In the tropics, along the band of sky near the equator, clouds and wind run the show. These are juicy clouds that aggregate and disaggregate in agglomerations and that can live a long time, as far as clouds go. In the summer, when the ocean is especially hot, they can pile up high, breeding hurricanes; at all times of year, the behavior of tropical cloud systems drives global atmospheric circulation, helping determine the weather all over the world. And still, clouds remain one of the least understood—or least reliably predictable—factors in our climate models. “They are among the biggest uncertainties in predicting future climate change,” Da Yang, an atmospheric scientist at the University of Chicago, told me.
Yang is a cloud expert—a cloud guy, really, drawn to their mysteries. He recently moved from California to Chicago, where he gets to see a lot more clouds on a daily basis. “I find clouds are beautiful to watch,” he said. “If I take an airplane, and I can see clouds down below or far away, I’m always fascinated by how rich the cloud organizations are. How they interact with each other …” He trailed off. Clouds are complex and ephemeral, which makes them difficult to fully understand. Yang listed for me key aspects of clouds for which we still lack comprehensive understanding: how they form, what determines their spatial scale, how long they can last. “Those sound like simple questions,” he said, “but they are actually at the forefront of the field.”
The cloud problem has persistently plagued climate models. Although these models do many jobs extraordinarily well—understanding the energy balance of the planet, describing a trajectory of warming from human-made greenhouse-gas pollution—they can’t seem to get clouds right. Models will sometimes produce cloud-related projections that are simply incorrect, and some models “run hot,” meaning they predict catastrophic warming, possibly because of cloud dynamics.
One major stumbling block is the resolution of climate models, or how finely or coarsely they represent the Earth; to represent individual clouds, which can be the size of a minivan or the state of Minnesota, would require models at a resolution finer than the current finest model. Climate modelers have recently begun to produce fine-scale models at the regional level, where they can zoom in on the individual details of clouds. But, Yang told me, stitching such snapshots together into a picture of the whole globe would exceed the capacity of the largest existing supercomputer.
Even if computers did have the capacity to do these analyses, scientists would need more tools to understand the results. For that, Yang said, we need more cloud theory. “Without theoretical understanding, we would not be able to interpret the model results,” he told me. “Without these first-principal-based understandings, we don’t really know whether the model is accurate.”
Tiffany Shaw, a climate physicist at the University of Chicago, told me that some models are producing inaccurate visions of entire regions, possibly because of the cloud problem. For example, models predict more warming in the east Pacific than the west; the opposite is true in reality. Another example is the narrow belt of rainfall that rings the planet in the deep tropics and produces some of Earth’s strongest thunderstorms—and, as such, many clouds. Our planet generally has one such belt, but atmospheric-ocean climate models have been insisting for decades that it has two. This may, in part, be an issue of undercooked cloud modeling.
To Shaw, these irregularities are not a sign of something amiss; rather, they show the maturation of climate science. The field has gotten many of the big things right, and now it is learning to incorporate the smaller, more granular things into its vision of the world: things like clouds. Because of their complexity, Shaw is also excited about the possibility of using machine learning to understand them. “They’re data-hungry algorithms, and we have a lot of data,” she said.
One big question haunts all cloud research: Scientists know that there’s a lot of uncertainty about how to predict future cloud dynamics, and that those dynamics will likely have some bearing on how climate change progresses. But how significant of a bearing? For now, initial indications point to reassuring conclusions rather than catastrophic ones. “What we’re learning is that not everything matters for climate change. Which is good!” Shaw told me. For example, losing shallow cumulus clouds as the ocean warms—which some computer models have suggested could happen—would have a destabilizing effect on the tropics, potentially provoking runaway warming. But, Shaw said, a recent observational study found that the clouds aren’t as sensitive to warming as the computer models thought; the feedback between heat and clouds does amplify global warming, but not to the extreme degree suggested.
One of the keys to reconciling modeling and reality is simply more observations. Chris Fairall, a research physicist at the National Oceanographic and Atmospheric Association, has been studying clouds since the 1970s, when he worked on fog forecasting for the U.S. Navy, in highly foggy Monterey, California. “Fog is a cloud that sits on the ground. The Navy is very interested in fog, because they don’t want their ships running into things,” he told me. Fairall has seen the field of cloud science improve dramatically, in part thanks to efforts, including his own, to measure them. In 2020, he was the lead scientist on NOAA’s ATOMIC project, which flew one of the agency’s “Hurricane Hunter” planes and sent a ship to survey cumulus-cloud formations off the east coast of Barbados, as part of a larger joint cloud project with European researchers. Over the next few years, the data from those missions will help improve cloud models. Although Fairall likes studying relatively shallow cumulus clouds, he told me that the biggest cloud questions are about deep convective clouds, the ones that go all the way up into the troposphere, where they begin to develop complex ice, snow, hail, and supercooled water interactions. Cumulus clouds are complex enough; those deep clouds “have 100 times the complexity,” he said.
In his view, the U.S. is devoting a tremendous amount of effort to cloud research; it’s only up from here, in terms of cloud knowledge. NASA, NOAA, the Department of Energy, the Navy, and the Army all have researchers working on cloud problems, he said. Clouds envelop two-thirds of the Earth in their moist embrace, and in every moment help determine our collective physical reality. Surely the quest to understand them is among the most salient scientific endeavors of our time.
