Tutorial

The microwave radiation is transmitted from the JERS-1 radar antenna in Earth orbit, reflect off the Earth's surface, and some of the microwaves then make it back to the receiving antenna where the data is later transmitted back down to Earth. Because SAR is a side looking instrument, most of the radar pulses are reflected away from the the radar (except if there is a double bounce - I will get back to that). The sigma0 value is the ratio of the received backscattered energy to that of an isotropic scatterer (scatters the radar waves uniformly in all directions).

Figure from "What is Imaging Radar"

Usually the sigma0 values, because they are power ratios, are expressed in decibels (db). The db values are just 10*log(power), where power is the processed image values.

We have taken the sigma0 values and compressed them to 1 byte per pixel and saved the result as either binary or GIF images. Since we have saved them to 1 byte per pixel, the actual sigma0 values had to be scaled to range from 0 to 255. The formula used to do that is described in the Data Products Information Sheet .

The sigma0 values depend in general on the illuminating geometry and the nature of the scattering of the radio waves from the targets. Because the JERS-1 SAR is an L-band radar, the radar is sensitive to structures on the order of 23 cm (the wavelength of L-band). Often, the JERS-1 SAR can peer past the top layers of vegetation canopies to the surface below, though the tree trunks are good scatterers. The sigma0 values typically vary from -18 db (the noise level of the data) to values greater than one. Sigma0 can be greater than one (0 db) if the radio waves are preferentially reflected back to the radar (if there is a mountain, and the mountain face is toward the radar, for instance, i.e. specular reflection). Often urban areas are dominated by double bounce returns, as the radio waves relect off the streets, and then off the buildings, and back to the radar. In general, the brightness of the image depends on the illuminating geometry and the nature of the scattering of the radio waves of the targets:

Figure from "What is Imaging radar"

We can also get double bounce reflections off flooded forests when the radio waves reflect off both the tree trunks and the smooth reflective surface of the water. This means that the JERS-1 radar images often indicate where there are flooded forests.

On the other hand, when there is not much vegetation, the ground will be pretty smooth looking at the 23cm wavelength of JERS-1, and most of the radio waves will be reflected away from the radar - this will cause these areas of the image to look dark, and have small sigma0 values. For this same reason, rivers and lakes are quite dark in the image.

The forest areas usually are pretty bright, because the radio waves are diffusely scattered, with a goodly amount being reflected back towards the radar.

Interpreting the radar images

These images are not optical images like you get with a camera! They are images made from radar signal data - as you might imagine, the process to transform this radar signal data into images is a complex one. But, we are not going to go into that right now. Instead, let's discuss how to interpret the images after they are made.

There are several differences between these radar images and optical images such as you get from the Landsat series of satellites.

Radar	Optical
See through clouds	Can't see through clouds
Provides illumination	Illumination from Sun
Wavelength : 23 cm	Wavelength : really small!
"sees" how well radio waves reflect and scatter off structures	sees how well different colors of light are reflected
Radio waves can penetrate a forest canopy	Sees the color of the tops of the trees

There are two important things to keep in mind when looking at radar images that you don't usually have to think about when you are looking at an optical image : 1) the direction of radar illumination affects how the image is going to look; and 2) the radio waves interact with the surface - sometimes penetrating, sometimes scattering, sometimes reflecting off more than one target - how they interact determines what the image looks like.

So, what can we determine by looking at the radar images of the Amazon:

• If the image is dark, then this probably indicates the surface is smooth. Usually, this will be water or roads or pasture land. It is difficult to distinguish with the JERS-1 SAR the difference between these just by looking at the brightness (though with other radars with better SNR we can).
• If the image is really bright, then this indicates (usually) that we are getting some double bounce reflections. This can be caused by buildings in urban areas and flooded forests, among other things.
• If there are mountains or hills, the radar image will tend to be brighter on the side of the mountain or hill facing towards the radar. It will tend to be darker on the side of the mountain or hill facing away from the radar. The radar on these images is traveling from North to South and is looking at the Earth from the East.
• The rainforest is the medium brightness that you see in most of the images. The trees diffusely scatter a lot of radio waves back to the radar, but not as much as when you have double bounce reflections.
• The moisture content will affect how bright the image is. For instance, if pasturelands are wet, they will tend to be brighter than if they are dry (unless it is completely flooded, in which case it will be dark like a lake would be).
• Sometimes you have to guess what could be causing the image to look different than the surrounding pixels.