Plate Nº 56 · recorded November 12, 2026
Earth & ClimateReported finding
250 Climate Simulations Show Why Cloud Cover Numbers Mislead
A 250-simulation University of Wyoming study finds identical cloud-cover changes can warm Earth differently depending on cloud reflectivity, and exposes a model flaw.
By James Calloway3 min read696 words
In brief
- The study ran 250 configurations of the DOE's E3SMv3 Earth system model, published Oct. 18 in Geophysical Research Letters.
- Cloud reflectivity depends strongly on how efficiently the model converts cloud water into rain.
- None of the 250 simulations reproduced the observed cloud response to winds moving cool air over warmer oceans.
- Simulations and satellite observations analyzed cover the period 2010–2016; the study ran February through August.
- The study identified improvement opportunities but did not implement or test them.
Two hundred and fifty climate model simulations, run by University of Wyoming researchers, have produced an uncomfortable conclusion: two identical-looking losses of cloud cover can warm the planet by very different amounts. The difference comes down to how reflective the clouds were before they started disappearing.
The study, led by Timothy Myers, an associate research scientist in UW's Department of Atmospheric Science, was published Oct. 18 in Geophysical Research Letters. It examined marine low clouds over the world's eastern subtropical oceans — the flat, gray decks that cool the planet by bouncing sunlight back to space. How those clouds behave as Earth warms remains one of the biggest uncertainties in predicting how severe future warming will be.
"The same loss of cloud cover can have very different effects on climate depending on how reflective the clouds were to begin with," Myers said. "Losing highly reflective clouds allows more solar energy to remain in the climate system than losing the same area of less reflective clouds."
How the study worked
The work ran from February through August and builds on Nephele, a UW-led effort using hundreds of climate model simulations to probe uncertainty in cloud processes. The team ran 250 versions of the Energy Exascale Earth System Model version 3, or E3SMv3 — a U.S. Department of Energy model that simulates interactions among atmosphere, oceans, land and ice. Each run varied settings for uncertain cloud, convection and aerosol processes.
The main simulations and the satellite observations they were compared against cover 2010 to 2016. The computations ran on the Derecho and Casper systems at the NSF's National Center for Atmospheric Research-Wyoming Supercomputing Center.
Co-authors included UW's Travis Aerenson, Jacqueline Nugent, Hunter Brown and Daniel McCoy, plus Mark Zelinka of Lawrence Livermore National Laboratory.
What controls cloud reflectivity?
Here the story turns to rain — or rather, to how eagerly clouds make it. Small-scale turbulence and rain processes are too small for climate models to simulate directly, so models approximate them. That shortcut matters.
"We found that how readily the model converts cloud water into rain strongly controls cloud reflectivity," Myers said. "When that conversion is less efficient, more water remains in the clouds, making them more reflective and increasing the climate impact of a given change in cloud cover."
In plain terms: clouds that hold their water stay brighter, and brighter clouds punch harder on the climate ledger.
The model didn't fail across the board. It captured the direction of most observed cloud responses to meteorological and aerosol changes. But it often exaggerated their strength. And one response it consistently missed entirely.
What does this change for climate predictions?
The persistent failure involves what scientists call surface temperature advection — winds moving air across ocean regions with different surface temperatures. When cool air flows over warmer water, the ocean feeds the air extra heat and moisture, stirring mixing near the surface. Observations show this supports more low cloud cover and greater sunlight reflection.
The model doesn't reproduce that relationship. Not sometimes. Never.
"None of the 250 simulations recovered the observed response across the settings we tested," Myers said. Fixing it, he added, "will require closer examination of how the model fundamentally represents the underlying processes."
That is a structural flaw, not a tuning problem. And a caveat belongs here: the study identified where improvements are needed but did not implement or test any of them. Nobody should expect better forecasts next month.
Why the distinction matters
Myers frames the practical stakes plainly. Immediate beneficiaries are climate researchers and model developers. Downstream, better simulations support climate assessments and help governments, water managers, energy planners and communities prepare for shifts in temperature and rainfall.
"To understand how clouds affect warming, we need to know both how their coverage changes and how strongly they reflect sunlight," he said.
The message to model evaluators is blunt: measuring cloud coverage alone is not enough. Two models can predict identical coverage changes and still diverge sharply on warming — if their starting clouds differ in brightness. Any assessment of cloud feedback, and of how much warming greenhouse gases will ultimately deliver, has to check both numbers.
via phys.org/rss-feed/ (Source)
More from James Calloway
Show full bio
Staff writer covering regulation, policy and their impact on general science news at SciBeat.
346 articles
Nearby plates
- Arctic winter sea ice loss may exceed forecasts by 1 million
- Global warming has accelerated to 0.35°C per decade, study finds
- Marine Heatwaves Hide Months of Extra Ocean Warming, Study Finds
- Hot Droughts Triple Extreme Downpour Frequency, Study Finds
- Mars' North Pole Holds Far Less Dust Than Scientists Believed