Lake Bakun, Malaysia. RapidEye image ©2016 Planet Labs, Inc. cc-by-sa 4.0.

Infra-Vision, a Satellite Superpower

Satellites have superpowers — they can survive in a vacuum, feed on sunlight, tirelessly observe the most remote corners of the Earth, and speed through space at over 17,000 miles per hour.

They can also see the unseeable.

Although many of the pictures from space you commonly see — Google Maps’ satellite view, NASA’s Blue Marble, or snapshots from the International Space Station — are ‘true color’ (taken using visible light), most Earth observation satellites also collect data with light invisible to humans — most commonly near infrared. Among other things, this type of light makes it easy to tell the difference between water and land (important for mapping), and can give a clear view of the Earth’s surface through smoke and haze that might otherwise block a satellite’s view.

Near-infrared light is just outside the range of human vision. It’s only a tiny bit redder (longer wavelength) than a red LED. This redder-than-red light behaves mostly like the light that makes up the colors of the visible spectrum. Near-infrared light travels through the air like normal light, and is usually observed after it’s been reflected off a surface. Other colors of infrared light—long-wavelength thermal infrared, for example—are less like visible light. In some ways, they’re downright weird—midwave infrared is blocked by water vapor, and brightness in thermal infrared is proportional to the temperature of a surface.

East Asia, Australia, and the Pacific Ocean seen in green (0.51 µm), near infrared (0.86µm), midwave Infrared (6.2 µm, sensitive to water vapor), and thermal infrared, by the Himawari-8 satellite. Notice how the green and near infrared views look similar, but the midwave and thermal infrared images are radically different. Provided courtesy of University of Wisconsin-Madison Space Science and Engineering Center.

While more exotic infrared wavelengths have their uses (read NASA’a Tour of the Electromagnetic Spectrum to learn more), a few key properties of near-infrared light makes it nearly indispensable for observing Earth from space. Most importantly, differences between the red & near-infrared light reflected by vegetation reveal how healthy the vegetation is. However, that’s a deep rabbit hole, so we’re going to explain vegetation indices in depth later, and focus on a few other properties of near infrared light at the moment.

Planet Labs operates two constellations of satellites, and both collect near infrared data—5 RapidEye satellites launched in 2008, and a flock of Doves that’s still growing. The RapidEye satellites provide a long-term archive, and the Doves will supply fresh data of the entire globe.

Because near infrared light is in principal similar to visible light—both are reflected from surfaces—most near infrared images of the Earth can be interpreted similarly to normal photographs.

True-color (left) and near-infrared false-color composite (right) images of Clinton, Iowa collected by a Planet Labs Dove on September 9, 2015. Vegetation strongly reflects near infrared light, so green fields and forests in the true-color image are bright red in false-color. PlanetScope images ©2016 Planet Labs, Inc. cc-by-sa 4.0.

Because near infrared light is invisible, it is usually displayed as false-color. The infrared picture (which is black-and-white) is substituted for red, while red is swapped with green, and green with blue. The results look weird, especially at first, but the images become familiar with a little practice.

True-color (left) and near infrared, red, green false-color images of the Earth from the Messenger spacecraft on it’s way towards Mercury. The heavily-forested Amazon is green in true-color, but bright red in near infrared false-color. Also notice that the oceans are a bit darker in the near infrared view, and the smoke over the Amazon virtually disappears. NASA/Johns Hopkins University APL/Carnegie Institution of Washington.

As you might have noticed from the Messenger pictures above, infrared light has some interesting (and useful) properties in addition to being sensitive to subtle changes in vegetation. In contrast to highly reflective vegetation, water strongly absorbs infrared light. In infrared false-color, the forested coast of Brazil (red) is easy to distinguish from the Atlantic Ocean (blue-black. Infared light also travels through haze more easily than visible light, so the pall of smoke over the Amazon Rainforest is easier to spot in true-color than false-color.

