Charlie Wood writes:

In October 2008, Chris Bretherton lifted off from the coast of northern Chile in a C-130 turboprop plane. It was too dark to see the sandy hills of the Atacama Desert below, but the darkness suited Bretherton just fine. The researcher wasn’t going sightseeing. Seated directly behind the pilots, he kept his focus entirely on the sky.

The plane was stuffed with instruments, and its wings bristled with sensors and other devices. Bretherton’s immediate mission was to help the pilots collect information about the ice, water vapor, and air pressure around them. His longer-term goal was to use that data — as well as data he would collect over California, Hawai‘i, and Antarctica — to deal with one of the most challenging factors in climate science: clouds.

The plane passed a fluffy cumulus, and Bretherton spotted a rainbowlike prism of colors. This was strange; the cloud seemed too thin to host the large droplets required to refract light in this way. “The six-to-nine-hour flights rarely get boring,” Bretherton said, “because we always run into surprising cloud structures that rattle our scientific preconceptions.” He would later conclude that the air must have been so pristine that the cloud’s vapor was condensing into unusually large droplets on an unusually small number of particles.

In the nearly two decades since Bretherton boarded that plane, the globe has warmed by roughly half a degree Celsius. And clouds, which both reflect sunlight and trap heat, are still the biggest source of uncertainty in climate predictions. The world’s top supercomputers aren’t nearly super enough to include tiny digital clouds in the gigantic digital Earths they simulate. So climate scientists are developing workarounds, techniques for coaxing relatively cloudless climate simulations to swirl, storm, and warm as if they contained a full portfolio of realistic clouds.

Over the last few years, a competition has broken out among physicists to build the next generation of these crystal balls for climate. Bretherton, now working at the Allen Institute for Artificial Intelligence (Ai2), is one prominent entrant. Tapio Schneider at the California Institute of Technology is another.

Galvanizing these new efforts is the rise of machine learning techniques categorized as artificial intelligence. Schneider leans on AI to better incorporate the effects of clouds into climate models that use physics equations to see what’s ahead. Bretherton, worried that these equations will never fully capture clouds’ behavior, is developing new AI tools that can predict the future directly from real-world data, barely relying on physics equations at all.

While Schneider, Bretherton, and other physicists differ in their approach, they share a sense of urgency. “Climate is changing fast,” Bretherton said. “Having a perfect model in 100 years will not be useful for solving the climate crisis.” [Continue reading…]