at its outer edge, rising to about 1.45 m from −650 to −550 m; those three stretches cannot be told apart, so it is a bottleneck about 150 m long, not a single spot. Because both gaps must be open at once, the whole ride needs more swell than either. When only one gap shows whitewater, it is more often the middle one (38 images) than the one near the point (23). A snapshot shows which states occur, not the order in which a set switches them on; that order needs the camera.
Each image is one instant. A big set arriving at the moment of the pass can light up a gap that is dark a minute later.
The two gaps are open together more often than the swell alone explains: 41 images against about 32 expected. The likeliest reason is exactly this, a set crossing the whole point at once. The other possibility is a shared sandbar that makes both gaps easier. A fixed camera, filming wave by wave, will tell the two apart.
A rising threshold would be an early warning.
If the seabed under a gap changed, the first sign would not be a missing wave. It would be a wave that needs more swell before it links up: the same good days, but fewer of them. The swell needed for the whole ride to connect is the number to track. From satellites alone, a shift of about 0.4 m would show within roughly 30 new clear images, and 0.2 m would take two to three years; a camera would see it in weeks.
Higher tide makes the gaps harder to fill: 0.3 m more water needs about 0.2 m more swell. The satellite always passes at the same time of day, so it only sees tides between about −0.3 and +0.1 m around mean sea level. The full tide effect needs the planned pressure logger and camera.
Dry-season snapshots at about 11 am. Every threshold here comes from Sentinel-2 images taken around 11:10 local time, almost all between October and March. They describe the dry-season break at that hour, not the whole year or the whole day. The middle gap is well supported: 121 images fall within 0.2 m of its threshold, 51 connected and 70 not. The whole-ride threshold is thin: only 30 images fall within 0.2 m of it and only 10 have more swell than it, so treat 1.65 m as a rough figure.
Exploratory. How "breaking" is defined (whitewater reaching at least 25 m from the waterline) and how the gaps are drawn were chosen after looking at the data. They must be fixed in advance, with 10 m and 50 m versions, before this is used to compare any before and after, including the September 2026 storm. For that storm they were fixed on 27 September 2026, before any post-storm image existed, with a first look in September 2027 and the deciding read in September 2028. Satellite images alone can only detect a large change, of roughly 0.4 m or more, by then; anything smaller will be reported as not enough information, not as no change (natural experiments).
Satellite images measure whitewater at one instant, not ride continuity. What is estimated here is the chance of seeing whitewater at a moment, not the chance a surfer can make the section.
Processing noise is largest exactly in the gaps. Reprocessing the same image moved the whitewater edge there by 70–88 m on average. The standout stretch at −600/−550 m still exceeds what the measured processing error can produce.
The swell comes from a wave model offshore, not a measurement at the point, and nothing here shows what the seabed looks like under the gaps.
Data: 241 Sentinel-2 images, 2015 to 20 September 2026 (nearly all from October to March: sun glare and cloud leave almost no clear images from April to September, so the thresholds describe the dry-season beach), after removing duplicate processings of the same acquisition and screening out sun glint and cloud; whitewater measured along 50 m transects over the point. Offshore wave height, period and direction from the ERA5 ocean-wave hindcast at the time of each image; tide from this site's harmonic model.
Method: a stretch is "breaking" when whitewater extends more than 25 m seaward along its transect. For each stretch, a logistic fit of breaking against offshore wave height gives the height for a 50% chance (H50), with 95% ranges from resampling whole calendar years. Gap 1 = both −950 and −900 m breaking; gap 2 = at least 3 of the 5 stretches from −750 to −550 m breaking; whole ride = both gaps. Tide effect from a fit that also includes period, direction and tide. Processing error from 50 acquisitions processed twice. ERA5 wave heights after 1 February 2024 read about 9% low; correcting them, or using only images before that date, moves each threshold by 0.02–0.04 m, inside the ranges shown. A threshold only has meaning with the wave model it was fitted on: at these images ERA5 heights run about 28% above those of the Météo-France model at the grid point nearest the break. On Météo-France heights, October 2021 to March 2026 (132 images, tide included), the whole ride connects half the time at about 1.44 m (1.32–1.62), the middle gap at 1.21 m and the gap near the point at 1.31 m. The before-and-after test for the September 2026 storm uses the Météo-France scale.
Analysis run: 27 September 2026. Exploratory; not a pre-registered test.
Cite as: "How the wave links up: swell needed for each section of La Saladita", La Saladita Field Guide, lasaladita.com/findings/how-the-wave-links-up/, 2026-09-27.