Take a good look at the photo below. Shot from a satellite, it shows a section of the Grand Canyon, with the dark ribbon of the Colorado River winding through it. Notice anything “off” about the image? Especially in the upper portion and in the area of the big hairpin turn at lower right?

The river lies about 2,000 feet below the surrounding plain. Yet in this photo, the down-cut sides of the canyon appear to rising UP from the surrounding plain.

This photo ran in the April issue of Research|Penn State, as part of a news brief about research on how much water flow would have been needed to cut a major valley on Mars. It looked a little odd to me, which I chalked up to not having much experience looking at such pictures. But reader Grady Meehan, who graduated from Penn State in 1964 and earned a Master’s in geography here in 1970 and a Ph.D. in the field at the University of North Carolina-Chapel Hill, recognized right away what was going on.

“I perceived the image of the Colorado River to be inverted, with the river appearing as a ridge rather than as a valley,” he wrote to me. “I rotated the page 180 degrees and it then appeared normal.”

What we had, he said, was a common phenomenon called “relief inversion,” an optical illusion that occurs when shadows are not where the mind expects them to be. Meehan sent a link to a short article that explains it:

“When looking at aerial and satellite imagery, telling the difference between a canyon and a mountain can sometimes be tricky…Most people have a subconscious interpretation of images based on the assumption that objects are illuminated from above. We expect the light source to come from the upper left corner of an image and when it doesn’t our brains flip mountains into canyons and vice versa to compensate.”

I’d never heard of relief inversion before, and even after Meehan pointed it out, it took me a few minutes to “see” the inverted version of the image. Now that I have seen it, it’s easy to see—but so is the original version. My mind keeps flipping back and forth between them, like looking at that optical illusion that shows two faces in silhouette—or wait, no, it’s a goblet!—no, now it’s faces again.

I asked the geoscientists who did the research described in the story, professor Jim Kasting and graduate student Sonny Harman, if relief inversion is ever a problem when trying to interpret images of other planets. Turns out, it is, and it requires careful examination to know what you’re looking at.

“As I understand it, you can solve the problem of topographic highs and lows by looking for shadows and knowing where the Sun is located relative to the spacecraft,” Kasting wrote. But even that isn’t always enough, he said: There are times when a riverbed looks higher than its surroundings because it is higher, as in these photos.

These shots, from a 2003 paper in Science, show meandering river channels in an alluvial fan deposit within a crater on Mars. Because the riverbeds are rocky, they are more resistant to wind erosion than the surrounding material. As a result, Kasting said, “the riverbeds are now higher than the remaining terrain—just the opposite of what would have been true when they formed.”

Harman said there have been many studies looking at how we interpret various kinds of images. According to this one (which provided the image at the top of this post), we not only tend to assume the light source is “overhead” (that is, coming from the top of the image) but also that main features in the image are convex. So we tend to interpret the shadows we see as being cast by a feature protruding from the surface, rather than falling into a depression in the surface. If the source of light in an image is anywhere other than the top of the photo, we’ll see a dip as a rise—or a canyon as a ridge.

It’s just really interesting how, even with the incredible technical tools we have today, we still depend very much on our brain’s perceptual systems and our ability to correctly interpret the data our tools provide us.

Members of the media who want to learn more about relief inversion or how much water there might once have been on Mars are invited to contact Jim Kasting at jfk4@psu.edu and Sonny Harman at ceh5286@psu.edu.