GENEVA OBSERVATORY · ANNUAL MEAN TEMP · 1753–2025 · ONE STRIPE PER YEAR · METEOSWISS HOMOGENEOUS SERIES
Un guide de terrain · Lausanne edition · July 2026

The Shape
of Heat

Why warming feels faster than the numbers say, what the models actually predict, and what it means for a life lived at 46.5°N. Built on 273 years of real measurements.
Chapitre 00 · Pourquoi, au juste

The one-way blanket

Before any charts: why does an invisible gas warm a planet at all? Because the atmosphere is a window for sunlight but a maze for heat. Sunlight arrives as visible light and passes straight through. The warmed ground re-radiates that energy as infrared, and infrared is exactly the frequency CO₂ molecules vibrate at. They catch it and re-emit it in a random direction. Half of "random" is back down.

Solid — 19th-century physics, measured from orbit

Photons in, photons trapped

Yellow = sunlight (passes through) · red = infrared from the ground · rings = a CO₂ molecule catching and re-emitting · drag the slider
430 ppm
+1.4°Cequilibrium warming vs 280 ppm (≈3°C per doubling)
0infrared photons sent back down so far

More CO₂ means more catchers in the maze, so the surface must warm until enough infrared still escapes to balance the incoming sun. That's the whole engine. One more piece: warm air holds more water vapor, which is itself a greenhouse gas, roughly doubling the effect. Keep that sponge in mind, it returns in Chapter 08.

Chapitre 01 · Ce qui s'est déjà passé

The thermometer doesn't negotiate

Geneva has one of the longest instrumental temperature records on Earth, measured continuously since 1753, before the French Revolution, before the theory of the greenhouse effect existed. The record is boring for two centuries. Then it isn't.

Two centuries of nothing, then a staircase
Temperature anomaly vs the 1871–1900 average · thin line = yearly · thick line = 20-year mean
Geneva station (MeteoSwiss homogeneous) World (NASA GISTEMP, land+ocean)

The gap between the two lines is the first thing to understand: Geneva has warmed roughly two and a half times more than the planet. That is not a Swiss curse. The global average is dragged down by oceans, which warm slowly. Every inland landmass runs hotter than the mean, and the Alps add two amplifiers of their own: lost snow reflectivity and cleaner air letting more sun through.

+3.2°C
Geneva, 2020s vs 1871–1900 (measured, not modeled)
+1.3°C
World over the same period
2.9°C
Official all-Switzerland warming since pre-industrial (MeteoSwiss, 2024)
Chapitre 02 · Le vrai truc

The mean walks. The tail runs.

Daily temperature is a bell curve. Climate change does two things to it: slides it right, and stretches its hot side. A small shift in the middle multiplies the days beyond any fixed threshold. Your body keeps score with the threshold days, not the average, and that's the whole illusion of "exponential" warming.

Drag the planet's thermostat
Stylized Lausanne summer daily-max distribution, calibrated to the CH2025 anchors · blue = today (+1.3°C world) · red = your chosen world
+1.3°C ← we are here
1.0×
days ≥ 30°C
—
1.0×
days ≥ 35°C
—
1.0×
days ≥ 38°C
—
The hotter the threshold, the bigger the multiplier. This is why CH2025 can say "the average day warms 4°C" and "extreme heat days become 17× more frequent" about the same world.

Theory is nice. Here is the same effect in the actual Geneva measurements:

The tail, measured
Hot days (max ≥ 30°C) per year at Geneva, by decade · real station data
While the summer average rose ~3°C, days above 30°C went from 9 to 28 per year. Mean × 1.2, tail × 3. You noticed the tail.
Chapitre 03 · La machine à transpirer

The land forgot how to sweat

Why does the hot tail stretch instead of just sliding? Because of a literal phase change. While soil is moist, part of the sun's energy goes into evaporating water (latent heat) instead of heating air (sensible heat). The ground sweats, like you do. When soil dries out, the sweat budget hits zero and every extra joule becomes temperature.

Maritime
Britain, coasts. The ocean never stops sweating. Extremes stay buffered. Warms least.
Transition
Central Europe — us, now. The buffer fails in exactly the hot, dry years. Extremes amplify fastest here.
Dry
Interior Spain in summer. Buffer fully spent. Warming maps ~1:1 onto extremes again. Hot, but no longer accelerating relative to the mean.

Central Europe is mid-transition, which is why the last fifteen years felt like something snapped. 2003, 2015, 2018 and 2022 were all dry-soil summers with the brake missing. The important part: this is a battery that discharges once, not compound interest. Same for snow-albedo: once the low snow is gone, that feedback is spent too. The nonlinearity is real, local, and bounded.

