Spies That Fly homepage

In a way, SAR is like a camera. But instead of using light to create an image, SAR relies only on radar signals it sends out. This allows it to produce images in total darkness. It also takes advantage of radar's penetrating ability, allowing it to "see" through clouds, rain, snow, and fog. Creating a SAR image, however, involves mounds of data and tremendous computing power.

In general, radar works by measuring the time it takes a radio wave to travel from an antenna to an object and back to the antenna. Based on the time interval, radar calculates the distance to the object. But this type of radar can only locate an object and assess its speed. To create an image, a significantly greater amount of data is needed, and that requires an extremely long antenna—much longer than any plane could carry. SAR synthetically creates a long antenna by moving a short antenna a long distance. It stores all of the data it receives until it has enough to start processing it.

As the plane flies, its SAR antenna sends out pulses of high-frequency radio waves (radar waves) toward the ground. In between the pulses, the antenna receives "backscatter"—radar waves that have bounced off objects on the ground. All backscatter pulses hold information, including how long each pulse took to make the round trip to and from the plane and if the SAR is moving toward or away from the object on the ground. If the pulses are spaced ever closer together in time after bouncing off the object, the SAR antenna is moving toward the object. If they are spaced ever farther apart in time, the SAR antenna is moving away.

From the travel time of each radar pulse a computer can calculate the distance to the object it bounced off, and from the time interval between adjacent pulses it can calculate how fast the pulse was moving toward or away from the object. The computer now has to figure out, from all the data it has stored, which pulses have bounced back from the same location. It does this by looking for a pattern. The computer is programmed to know that if an object is moving closer and closer at a set rate (and then farther and farther away at a set rate), then the time interval between pulses should also change at a set rate.

When the computer locates, from the huge amount of pulse data stored, certain pulses that fit a specific pattern, it knows that these pulses bounced off the same location. Since it knows the distance from the exact location where the plane was when it sent and received the pulses that fit the pattern, it's a simple matter to plot the point of the object on the image it's creating.

The point on the image is plotted, but the brightness of the point is still not known. SAR determines the brightness by the intensity of the signal it receives. An object such as a tree absorbs some of the radar energy and so it appears gray. A metal object oriented toward the SAR antenna reflects a lot of the energy back, so it appears bright.

With actual SAR processing, about a thousand calculations are performed for every single pixel. And an image like the one shown above of Washington, D.C., is made up of several million pixels.