(ERA5 T2m, 1979–2025)
arrival (1978–2025)
within 500 km (1978–2025)
trend (1979–2025)
ERA5 reanalysis (1979–2025) and NHC HURDAT2 storm tracks (1949–2025) at 17.8375°N, 101.7606°W. Linear regression; p-values approximate. One signal is clear in every product: the surface is warming. Nine temperature series put the rate between +0.13 and +0.31°C per decade, and this page’s ERA5 figure, +0.133, is the lowest of them. Over 1949–2025 the first storm within 1,500 km arrives earlier (−2.3 days a decade) and storms are more frequent, but both trends come from the pre-satellite years. 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 within 500 km keeps a similar slope (+5 to +6 kt a decade) but is not significant on the satellite era (p≈0.07–0.09). No trend is detected in rainy-season total, timing or 99th-percentile daily rain.
Revised 26 September 2026: ERA5 metrics rerun at the corrected grid cell 17.8375°N, 101.7606°W. The June 2026 version of this page used a cell 44 km to the south-east (17.5897°N, 101.4317°W — a different town). The storm-season results come from HURDAT2, not from the grid cell.
Revised 27 September 2026: the storm figures on this page had been pasted into the analysis script from an older HURDAT2 download. The script now reads the storm table rebuilt on 19 September 2026 from hurdat2-nepac-1949-2025-091426.txt. First storm arrival moves from −2.45 to −2.30 days per decade (p 0.005 → 0.009); storm-days from +6.1 to +6.2 per decade; peak intensity within 500 km from +4.6 to +4.75 kt per decade (p 0.001 → 0.0008); mean peak 85 → 85.5 kt.
Revised 29 September 2026: the warming rate is given as a range across nine temperature series, and the rainfall paragraph states which structure results hold across three daily rain products. The ERA5 numbers on this page are unchanged.
The coast at Saladita is warming.
ERA5 2m air temperature at the Saladita grid cell, annual means 1979–2025. Linear regression on 47 annual data points.
+0.13°C per decade in ERA5 at this cell (p=0.001, R²=0.21). Total implied change at that rate: about +0.6°C over 47 years.
Significant by any conventional criterion; calendar year explains about a fifth of the variance in annual mean temperature. The rate is in the range of regional tropical Pacific warming, and it is the low end of what nine series give for this coast (table below).
Every series warms. The rate runs from +0.13 to +0.31°C per decade, and this page’s figure is the lowest.
The +0.133 above is ERA5 at the grid cell the Open-Meteo archive returns by default for this coordinate: 18.0°N, 101.75°W, 19 km north of the point and inland. The nearest ERA5 cell, 17.75°N, 101.75°W, is the one whose footprint holds the break, and it gives +0.17. On the common window 1981–2024, annual means:
Intervals are from a 3-year block bootstrap. Over 1979–2025 the default-cell ERA5 series gives the +0.133 quoted above and the nearest cell +0.168. The airport record steps up against the reanalyses by 0.5 to 0.75°C between about 2015 and 2018, and its warming sits in daytime maxima (+0.33°C per decade) more than in night minima (+0.12), so it may hold a station change. There was no thermometer at the break before November 2023, so no series here is ground truth; the range is the finding.
Note on data source: NOAA OISST v2.1 satellite-derived SST returns no-data (NaN) at this grid cell due to land masking — the Saladita coast falls inside the coastal exclusion zone of the standard 0.25° SST grid. ERA5 2m air temperature is the next-best continuous long-term record for the ERA5 grid cell containing this coordinate. Surface air temperature and SST are highly correlated in the coastal tropics; the direction and approximate magnitude of this trend are consistent with regional SST trends reported in the literature.
No trend detected in rainy-season totals, timing or the heaviest days.
ERA5 hourly precipitation at the Saladita grid cell, aggregated into four annual metrics covering the May–October rainy season.
No trend detected in totals, timing or the 99th-percentile daily rain (+4.9 mm/decade, p=0.14).
Seasonal total, onset, end and the p99 day all fall short of p<0.05. The rainfall-structure analysis has the annual maximum one-day rainfall at +11.8 mm/decade (p=0.058), marginal and not significant. In ERA5 that page finds a change in how the rain is spaced, a shorter longest wet spell (−3.26 days per decade) and more separate wet runs. Two other daily rain products do not reproduce the shorter wet spell (CPC gauge analysis −0.44, interval spanning zero; MERRA-2 +1.97), so no rainfall-structure trend holds across products. Wet-run counts rise in all three, and only ERA5 clears significance. ENSO dominates year-to-year variance.
Mean seasonal rainfall over the full record: 969 mm (May–Oct total). Mean rainy-season onset: around day 146 (approximately May 26). Mean end: around day 303 (approximately October 30).
The storm catalogue: trends in the full record, none detected since the satellite era.