Chapitre 04 · Quel futur, exactement

Pick a world

Headlines mix scenarios freely. The honest framing: the world is currently tracking ~2.6–2.7°C by 2100 under existing policies (Climate Action Tracker, Nov 2025), roughly the "middle road" SSP2-4.5. The 5°C doom pathway (RCP8.5) needs a fivefold coal expansion nobody is building. And 1.5°C is already out of reach. Below, what each world does to Switzerland, from the official CH2025 scenarios.

—
Switzerland's own warming (amplified ~1.6× the global level)
—
Hottest day of the year vs 1991–2020
—
Frequency of former once-in-50-years heat events
—
Winter zero-degree line rises (now ~900 m)
—
Snowfall days at 1,000–1,800 m
—
Swiss glacier ice surviving to 2100
Anchor values at +1.5° and +3.0° are official CH2025 / GLAMOS figures; values between anchors are interpolations (~). "50-yr event" frequencies interpolated logarithmically.
Chapitre 05 · Ta ligne du temps

The planet's timeline is your timeline

Assume the current-policies pathway: ~2°C world around 2050, ~2.6–2.7°C by 2100. Anchored to a birthday in 1992:

2026
AGE 33 · +1.3°C WORLD
Baseline. Summers already ~1.5°C above your childhood's. The 2022-style summer still counts as "exceptional".
2035
AGE 43 · ~+1.65°C
2022 becomes the average summer. Lake regularly above 25°C. Skiing below ~1,200 m mostly artificial.
2050
AGE 58 · ~+2.0°C
Lausanne feels like today's Lyon. Heatwave peaks 36–37.5°C. AC standard. Former 50-yr heat events every ~7 years.
2065
AGE 73 · ~+2.3°C (3.0 IF POLICY STALLS)
Today's Milan, or Toulouse in the stalled case: 38°C+ peaks, droughts 3× as frequent. Heat becomes a personal health variable at this age.
High confidence to ~2050 (that part is already locked in by past emissions). The 2065 spread is political, not scientific.
Chapitre 06 · La carte

Winners, losers, and the lucky

Climate resilience = low physical exposure × high capacity to adapt. On the 187-country ND-GAIN index, the top of the world list is essentially Northern Europe plus one Alpine country, and the bottom is a map of where exposure meets poverty.

Faring best (global rank)

  1. 1 · Norwayhydropower, wealth, cold headroom
  2. 2 · Finlandcarbon-neutral target 2035
  3. 3 · Switzerlandlowest vulnerability score of ANY country on Earth
  4. 4 · Denmarkadaptation planning, wealth
  5. 6 · Swedenso far net GDP gains from warmth
  6. 9 · Austriathe other Alpine seat

Hit hardest in Europe

  1. Spain · rank 29interior heat, water scarcity for ⅓ of the south at 2°C
  2. Italy · rank 35heat mortality, Po valley drought, debt + ~zero disaster insurance
  3. Greece · rank 34wildfire burnt area +96–187% at 3°C
  4. Portugal · rank 24fire + drought, crop losses to −50%
  5. Southern FranceMediterranean regime moving north
  6. Low coasts (NL, Venice…)sea level; NL = exposed but best-adapted

Rest of the world, worst first: the Sahel and Horn of Africa (Chad ranks 190 of 190; climate stress converts directly into famine and conflict), South Asia's river plains (humid heat approaching the survivability limit, Himalaya-fed water), the Middle East and North Africa (50°C+ summers, structural water deficit), small island states (existential territory loss), and the tropical delta megacities. Depending on pathway, 1–3 billion people end up outside the climate that has supported human life for 6,000 years. Almost none of them are in Europe. That pressure reaches Europe through migration and politics, not weather.

Physical exposure and adaptive capacity are inversely correlated. The hardest-hit places can least afford to adapt, and contributed least to the problem.
Chapitre 07 · L'eau

A timing problem, not a quantity problem

"But the glaciers are melting, is Switzerland really fine for water?" Fine, with an asterisk worth understanding. Glaciers are not the water source, rain and snow are, and Switzerland stays one of Europe's wettest countries (winter precipitation actually increases). Glaciers are the savings account with perfect timing: they release water exactly in the hot, dry weeks when everything else has run out.

