The spine
Since the satellite era, the start of the hurricane season shows no detectable change. Neither does the rain's annual total.
In ERA5 the structure does: the longest wet spell is shorter and wet days come in more separate runs while the totals hold flat. Two other rain products do not reproduce the shorter wet spell, so no rainfall-structure trend holds across products. Peak wave energy shows no trend.
November is a mirage; September is real.
Mushrooms tell you the climate. Fish: 213 species on record within 50 km.
The lagoon is the dangerous water.
La Niña outscores El Niño in this small window of ocean.
30-year probability of a M≥6.5 within 100 km: ~97%.
Climate
Is the wave changing?
Forty years of satellite shorelines: the waterline at the break has not moved beyond the noise since 1986, while the north end of the cove retreated about 0.7 m a year. The whitewater at the point shows no detectable shift, and satellites alone cannot see a moderate change there for about a decade. 27 September 2026.
New →
Essay: The accidental audit
On 20 September 2026 every dataset was rebuilt and every claim re-run with no author in the loop. Wave power lost significance (+4.1 → +3.8 kW/m/decade, p=0.27) and rainfall structure changed shape; with the series later pinned to ERA5, warming is +0.13 °C/decade (the lowest of nine series that run to +0.31) and two of seven rainfall-structure tests are significant in ERA5, neither of them across three rain products. The HURDAT2 storm findings were out of the cell’s reach; a preprint re-ran them on 27 September and withdrew “widening”: the top of storm intensity rises, its spread does not.
New →
Essay: Three hours
The measured morning at the point: offshore at dawn, onshore at a median 10:25, a mixed tide with a solar clock (no dawn low September–April), 8% of wet-season rain before noon, UV ≤3 before 9. A clock, not a forecast.
New →
Field note: The driest August in eight years
A dry month and a flat swell, from two different oceans. The Open-Meteo archive puts August 2026 at ~46 mm, the lowest August since at least 2019. September usually breaks both.
New →
Essay: The polite story and the dangerous story
Three climate signals at this coast read for the same structural shape. Two hold it; the wave-power leg shows no trend. Revised 28 September 2026.
Essay →
Hurricane season timing
Over 1949–2025 the first storm arrives 2.3 days a decade earlier, but from 1971 or 1978, when the catalogue is consistent, no change in first arrival, storm count or storm-days is detected (p 0.13–0.97). Peak wind +5 to +6 kt/decade on the satellite era, not significant. Revised 28 September 2026.
→
Strongest storms getting stronger
Decade maximum: 120 kt (1950s) → 185 kt (2010s); interpolated p90 80 → 131 kt. Spread flat: the top is rising, not widening. Revised 27 Sep 2026.
→
A 1969 "regime jump" we then disproved
5/5 methods said yes (1969, step −17.0 days, ΔAIC −7.36). Satellite-era retest said no (1/5). Worked example of the chorus catching its own false positive. Revised 27 September 2026.
→
Climate trends at the coast
Surface warming +0.13 to +0.31°C/decade across nine series; this page’s ERA5 figure, +0.13, is the lowest. No trend detected in rainy-season totals, timing or the heaviest days. Revised 29 September 2026.
→
Rainfall regime structure
No trend detected in annual totals. In ERA5 two of seven structure metrics are significant: the longest wet spell is shorter and wet days come in more separate runs. CPC and MERRA-2 do not reproduce the shorter wet spell; no structure trend holds across products. Revised 29 September 2026.
→
Marine heatwaves
Days above the 90th-percentile SST threshold, 1982–2025. 2026 to date: 134 MHW-days, already more than any full year.
→
Was June 2026 the biggest swell in a decade?
ERA5-Wave at the corrected offshore node, storms matched. June 2026 ranks 4th of 13 Junes; two of the three bigger (2018, 2025) were cyclone swell. Revised 27 September 2026.
→
Wave power trend
Mean power flat; no peak-power trend clears p<0.05. The record hour is Hurricane Calvin, 129 km, July 1993; 34 of 47 annual peaks are cyclone days. Revised 27 September 2026.
→
Hurricane track shifts
Where storms form, where they recurve: 1,260 storms over 77 years. Satellite-era northward lean +0.26°/decade, p=0.08, not a finding. Revised 27 September 2026.
→
Wildlife
Phenology vs climate drivers
Mushroom fruiting tracks the previous month's rainfall (r²=0.72).