Compare these images of Bakun Lake in Malaysia. In true color the blue-green water appears similar to the surrounding green forest. This makes it difficult to trace the lake’s boundary. In the near infrared false color image, however, dark blue (almost black) lake water contrasts sharply with the bright red forest.

True-color and near infrared false-color images of Bakun Lake collected on April 28, 2016, by a RapidEye satellite. The images were processed to bring out maximum contrast and remove haze. Images ©2016 Planet Labs, Inc. cc-by-sa 4.0.

Notice how the water gets brighter where the Balui River enters the lake (lower right in both images). That’s due to the mud and silt suspended in the fast-flowing water. These sediments make the water more reflective in green and red, but not as much in infrared, so it is brown in true-color and blue-green in near infrared false-color. Likewise, water coursing through the Bakun Dam’s powerplant stirs up sediment in the river downstream (at the top of both images).

Blue, green, red, and near infrared images of Bakun Dam, Bakun Lake, and the surrounding forest. RapidEye images ©2016 Planet Labs, Inc. cc-by-sa 4.0.

The four images above show how the Bakun Dam looks in blue, green, red, and near infrared light. Each image was processed identically, and shows the relative reflectivity of different surfaces in each wavelength.

Clouds reflect almost all the visible and near infrared light that hits them, so they appear equally bright in all four images, and white in true- and false-color composites.

Clean water—for example the reservoir in the lower right corner—is somewhat reflective in blue and green, but dark in red, and nearly black in near infrared. As mentioned before, dirty, silt-filled water is brighter than clean water in all four bands, but still darker in near infrared than visible light, and easy to distinguish from its surroundings.

This is one of the superpowers satellites gain by being able to detect invisible light—the pictures they take have excellent contrast between water and land, making them great tools for mapping!

Composite true-color (red, green, blue) and false-color (near infrared, red, green) images of Bakun Dam. Each band was processed identically to correct for the color of sunlight and the human eye’s non-linear response to light, but without correcting the effects of the atmosphere. RapidEye Image ©2016 Planet Labs, Inc. cc-by-sa 4.0.

Another feature of these images is the color of the forest—it’s a tiny bit brighter in the blue band than it is in the green band, and appears aqua-colored in the true-color picture. Your eyes aren’t fooling you — the forest looks too blue in true-color, and the red vegetation is even a bit bluish in the false-color image.

The warm cliffs of the California coast near Santa Cruz, with the blue-ish hills on the shore of Monterey Bay visible in the distance. Photograph ©2016 Robert Simmon cc-by-sa 4.0.

The blueness is a result of the scattering of blue light from the molecules of the atmosphere itself—an effect so common that it’s built into our eyes and brains as a way to infer distance. Scattered light from the atmosphere fills in shadows, making distant objects appear washed out and pale. Blue (short wavelength) light is scattered more easily than red (longer wavelength) light. As a result, things in our field of view get bluer as they get further away.

The same things happen with satellite images, where light travels from the sun through the entire 60-mile- (100-kilometer-) thick atmosphere, reflects off the surface, and then all the way back through the atmosphere to a watching satellite. The redder (longer-wavelength) light is, the less it is scattered, so the view from space is often more clear in red and near-infrared wavelengths than it is in blue or green wavelengths.

Smoke from a large wildfire hovered over the Athabasca River and Fort McMurray, Alberta on May 14, 2016. In blue light (top) many features of the town and its surroundings were hidden by the smoke, but visible in near infrared (lower). RapidEye images ©2016 Planet Labs, Inc. cc-by-sa 4.0.

The ability to see invisible light is one of the qualities that makes remote sensing satellites so powerful. It allows monitoring of the Earth in ways that simply taking pictures in visible light does not—like detecting subtle changes in vegetation, accurately mapping shorelines, and peering through hazy skies. As Planet Labs’ archive of imagery continues to grow, these data will help people understand our planet, see it changing, and respond to that change—a satellite superpower.