Sometimes the most interesting things in the universe are hiding in plain sight. And sometimes they are in what would be plain sight but are totally unlit, so being in plain sight really doesn’t help.

On the Moon, there are a number of craters at the north and south poles that are in permanent shadow. This lack of sunlight means that ice has the ability to build up in these craters. If you’ve watched Apple’s For All Mankind, you know that ice is one of our best hopes for getting water for future astronauts and visitors to the moon.

Before sending people to explore, NASA wants the best possible understanding of the interiors of these craters, and currently, that means sending a rover named VIPER to explore. VIPER is actually an acronym for Volatiles Investigating Polar Exploration Rover, and it will visit the Moon’s south pole in (if all goes well) 2023.

Now, before sending the rover to explore before we send humans, NASA really wants high-quality data of those permanently shadowed craters, and to do that, they are taking images with the Lunar Reconnaissance Orbiter, but I think I mentioned those craters are in permanent shadow, and that makes them hard to image. Hard, however, is not impossible. 

Scattered light off the walls of the crater allow just enough light to enter the craters that with enough Photoshop-like image tweaking, you can start to make out what is there. Unfortunately, this process isn’t as fast as TV shows would have us believe, and having humans tweak every image is a time-consuming process. So researchers had humans tweak 70,000 images and then used those 70,000 images to train AI software to tweak future images, and it works. 

Researchers can now get a basic idea of what is in these craters, so thanks to images from an orbiter and processing from an AI, NASA will be able to send a rover to explore an area that humans wish to visit. I feel like we need to rewrite the children’s song “The Farmer in the Dell” to be “The Scientists in the Lab” and tell this wild story of trying to explore the Moon.

More Information

Max Planck press release

“Peering into lunar permanently shadowed regions with deep learning,” V. T. Bickel, B. Moseley, I. Lopez-Francos and M. Shirley, 2021 September 23, Nature Communications