Earthshots: Satellite Images of Environmental Change

You are here

» Selkirk Island, Chile

In September 1999, engineers at EROS were reviewing Landsat 7 sensor output. An image grabbed their attention. Over Selkirk Island, about 500 miles off the coast of Chile, they noticed an unusual cloud pattern. They referred the image to EROS scientists, who identified the striking series of swirls as a “Karman vortex street,” a fluid flow pattern rarely seen so clearly outside of a laboratory.

How did these Karman vortices develop? Selkirk Island is the tip of a volcanic peak, rising sharply from the ocean, its steep sides critical to the formation of these vortices. Though covering only 33 square miles, the island rises over a mile into the sky.

On the day Landsat 7 acquired this image, the wind was carrying northward a layer of stratocumulus clouds (flat-bottomed puffballs). The mile-high island caused this cloud layer to slow about the island, while remaining fast farther out on either side. So on each “wing,” left and right, the air started rotating toward the inside—clockwise on the left, counter-clockwise on the right. The rotational momentum made each side swirl in on itself.

The whorl-cores were clear because the swirling pulled dry, clear air (from above or below) into the wet layer, a bit like the funnel formed when you stir up a pitcher of orange juice. These clear, spinning pockets trailed off down the “street” from the island like soap bubbles from a toy wand—drifting downwind, weakening, filling with clouds, and breaking up.

We ordinarily show as few clouds as possible in Earthshots because they block our view of the land, but in this case they do have scientific, as well as aesthetic, value. The American Meteorological Society featured the Selkirk image on the cover of their monthly bulletin. This image also became one of the earliest selections in the Earth As Art collection.

Map of the featured area.

Since its launch in 1999, Landsat 7 has not again seen such nice vortices over Selkirk Island. A single vortex formed on March 25, 2000, from a similar southerly wind, but then the pattern broke up. That morning the clouds were too unstable, as shown by the turbulent convection cells in the northeast.

In its first year of acquiring images, the whole island was never visible to Landsat 7. Even on a clear day like November 18, 1999, the island's heat and elevation heave up damp marine air into the cold, until it reaches its dew point and condenses into a kind of permanent parasol of clouds. This sometimes trails downwind a short distance on windy days, as on February 22, 2000.

On November 10, 2008, the island looked a bit more like an icebreaker ship plowing through ice pack; no vortex street this time, but the effect the tall mountain island has on the clouds is clearly demonstrated. The February 14, 2009, image almost displays a vortex, but it didn’t quite form completely.

April 19, 2009, could have been a rare clear view of the island, but some of the data are missing because of Landsat 7’s Scan Line Corrector (SLC) malfunction of May 2003. Landsat 8, which launched in February 2013, has been more fortunate in its relatively short time in orbit to capture at least three clear views of the island: on March 8, 2014, March 11, 2015, and January 11, 2017. Those three images look about the same except for subtle changes to the mountain shadows caused by a different sun angle at different times of year.

The Landsat images that are shown on Earthshots are combinations of three different wavelengths of the electromagnetic spectrum. Typically, the images show a combination of visible and infrared light; therefore, they are false color images. When these wavelengths, or bands, are shown individually instead of combined, different details sometimes emerge.

In the case of this Selkirk Karman vortex street scene, Landsat 7’s band 5 hints at a feature not visible in the image featured in the first section, which is a combination of bands 5, 4, and 3. Do you see it? Band 7 shows it even better. Northeast of the island, just east of the vortices, there is a bright spot and a bright plume trailing from it to the northeast. This may be a westbound ship, the effect of the ship’s exhaust shown in the clouds.


Have you noticed that in these images, clouds look white but the ocean looks black? Both are made of the same substance, water—why would they appear as opposite colors?

Landsat satellites see solar energy that reflects off the earth (or atmosphere) and back to the satellite. When light hits water, whether ocean or cloud droplet, most of the light reflects at the same angle it came in, like a basketball bounce-pass. Calm water lets the light bounce away, like a mirror, so little light reflects toward the satellite; the ocean looks dark. But a cloud’s millions of droplets bounce the light around like pinballs, so some light always scatters toward the satellite; therefore, the cloud looks bright. This is the same reason a choppy ocean with whitecaps looks brighter than a calm ocean.

Selkirk Island is not the only place where Landsat has seen vortex streets. Other places where this phenomenon is common are nearby Robinson Crusoe Island; the Kuril Islands, Russia; Guadalupe Island, off the western coast of Mexico; and the Aleutian Islands, Alaska. The reason for the cloud patterns is the same: tall, steep-sided islands that affect the motion of passing clouds. That first image over Selkirk Island in September 1999, however, is still the clearest vortex street Landsat has recorded.


Have a question or comment? Please contact us at