−40%
Swiss glacier volume lost since 2000 (−10% in 2022–23 alone)
~25%
ice volume that survives 2100 if warming holds at 1.5–2°C · ~0% on the high path
6%
of Europe's freshwater reserves sits in Switzerland
№1
most dam-equipped country per capita — storage is exactly the fix for a timing problem

What actually breaks: late-summer rivers run low and warm (fish, cooling water, irrigation), and hydropower shifts from summer toward spring. What fixes it: reservoirs, including new ones planned in the very valleys the ice is vacating. Expensive, political, solvable. The people with the real glacier problem live downstream: Rhine shipping, French reactor cooling, Po valley farms. When the water tower's buffer shrinks, it's the bottom of the pipes that feels it.

Chapitre 08 · La tuyauterie

Three machines that move the heat

Everything so far was thermodynamics: how much heat. But weather is circulation: where the heat goes, and how long it parks. Three machines run Europe's plumbing, and they carry very different confidence labels.

Debated — the mechanism is real, its strength is contested science

№1 · The river of wind

The jet stream: a ~200 km/h river of air powered by the cold-Arctic/warm-tropics contrast · drag the slider to warm the Arctic
weak
100%jet speed (schematic)
lowwaviness → blocking risk

The jet is powered by the temperature difference between Arctic and tropics. The Arctic warms 3–4× faster than anywhere, so the difference shrinks, and a weaker river meanders and slows. A stalled northward bulge is a heat dome (2003, Pacific-Northwest 49.6°C in 2021, 2022); a stalled dip is a flood week (Ahrtal 2021). This is why extremes increasingly arrive as week-long sieges instead of hot afternoons, something no bell curve can show. Honesty chip applies: whether waviness is actually increasing is one of climate science's livelier fights; the stuck-weather damage, however, is observed.

Solid — Clausius-Clapeyron, 1834. Nobody debates this one

№2 · The sponge

Air holds ~7% more water per °C. The only true exponential in this story · drag to warm
+0.0°C
+0%more water in the same storm
+7%/°Ccompounding, like interest

This resolves the paradox sitting in plain sight on this page: drier summers AND heavier rain. Warmer air is a bigger sponge, so it sucks more moisture out of soil between rains (drought), and when a storm finally wrings the sponge, more comes out at once (CH2025: up to +30% rain intensity). It's also the machine behind Valencia 2024: a hot Mediterranean loading the sponge before one town gets a year of rain in a day.

Wildcard — low probability this century, rewrites the page if wrong

№3 · The conveyor

The AMOC: warm salty surface water flows north, cools, sinks near Greenland, returns at depth · add meltwater
85% conveyor strength
~15%estimated weakening since the mid-20th century
mildEurope's winters (the conveyor is why Lausanne isn't Quebec City)

The sinking is the engine, and it needs the water dense: cold and salty. Greenland meltwater is fresh, so it sits on top like oil and blocks the dive. Slow the dive, slow the whole loop. Europe sits at the warm end of this conveyor, which is why a serious slowdown is the one mechanism that could flip the sign of everything above: milder winters becoming much colder ones, storm tracks rearranged. Current science: weakened, not collapsing; collapse this century judged unlikely but no longer dismissed. This is the single item on the watch-list.

Chapitre 09 · Borné, pas exponentiel

Scary, bounded, and worth preparing for

The physics summary that dissolves the doom-scroll: warming scales roughly linearly with cumulative emissions, and it largely stops when net emissions stop, because the ocean and the land keep quietly swallowing about half of everything we emit. The "it's accelerating!" feeling is three stacked one-time effects: the mean shift, the tail stretch from drying soils, and perception that counts threshold days. Batteries discharging, not interest compounding.

The honest asterisk lives elsewhere: tipping points. For Europe the one that matters is the conveyor, machine №3 above, the only mechanism that could genuinely rewrite this page. Current science treats collapse this century as low-probability, high-impact. Check the research every few years, not every few headlines.

01
Stay. On climate grounds Switzerland is a top-3 seat on the planet. If relocation ever happens for climate, the direction is north, never south.
02
Housing that handles heat. Thermal mass, shading, or AC. By 2050 this is the single biggest daily-life variable, and by your 70s it's a health variable.
03
Keep altitude access. The lake plus a train to 1,500 m is the Swiss adaptation superpower. Value it when choosing where to live.
04
Treat residency as an asset. The resilient zones become more desirable, not less. Swiss residency/citizenship optionality appreciates.
05
Watch AMOC, ignore weather-porn. One scientific question on the watch-list. Everything else on this page moves slowly and predictably.
You were worried it was exponential. It isn't. It's a staircase you can see the shape of, and you happen to live on the best-placed step.