East Pacific storm tracks within 1,500 km of La Saladita (the "swing zone" covering activity from Mexico's coast to well offshore), with separate tracking of storms within 500 km (direct-impact range). Seventy-seven years, every named East Pacific storm 1949–2025. The last column refits from 1978, when geostationary coverage makes the catalogue consistent; its p-values allow for year-to-year correlation.
No change in first arrival detected since 1971 or 1978. Mean first arrival: day 156 (June 5).
Over 1949–2025 the fitted line moves 2.3 days earlier a decade, but refit from 1971 (+0.62 days, p=0.59) or 1978 (+0.55 days, p=0.72) it is flat, so the long-record slope comes from early-season storms the pre-satellite catalogue missed.
Peak wind within 500 km: +4.75 kt/decade over 1949–2025; +5 to +6 kt/decade from 1971 or 1978, not significant (p≈0.07–0.09). Mean peak: 85.5 kt. 97% of years had at least one storm in the 500 km ring.
The slope holds its size in the satellite era, but with fewer years it does not clear significance. See the ENSO-conditioned breakdown in the hurricane-phenology analysis.
Storm dwell time near Saladita, 1949–2025.
Two proximity metrics derived from the per-year storm aggregates. The artifact does not record minimum closest-approach distance per storm, so we use storm-days-per-storm as a proxy for track proximity and dwell time.
Over 1949–2025, storms stay longer in the approach zone. 500 km frequency was already at ceiling (97%).
Mean dwell: 4.2 days per storm, rising at +0.13/decade over the full record. This fit includes the pre-satellite years and has not been refit on the satellite era.
Pre-1966 storm records carry systematic undercount bias. Before geostationary satellite coverage, storms in open water that did not strike land were frequently missed. Excluding those years removes the trends: from 1966, first arrival gives p=0.44 and storm counts p=0.08; from 1971 or 1978 no change in first arrival, storm count or storm-days is detected (p 0.13–0.97).
NOAA OISST v2.1 returns NaN at this grid cell. The standard 0.25° SST grid land-masks this near-coast point. ERA5 T2m is used instead. The warming direction is consistent with regional SST records, but the magnitude is a proxy, not a direct ocean measurement.
p-values are approximate. All regression tests assume annual independence. Storm counts, precipitation totals, and temperature in adjacent years are weakly autocorrelated; the true effective sample size is probably 20–40% smaller than the nominal n. The trends that clear p<0.001 are robust to this correction; those near p=0.05 should be treated with more caution.
77 years is a long record but still short for climate. Individual decades with anomalous ENSO sequences — like the persistent La Niña of 2020–2022 — can distort apparent trends. The results here describe the observed 47–77-year period and do not constitute projections.
Rainy-season non-results are also results. The absence of a significant trend in seasonal total, onset and end is informative, not a failure. It means the amount of rain and the calendar of the May–October window have not demonstrably shifted over this record. The top of the daily distribution does not clear the bar either: p99 +4.9 mm/decade (p=0.14), and the annual maximum one-day rain on the rainfall-structure page is +11.8 mm/decade (p=0.058), marginal.
Surface warming: +0.13 to +0.31°C/decade across nine series, this page’s ERA5 figure the lowest. Storm season: since the satellite era, no change detected in its start or its storm count. Rain: no trend detected in totals, timing or the heaviest days.Synthesis · Revised 29 September 2026
Sources
Temperature: ERA5 hourly T2m at 17.8375°N, 101.7606°W via the Open-Meteo archive (models=era5), 1979–2025 (47 years). NOAA OISST v2.1 returns NaN at this coastal cell; ERA5 T2m substituted. Precipitation: ERA5 hourly precip same cell; rainy season May–Oct; onset = first day cumulative >50 mm; end = last day ≥10 mm. Storms: NHC HURDAT2 East Pacific best-track, 1949–2025 (nhc.noaa.gov); haversine distance; pre-1966 records carry undercount bias. OLS regression via scipy.stats.linregress; p-values approximate (autocorrelation reduces effective n by ~20–40%). Over 1949–2025 the storm results clear the Bonferroni threshold (p<0.0125); refit from 1971 or 1978, with p-values adjusted for lag-1 autocorrelation, first arrival, storm count and storm-days do not (p 0.13–0.97) and peak wind gives p≈0.07–0.09. Scripts: scripts/analyze_climate_trends.py → functions/api/_findings_climate_trends.js.
Analysis run: 28 September 2026 at the corrected cell. The June 2026 ERA5 numbers on this page came from the 17.5897°N, 101.4317°W cell and are superseded; HURDAT2 results were re-run on 27 September 2026 on the current storm table.
Corrected 28 September 2026: the ERA5 series is now requested as ERA5 itself (models=era5); the default Open-Meteo product splices two models at 2017, and without that splice warming is +0.13°C per decade rather than +0.28, and the p99 rain trend is no longer significant.