August
(after spike filter)
(daily minimum)
August, median
Air temperature at Saladita is unremarkable — the median daily maximum is 87–93 °F in every month. What the body feels is not. Humidity of 70–80% at the afternoon peak pushes the heat index, the NWS's estimate of apparent temperature in the shade, into the 110s on an ordinary summer day. This page is the station's own climatology of that: heat index, night minimum, ultraviolet and solar energy, from 297,356 five-minute readings over 858 days with adequate coverage.
Every number comes after the instrument corrections the station needs: 68 temperature readings removed as spikes or impossibilities (including an 80-minute plateau at 109 °F at midnight), the stuck-array window of August 2024 blanked, UV=15 and solar ≥2000 W/m² treated as sentinels, and 89 dead-solar-sensor days (mostly March–April 2024) excluded from the UV and solar tables.
Ten degrees of humidity on top of the thermometer, every summer afternoon.
Daily maximum of the NWS heat index (Rothfusz regression, with its low- and high-humidity adjustments) on days with ≥80% of readings; the median and 90th percentile across all days of that month in the record, and the share of days that reached 105 °F. tmax and tmin are the daily air-temperature extremes.
| Month | Days | Median daily-max HI (°F) | p90 daily-max HI | Days HI ≥105 | Median tmax (°F) | Median tmin (°F) | p10 tmin |
|---|---|---|---|---|---|---|---|
| Jan | 93 | 100 | 105 | 10% | 88.0 | 68.5 | 65.4 |
| Feb | 81 | 99 | 101 | 0% | 87.4 | 68.5 | 65.3 |
| Mar | 62 | 98 | 102 | 2% | 87.4 | 65.9 | 63.1 |
| Apr | 68 | 96 | 105 | 7% | 87.3 | 70.2 | 63.9 |
| May | 40 | 106 | 112 | 62% | 90.7 | 75.6 | 71.9 |
| Jun | 55 | 107 | 112 | 69% | 89.4 | 76.3 | 71.8 |
| Jul | 75 | 112 | 118 | 91% | 92.5 | 75.6 | 73.6 |
| Aug | 87 | 112 | 117 | 97% | 93.0 | 75.7 | 74.0 |
| Sep | 75 | 111 | 116 | 88% | 92.1 | 75.9 | 74.3 |
| Oct | 61 | 110 | 115 | 84% | 91.0 | 74.8 | 73.6 |
| Nov | 71 | 108 | 111 | 86% | 90.9 | 73.0 | 69.6 |
| Dec | 90 | 104 | 107 | 33% | 89.1 | 70.4 | 66.4 |
The same threshold by year, July–October:
| Year | Jul (days ≥105 / days measured) | Aug | Sep | Oct |
|---|---|---|---|---|
| 2024 | 28 / 31 | 25 / 25 | 22 / 28 | 26 / 30 |
| 2025 | 29 / 31 | 28 / 31 | 27 / 29 | 25 / 31 |
| 2026 | 11 / 13 | 31 / 31 | 17 / 18 | — |
Median daily-max heat index 110–112 °F July–September, 98–100 °F January–March. HI ≥105 °F on 22–31 of the measured days in every full July–October month of every year.
The NWS "danger" band starts at 103 °F; the station crosses it on most days from May to November. The 2026 partial months (July restarted on the 18th, September ends on the 19th) are 11/13 and 17/18 — the same rate. The winter median of 98–100 °F is the number visitors should expect in the "cool" season: still a hot afternoon, with a p90 near 105.
| Date | Heat index (°F) | At | Air temp (°F) | RH |
|---|---|---|---|---|
| 2024-07-28 | 123.9 | 16:00 | 94.5 | 73% |
| 2026-09-17 | 123.4 | 13:55 | 95.2 | 70% |
| 2024-07-15 | 122.9 | 14:25 | 93.2 | 77% |
| 2024-08-08 | 121.7 | 16:00 | 93.4 | 75% |
| 2024-07-16 | 120.6 | 12:40 | 91.9 | 80% |
| 2024-07-19 | 120.1 | 15:50 | 93.0 | 75% |
The three highest are 2024-07-28, 2026-09-17 and 2024-07-15, each a mid-90s air temperature at 70–77% humidity. No 2025 day makes the top six.
