Findings · Original analysis

The strongest storms near Saladita are getting stronger, not just more frequent.

The 90th-percentile of EPAC peak intensity within 500 km of La Saladita climbed from 120 kt in the 1950s to 185 kt in the 2010s. Five statistical methods agree. The signal survives restriction to the satellite era — it isn't a catalog artifact.

HURDAT2 1949–2025 · Five-method ensemble · Satellite-era retest passed · 9 June 2026
Headline finding

Mean intensity has been rising at +4.6 kt per decade — already documented. This page is about the upper tail. The 90th percentile of decade-level max intensities is rising significantly faster than the mean. The distribution is widening, not just shifting.

Decade-by-decade

The upper tail jumped.

Each year contributes its peak max-intensity within 500 km of Saladita. Group by decade. Compute mean, median, and the 90th percentile. The mean creeps. The p90 leaps in the 2010s.

Decade
n yrs
Mean (kt)
p90 (kt)
1950s
10
76.5
120
1960s
10
66.5
75
1970s
9
87.2
125
1980s
10
80.0
115
1990s
9
91.1
130
2000s
10
85.5
135
2010s
10
99.5
185
2020s
6
107.5
145

2010s p90 reached 185 kt — Category-5 territory. The 2020s reading (n=6 so far) is 145 kt, still above every pre-2010 decade.

Method ensemble

Five methods, all real.

Method
Statistic
p
Decade-level Pearson (p90 vs decade)
r = 0.73
0.009
Decade-level Mann–Kendall
z = 2.35
0.019
Decade-level permutation null
0.029
Rolling-15-yr Pearson
r = 0.90
<0.0001
Rolling-15-yr Mann–Kendall
z = 9.25
<0.0001
Catalog-artifact check

Satellite era only — still real.

Pre-1969 hurricane records suffer from satellite catalog incompleteness; earlier chorus iterations on this site found that "regime change" tests on the full record could be artifacts of detection improvement. The same retest applied here.

Restricting analysis to 1978–2025 (consistent satellite surveillance), the rolling-15-yr p90 trend gives r = 0.87, p < 0.0001. The signal persists. The widening is not a catalog artifact.

The strongest storms reaching this coast are getting stronger faster than the average storm is. The mean and the median tell the polite story. The 90th percentile tells the dangerous one.
Synthesis · June 2026
Relation to prior findings

The "extremes shift, means hold" pattern shows up again.

The site already documents the same shape in two other signals at Saladita. Wave power: annual mean is statistically flat over 47 years, but annual peak climbed from ~24 kW/m in 1979 to 81 kW/m in 2024. Rainfall: annual totals are unchanged, but wet-event clustering is rising +1 multi-day run per decade. Now hurricanes: mean intensity rising slowly, p90 rising substantially faster.

Three independent climate signals, same structural pattern. The means are moving. The extremes are running ahead of them.

Methodology & sources

Data. HURDAT2 East Pacific best-track database (NHC), 1949–2025. Per-year max sustained wind for any storm passing within 500 km of Saladita (17.5897°N, 101.4317°W). See _storm_climatology.js.

Methods. Years grouped by decade (n=8 decade buckets). Pearson, Mann–Kendall, and a 3,000-shuffle permutation null applied to the decade-p90 sequence. Independently, rolling 15-year windows across all 75 years with intensity data (n=63 windows), Pearson + Mann–Kendall on rolling p90. Both decade and rolling-window views agree.

Catalog robustness. Same rolling-15-yr analysis restricted to 1978–2025 (satellite era). Rolling-std and rolling-p90 both significantly rising in the satellite-only subset.

Reproducibility. python3 scripts/loop_iter8_intensity_distribution.py. Artifact: _findings_intensity_widening.js. This finding was generated by chorus-stack iteration 8 of the self-paced pattern hunt; iterations 1–7 produced one shipped finding (later walked back as a worked example) and five honest nulls.

Related findings