Explained: Giant "Contrails" going from Africa to Australia

Mick West

Administrator
Staff member


Some strange white lines appeared on the iPhone "Living Earth" app this morning.

These are obviously not aircraft contrails, as they are far too long and straight. The most likely explanation is some kind of image processing glitch. But how did it get there?

Shortly after I took those screenshots, the lines were gone. Living Earth seems to update the cloud cover image every three hours or so. I snooped the traffic and found it was fetching the images from a CDN server with a URL like:
http://cdn.livingearthapp.com/maps/clouds/clouds-09-03-2014_12_00.jpg
Which looks like:

This is an equirectangular projection, a square image that can be wrapped around the globe simply by converting latitude and longitude into X and Y position in the image.

The actual lines are on the previous image
http://cdn.livingearthapp.com/maps/clouds/clouds-09-03-2014_09_00.jpg



The full globe cloud images are created from a patchwork of multiple sources. So it looks like this glitch came from the stitching together of a corrupt image with a good one. Sometimes there are transmission errors between satellite and ground. It looks here like it lost a few lines of the image.

Confirmation that this happens can be found in older images, like:
http://cdn.livingearthapp.com/maps/clouds/clouds-09-02-2014_21_00.jpg
(showing just the north half)


Here it's more obviously a digital glitch. You can also see it fade out at the left and the right as it is blended into adjacent images. Notice also the curvature, which matches the curve of the "trails", except inverted, being on the other hemisphere.

There are only a few sources of these images, most of which can be found here:
http://www.ssec.wisc.edu/data/geo/
You can select the satellite and the type of image from the drop-down box:


And a little digging gives us this image, from China's FY-2E, Sept 14, 10:02 UTC


And a bit more digging gives us:
http://dcdbs.ssec.wisc.edu/inventory/image.php?sat=FY-2E&date=2014-09-03&time=10:02&type=1 byte VIS&band=1&thefilename=fy-2e_1_2014_246_1002&coverage=FD&count=1&offsettz=0


So all we have here is some glitched data, which is creating these interesting lines when stitched together
 
Last edited:
Back
Top