This week vs this city's climate (warm-season 2005–2024)
“Percentile” places each night's overnight low temperature within the distribution of all May–Sep nights for this city. A high percentile means a night that is hot for here — the acclimatization-aware framing the epidemiology favours.
| Night evening → next morning |
Min temp* overnight low |
Percentile for this city |
Rarity summer nights |
Hours ≥ 20° (range) |
Hours ≥ 25° (range) |
|---|
Hot nights per summer (May–Sep), 1985–2024
All figures here — both night counts and the mean overnight low — are over the May–September warm season. The chart counts nights whose temperature stayed at or above 20° all night, every year for 40 years: this is why nights matter, they are warming fast. Dashed line is the recent (2005–2024) average.
Summer excess mortality
All-cause excess deaths in the hot season (weeks 22–36) vs a trailing 5-year baseline. Country-level; summer spikes often coincide with heatwaves but are not heat-attributed.
Why this view exists
The main site answers a personal question — will tonight wreck your sleep. This one is for the public-health question: are we worried about people dying? Heat mortality falls mainly on the elderly, the chronically ill, the socially isolated and outdoor workers, and it is driven less by a single hot night than by consecutive nights with no overnight recovery. The 2003 Paris heatwave study found multi-day minimum (night-time) temperature — not daytime heat — predicted elderly deaths; a 2025 study across 34 European countries found back-to-back day-and-night “compound” heat carried >2× the mortality risk of daytime-only heat.
Absolute vs relative thresholds
The everyday view uses fixed 20°/25° thresholds because they are easy to read. But whether a 20° night is alarming depends on the city: it is a ~98th-percentile night in Paris (≈4 a year) and essentially unheard of in London, yet routine in Rome (~60 a year). Populations acclimatize, so the epidemiology uses location-specific percentiles. The “Percentile” and “Rarity” columns here put tonight on each city's own scale — the anomaly, not the absolute number, is what tracks excess deaths.
Methods & sources
- * Temperature = “feels like”: every temperature shown is the Steadman apparent temperature (shade formula, no solar term), from 2 m air temperature, relative humidity and 10 m wind — the humidity-inclusive index that holds up for temperate-European heat, unlike WBGT/wet-bulb.
- Forecast: Open-Meteo deterministic + 40-member ICON ensemble (for the hours-above ranges).
- Climatology: Open-Meteo historical archive (ERA5 reanalysis), warm-season (May–Sep) overnight windows. The yearly/decade counts span the full 1985–2024 record; percentiles use the recent 2005–2024 window as a stable normal.
- Excess mortality: Eurostat Deaths by week and sex
(
demo_r_mwk_ts), national level. Summer = ISO weeks 22–36; expected = mean of the same week over the prior 5 years (≥3 required); excess = observed − expected; P-score = 100·excess/expected. This is a deliberately simple baseline, not the official EuroMOMO model, and is all-cause, not heat-attributed.
Caveats
Excess mortality is country-level and all-cause: a summer spike is consistent with a heatwave but also with epidemics or other causes; attributing it to heat needs a proper exposure model. Coverage varies — France's weekly series is short (from ~2016), the UK is sparse post-Brexit, and Russia, Ukraine and Türkiye are not in Eurostat so those cities show no mortality panel. Landmark heat summers for reference: 2003 ~70,000 excess deaths across Europe (~15,000 France), 2022 ~60,000, 2023 ~47,000 (Ballester et al., Nature Medicine). As on the main site, we forecast outdoor temperature while the exposure that kills is largely indoor.