→
Observer-effort bias
November has 10.7× more observations than May. Four taxa flip peak month after normalizing.
→
Biodiversity saturation
Species on record by group; how many remain undocumented cannot be estimated from these files yet. Where to look next.
→
Co-occurrence + range shifts
Top pairs separated from sampling artifacts. No robust range shift.
→
Diversity index trends
Shannon-Wiener + Pielou's evenness per taxon, rarefied.
→
Diel patterns
Hour-of-day clustering for three lagoon reptiles within 50 km. The peaks show when visitors photograph them, not when they are active.
→
Earth + methodology
The wind clock
Land breeze from 70° all night; sea breeze on at a median 10:25, 09:40 in March to 11:00 in January; clean window from first light 3.3 h. 393 station mornings. 26 September 2026.
New →
Nest temperature this season: a proxy
No sand is measured here, so this is a stated proxy: 25 Jul–18 Sep 2026 air ran +1.1/+1.2 °C over 2024/2025 with less rain, sea +1.21 °C above normal. Expect more female Olive Ridleys, mostly from late-season nests; one year matters little, a run of them would. 27 September 2026.
New →
Take part: Beach watch
Help record the beach, new rock walls and the wave: a photo post, twice-yearly beach profiles, wall and storm-damage reports, surf ratings. Open data.
→
The air tide
The barometer rises and falls 2.0–2.5 mb twice a day (12-h amplitude 1.01 mb Jun–Aug to 1.25 mb Dec–Mar, peak ~10:15–10:45), and the Moon's own 0.048 mb air tide shows in three years of readings. Corrects the site's earlier 3.3 mb. 27 September 2026.
New →
Where the sand goes
Net alongshore drift is north-west on both sides of the cove (net/gross +0.90) and at the point (+1.00), in every year and season since 2015: which would feed the break from the south-east, the Troncones side. Modelled, not yet measured; moderate confidence. 27 September 2026.
New →
How the wave links up
Each 50 m of the point breaks half the time at its own swell height; two gaps need 1.3–1.5 m, and the whole ride connects half the time at about 1.65 m of offshore swell (1.55–1.78). Exploratory. 27 September 2026.
New →
The connection season
Nearly all clear satellite images are from October to March, so they describe the dry-season break: the whole ride is seen connected in about 9–28% of snapshots from November to March. Summer values (about 20–47%) are inferred, not observed. Exploratory. 27 September 2026.
New →
The tide clock
Lows at 1–4 am and 1–4 pm in every month; a dawn low on 6 mornings a year, none Sep–Apr; Oct–Mar dawn is a high tide on 71–100% of days. 26 September 2026.
New →
How it rains
Night rain, hard rain: steps ≥1 mm/5 min carry 47% of the water, 06–12h gets 8%, outages hide the peaks, and the afternoon does not predict the night (AUC 0.41–0.55). 26 September 2026.
New →
Heat and UV
Heat index 110–112 °F on a median summer day, ≥105 on 84–97% of Jul–Oct days, five nights above 80 °F in 858; UV ≥8 only 11:00–15:00; 2026 ran warmer and drier. 26 September 2026.
New →
Preprints
Manuscript-length versions of the strongest findings, with methods, references and a numbers appendix mapping every figure to its artifact. Not peer reviewed. HTML and PDF.
→
Natural experiments in the record
Sealed first, read second. The coordinate fix cut daily-high error 0.9 °F (interval clear of zero, under the 1 °F bar it set). Hurricane years raise peak wave power +17 kW/m with no change in the mean detected. Warm-water years bring 3.3 more clustered rain runs. Three deploy experiments sealed for grading in a week.
New →
Forecast skill, scored against the beach
76 settled days: rain-day calls right 67% of the time, daily high within 2.35 °F on average. 13 owner-rated surf days set beside the engine's call; the engine was re-tuned twice on those ratings, so none of it is out-of-sample yet.
New →
Seismic hazard + b-value
Gutenberg-Richter per decade. Guerrero gap recurrence.
→
Citizen-science growth
Saladita observations vs Mexico national trend.
→
Pending: bird arrival-date trends per species (when GBIF deep pagination completes). One clear change already: the Eurasian Collared-Dove, first recorded within 25 km in November 2013 (none in 5,030 earlier eBird records). Heckscher EPAC test of trans-Pacific shorebird timing vs hurricane severity. Forecast skill vs naive baselines (when n≥20 ground-truth observations).