The night is the relief. Almost always.
Median tmin runs 74.8–76.3 °F from June to October and 65.9–68.5 °F January–March (table above). A night that never drops below 80 °F is rare: five in the record, counting only days with ≥80% coverage overall and ≥80% of the 00–07h readings present. Sixteen days show tmin ≥80 before that coverage test; eleven of them are outage nights where the minimum was simply never measured.
| Local day | Minimum (°F) |
|---|---|
| 2025-05-15 | 80.6 |
| 2025-05-23 | 80.6 |
| 2025-06-04 | 81.1 |
| 2026-09-09 | 80.6 |
| 2026-09-13 | 80.1 |
UV 8 lives between eleven and three, and takes the winter off.
| Month | Days | UV ≥8 window (median first–last) | Hours/day UV ≥8 | Days reaching 8 | UV before 09:00 (p90) | Median daily max UV | Solar MJ/m²/day |
|---|---|---|---|---|---|---|---|
| Jan | 93 | 12:22–13:20 | 0.0 | 15% | 0 | 7 | 14.8 |
| Feb | 81 | 12:35–13:40 | 0.5 | 56% | 0 | 8 | 18.2 |
| Mar | 62 | 11:50–14:25 | 2.5 | 100% | 0 | 8 | 22.0 |
| Apr | 66 | 11:30–14:35 | 2.8 | 97% | 1 | 9 | 22.6 |
| May | 66 | 11:15–14:20 | 2.4 | 91% | 1 | 9 | 21.8 |
| Jun | 55 | 11:50–14:30 | 1.3 | 80% | 1 | 8 | 18.6 |
| Jul | 76 | 11:20–14:35 | 2.5 | 93% | 1 | 9 | 21.0 |
| Aug | 87 | 11:10–14:42 | 2.5 | 97% | 1 | 9 | 21.0 |
| Sep | 75 | 11:10–14:30 | 2.0 | 89% | 1 | 9 | 19.0 |
| Oct | 61 | 11:38–13:50 | 1.3 | 82% | 1 | 8 | 18.6 |
| Nov | 71 | 11:55–12:25 | 0.0 | 13% | 0 | 7 | 16.4 |
| Dec | 90 | 13:32–13:40 | 0.0 | 9% | 0 | 6 | 14.4 |
97% of all UV ≥8 readings fall between 11:00 and 15:00. Before 09:00 the 90th-percentile UV is ≤1 in every month. Mean hours/day at UV ≥8: 2.4–2.8 in March–May and July–August, 1.3 in June and October, essentially zero November–January.
June dips because it is the cloudiest month of the wet season, not because the sun is lower. Solar energy peaks at 22.6 MJ/m²/day in April and bottoms at 14.4 in December — a 36% swing, small for the latitude, because the wet-season cloud flattens the summer.
The station's median summer daily-max UV is 9. A clear-sky noon at 17.8°N in July is closer to 12.
The sensor very likely reads low (ageing, dust, or partial shading by the mount), so the window is trustworthy — the timing of a sensor's own peak does not depend on its gain — but the absolute index is not. A day the station calls "8" may be an 11. Nothing on this page should be read as "the UV is moderate before eleven"; it says the UV is lower before eleven.
This summer ran warmer and drier. Probably real. Watch it.
Mean temperature and relative humidity at seven local hours, July–September, 2026 against 2024 and 2025 pooled. The 2026 array restarted on 18 July after two dead months, which is exactly the kind of coincidence that should make you suspicious.
| Local hour | Temp 2024–25 (°F) | Temp 2026 | Δ | RH 2024–25 | RH 2026 | Δ pts |
|---|---|---|---|---|---|---|
| 03h | 77.0 | 78.5 | +1.4 | 94% | 87% | -7 |
| 06h | 76.3 | 77.3 | +1.0 | 95% | 88% | -6 |
| 09h | 85.5 | 87.4 | +1.8 | 79% | 71% | -8 |
| 12h | 89.2 | 91.5 | +2.3 | 73% | 64% | -8 |
| 15h | 89.0 | 92.1 | +3.1 | 74% | 64% | -10 |
| 18h | 85.1 | 88.5 | +3.4 | 81% | 71% | -10 |
| 21h | 80.2 | 82.6 | +2.4 | 91% | 84% | -8 |
+1.0 to +3.4 °F and 6–10 RH points drier at every hour, largest in the afternoon and evening.
