What do the Color Images Mean?

Let's say that you are flying on an airplane between Philadelphia and Pittsburgh. You look out the window and see the West Branch River joining up with the Susquehanna River near Sunbury, Pennsylvania. Some of the diffuse scattered light from the Sun reflects off the ground and enters the pupil of your eye. Light from the Sun contains all the colors of the rainbow, and different objects on the ground absorb and reflect the different colors of light. Your eyes tell your brain what they saw, and your brain figures out that the rivers look blue and the trees look green.

Your camera does the same thing. When you take a picture with your camera, the camera shutter opens briefly, exposing the film to this reflected light. Then you take the film to be processed, and when you get the pictures back, the picture looks very similar to what your eyes saw. Now let's suppose that instead of a camera you have an imaging radar like SIR-C/X-SAR, which uses different wavelengths to observe the Earth. You are the envy of everyone on the airplane (and you have probably exceeded your baggage allowance).

As we have seen earlier, because radar operates at radio wavelengths, the picture you will take is a kind of "FLASH" picture. The radar will illuminate the ground with radio waves. The radio waves reflect off the ground, and some of the radio waves will be reflected back to the radar where it will record the result (like the film in your camera). If you have a single-wavelength imaging radar, you can visualize the radar image as a black and white image, where the whiter the image the brighter the radio reflection from the ground. But if you have a multi-wavelength or multi-polarization radar, you can make a color picture.

Unlike a camera which can take a picture of the entire spectrum of the rainbow, imaging radars have discrete wavelengths of radio waves that they can transmit. If your radar is a three wavelength radar like SIR-C/X-SAR, then it is like a camera that can only capture three colors of the rainbow. But with three colors, we can still make a color picture, by assigning different colors to different radio wavelengths. This is significant because the brightness of the ground typically changes with wavelength. This is just like what happens at visible wavelengths. A tree is bright at the green colors (wavelengths), and not as bright at the other colors, therefore the tree appears green.

Combining Images from Different Radar Wavelengths

A color image can be constructed by assigning just three colors (red, green, and blue). Therefore, in making a color image from a three wavelength radar, we could assign one wavelength band to be red, one wavelength band to be green, and the third wavelength band to be blue. The assignment of these colors is completely arbitrary, as radio waves don't have a "COLOR". However, by making some kind of assignment, we can visualize the radar image. The brightness of each color will depend on the brightness of the radar image for that wavelength.

For instance, say we are flying on the space shuttle over Sunbury Pennsylvania and you turn on the SIR-C imaging radar. Here are three black and white images corresponding to the three wavelengths of the SIR-C radar. They look like this :

"X-BAND"

"C-BAND"

"L-BAND"

As you can see, there are subtle differences between these images. For example, the "L-BAND" image has more variation in gray than the "X-BAND" and "C-BAND" images. The reason for this is that lots of different types of terrain appear rough on the scale of the X-band and C-band wavelengths (3 cm and 6cm). So, apart from the smooth river, which appears dark at all three wavelengths, most of the variation in the X-band and C-band images is due to changes in slope on the ridges of the Appalachians, which run through this area of Pennsylvania. If the terrain slope towards the radar, the backscatter tends to increase; if it slopes away from the radar, the backscatter tends to decrease. At L-Band, the agricultural fields in the valleys are smoother than the trees on the forested ridges, so we see some contrast between the tree-covered ridges (which are light gray) and the valleys, which appear darker.

Now, if we assign the red color to the "L-BAND" image, the green color to the "C-BAND" image, and the blue color to the "X-BAND" image, we get the following color image when we combine them:

As can be seen from this color image, there are significant differences in the "BRIGHTNESS" of the ground at the different wavelengths. The tree-covered ridges are redder than the valleys, because we assigned that color to L-band and the trees give a strong backscatter at L-band. Yellow areas are brighter in both the red and green images (L-band and C-band). It is somewhat easier to see these differences by looking at the color image than by looking at the black and white images of each wavelength separately. This type of image is sometimes referred to as a 'false color' image, since it does not represent the 'true' colors your eye would see, but an artificial selection of different colors assigned to different wavelengths.

Combining Images from Different Polarizations

When we transmit the radio waves, we can choose the polarization. Usually we decide to transmit the radio waves so that they are vibrating horizontally or vertically with respect to the plane of propagation. When the radar receives the radio waves, we can also choose which polarization it will receive, again usually horizontal (H) or vertical (V). Depending on the transmit and receive polarization, the ground reflects the radio waves differently.

Let's try combining three images of different polarizations. Below are three black and white images of Sunbury in Pennsylvania. Each image was illuminated with the same wavelength of radio waves ("L-BAND"). But the polarization of the radio waves is different for each image:

"L-BAND HH"

"L-BAND HV"

"L-BAND VV"

As can be seen, the differences are sometimes very subtle between these images. It is difficult to compare them for specific areas. However, if we form a color image by assigning the blue color to HH, the green color to HV, and the red color to VV, we can see more clearly the differences in the radar brightness for different transmit and receive polarizations :

In this image we see the tree-covered ridges stand out as a green color. That is because the HV polarization we selected for the green color is the best for observing the scatter off randomly oriented objects such as tree branches and leaves. So not only does the L-Band wavelength pick out trees from agriculture well, using L-Band HH polarization is even better.

We can also combine images from different polarizations and different frequencies. In this way we can examine the same image in different ways, and by using color we can highlight differences in radar brightness due to different wavelengths and polarizations of radio waves. For instance, we could take an L-band HH polarization image, an L-band HV polarization image, and a C-band VV polarization image, assign them the colors red, green, and blue respectively, and get the following color image:

LHH, LHV, and CVV Image

This combination appears to show the contrast between the tree-covered areas and the agriculture in the valleys even better than the previous one. This combination always seems to show up differences in land use well, so it is used a lot in generating color images from SIR-C.