Somewhere in orbit, radiometers have spent years doing the least glamorous job in climate science: counting watts. Gergana Gyuleva is a doctoral researcher in climate physics at ETH Zurich. She took that ledger and used it as a scale to weigh the world’s climate models. The paper appeared in Earth’s Future in August 2026. Her co-authors are Erich Fischer and Reto Knutti of ETH Zurich’s Institute for Atmospheric and Climate Science, and Sebastian Sippel of the Institute for Meteorology at Leipzig University. Under current policies, their result puts warming at about 3.25 °C by 2100.

That number is not a prophecy. It is a measurement narrowing a range.

Why the heat ledger beats the thermometer

The physics starts with geometry. Earth catches sunlight on a flat disc but spreads it over a sphere with four times the area. A solar constant of about 1,361 W/m² therefore averages out to roughly 340 W/m² across the globe. The energy imbalance is the small difference between that income and what the planet sends back. Anadolu Agency summarised it as “the difference between incoming solar energy and heat radiated back into space”. A positive balance means the planet is still charging.

Climate models disagree on how quickly warming follows emissions. The team’s EGU26 abstract notes that models span “a wide range of 1-3K” in transient climate response. Earlier constraints relied on surface temperature alone, and the authors found natural cooling variability had pulled those estimates low. The satellite record of the imbalance is harder to fool, and the models that match its observed trend are the warmer ones. “For any emission trajectory without immediate and drastic emission cuts, we expect warming to be 25% higher,” Gyuleva is quoted saying. Anadolu adds that the remaining global carbon budget comes out about one-fifth smaller.

The credit goes to two groups. Gyuleva did the re-weighing. Before her, the instrument and calibration teams kept the radiometry stable for decades, which is what made a trend measurable at all. Without their patience, there would be no error bar to tighten.

The Swiss arithmetic

Switzerland warms faster than the global mean. Kristian Foss Brandt reported in swissinfo on 22 September that a 3 °C world means about 4.9 °C locally, using the MeteoSwiss/ETH CH2025 scenarios. Summer 2026 already brought 38 days above 30 °C in Zurich. Basel had 58, against 41 in 2003. Models project around 30 such days a year in Zurich in a 3 °C world; this summer has already passed that.

A tighter range here is also a higher floor. Any building designed today against yesterday’s weather data carries the difference for its whole life. The larger gap is political. Switzerland has cut emissions by about 27% against 1990, while its 2030 target is 50%, and Federal Councillor Albert Rösti has said the target will probably be missed. Scrutiny belongs on that gap, not on the researchers who did the sums.

←TODAY: Satellite heat budgets put the current-policy path at about 3.25 °C by 2100.
→3012: Zurich builds for the climate it has measured, not the one it remembers.
Fulcrum: A count of the planet’s watts reaches all the way down to the overheating hours in a top-floor flat in Kreis 5.

Atelier: For a 12-person studio in Zurich, this shows up in the summer overheating check. That check depends on a design weather file that may describe a cooler Switzerland than the one already arriving. PAZ has covered the façade side before: the ETH spin-off whose adaptive façade shades and generates power from one surface. Monday move: ask whoever runs your thermal simulations to list the weather file each live project uses, then add one warmer stress-test run to the next design review.

Hack: Count the heat days in your design weather file before you trust its overheating verdict. EPW files hold hourly dry-bulb temperature in the seventh column, below eight header lines, so one groupby gives you the daily maximum. The +2 K shift is an editorial stress test, not the paper’s figure.

import pandas as pd
epw = pd.read_csv("CHE_Zurich.epw", skiprows=8, header=None)
tmax = epw[6].groupby(epw.index // 24).max()  # daily max dry-bulb, °C
print((tmax > 30).sum(), "heat days;", (tmax + 2 > 30).sum(), "with +2 K")

If the file shows far fewer than the 38 heat days Zurich had this summer, your simulation is designing for the past.

From where this desk sits, the expensive mistakes were never the warnings. They were the buildings specified for a climate that had already moved on. Read the error bar, then design for its upper half.

The authors leave one lever open. “Severe emissions reductions could substantially limit future warming,” they write, and they are clear that their result does not change what a 3 °C world would look like. Whether we reach it is still open. Open your project’s weather file and count the hot days.

Sources & Further Reading

• Primary: swissinfo — ETH Zurich research on Switzerland’s climate scenarios
• Reinforcing: EGU26-17120 — Gyuleva, Fischer, Knutti, Sippel abstract