Two facts argue it is real rather than a sensor change: the forecast model's bias against the station at the correct grid point over a ten-day check was −0.12 °F, i.e. the model agreed with the station's absolute level after the restart; and the warm-dry pattern has a diurnal shape (largest at 15–18h) that a gain error would not produce. One fact argues caution: it began the day the array was reinstalled. Label: probably real, watch it, and re-run this table when the 2026 season closes.
The heat index is a shade estimate from one sheltered sensor. The NWS formula assumes shade and light wind; in the sun on the sand the apparent temperature is higher. The station's temperature sensor sits in a radiation shield among trees; it is representative of the shaded village, not the beach at 2 pm.
Corrections applied. 68 temperature readings removed: >5 °F from a centred 1-h or 3-h median, or outside 50–100 °F (the site has never recorded a genuine 100 °F; the 19 readings above it are a 109 °F midnight plateau on 2024-05-20/21 and the run-up to the stuck array). Stuck array 2024-08-18 12:00 → 08-21 06:00 blanked (793 rows). UV=15 and solar ≥2000 W/m² are sentinels. 89 days with max solar <50 W/m² are dead-sensor days, mostly March–April 2024.
The UV sensor reads low. Median summer daily-max UV of 9 against a clear-sky expectation near 12 for this latitude. Windows and shapes are trustworthy; absolute values are lower bounds.
Sample and gaps. 858 days with ≥80% temperature coverage across 2.9 years; April–May have 40–68 days, other months 55–93. May 2024 has nine covered days. The hot-night count depends on the coverage rule: five genuine, sixteen before the rule, most of the difference being nights the station was down.
2026 is a partial season and one season. July 2026 starts on the 18th; September ends on the 19th. The warm-dry comparison is one summer against two; a single hot year is not a trend.
The thermometer says 92. The body says 112, on a median August afternoon, on nearly every day from July to October. The night gives it back — five nights in three years did not. The sun is a four-hour problem, eleven to three, and the station's own sensor understates it. And the summer just ending was the hottest and driest of the three, by a margin the model agrees with.Synthesis · 26 September 2026
Sources and method
Data: on-site Ambient Weather station, 5-minute tempf, humidity, uv, solarradiation, 2023-11-09 → 2026-09-19. Local day = UTC−6.
- Heat index: NWS Rothfusz regression with the low-RH (RH<13%, 80–112 °F) and high-RH (RH>85%, 80–87 °F) adjustments; the simple average formula below 80 °F. Daily max on days with ≥80% of readings.
- Hot night: daily minimum ≥80 °F with ≥80% coverage overall and ≥80% of 00–07h readings present.
- UV window: first and last 5-min reading at UV ≥8 per day, medians over days with ≥80% UV coverage; hours/day = count of readings ≥8 × 5 min; p90 of readings before 09:00.
- Solar: Σ W/m² × 300 s over days with ≥90% coverage, MJ/m²/day; dead-sensor days excluded.
- 2026 comparison: mean tempf and humidity in the 5-min readings of each listed local hour, Jul–Sep, 2026 vs 2024+2025.
Script: scripts/analyze_station_climatology.py → functions/api/_findings_station_climatology.js (export STATION_CLIMATOLOGY, generated 2026-09-27).
Cite as: Conner, A. (2026). "Heat and UV at La Saladita — What the Body Feels, Hour by Hour." La Saladita Field Guide, lasaladita.com/findings/heat-and-uv/, 26 September 2026. Data: on-site station 2023–2026; artifact /api/findings-station-climatology.