Camargo Syncline

(154 kB, 512x512)

This is a perspective view of the Camargo syncline in central Bolivia. It was created from a true stereo pair of images taken one year apart (6 June 1985 and 16 June 1986) in Landsat channels TM-5, TM-4 and TM-2. The vertical scale has been exaggerated by about a factor of 10.

Image by K. Palaniappan, F. Hasler and H. Pierce (NASA/Goddard) from data provided by T. Gubbels (Cornell University).

How was this image made?

Landsat imaged the area from slightly different angles. (The images were taken about a year apart, but we assume the topography didn't change during that time.) The two different locations of the satellite act just like your two eyes, which are separated by a small distance and therefore can see in stereo.

A simple way to coregister the two images would be to mark corresponding points in each and then shift, rotate and/or stretch them to match. In this particular case, features were matched using an automatic pattern-recognition algorithm on a MasPar massively parallel computer.

A simple stereo image could be made by simultaneously displaying one in red and the other in green, and then viewing the result with inexpensive two-color glasses. A better, full-color result can be obtained by overlaying color images on a Silicon Graphics workstation and viewing with stereo glasses equipped with polarizing shutters which rapidly flick on and off for each eye in turn. (That type of stereo imaging is part of our activities here). Another option would be to view one image in each eye, say with a pair of virtual reality goggles.

The height of the terrain was computed from the two images using the altitude of the satellite, the angle between the two images, and basic trigonometry. Once the height data is known, the result can be plotted as a shaded surface or contour map. Various visualization software packages can be used to tilt and rotate the surface until the desired perspective view is obtained.

There were actually several images taken at different wavelengths on each date. The height field was computed in several channels (each more or less sensitive to ground, vegetation, and water) and an average of the result was taken.

For another perspective view created using satellite data, check out Hurricane Andrew.