How are SAR images acquired?

The family of satellites that have carried, are carrying, or will be carrying SAR sensors for commercial applications is illustrated in Figure 1.

All satellites equipped with SAR sensors orbit the Earth on a near-polar orbit at an altitude ranging from 500 to 800km above the Earth’s surface, depending on the satellite platform hosting the SAR sensor. The  time taken for a satellite to re-pass over the same area is called the 'revisiting time'. Learn more about the individual characteristics of available SAR satellites. 

The angle between true north-south and the satellite orbit varies slightly, depending on the satellite, but in general lies in the range of 10 degrees. The circumpolar orbits of all SAR satellites mean that for half of their trajectory they are travelling from the north pole towards the south pole. This direction is referred to as a descending orbit. Conversely, when the satellite is traveling from the south pole towards the north pole, it is said to be in an ascending orbit. Figure 2 illustrated this concept.

Data from different satellite sources can generally be purchased or ordered without limitation, with the exception of the Japanese Space Agency (JAXA). Although ALOS-PALSAR data can be purchased, JAXA selected the acquisition modes of the satellite at the beginning of the mission for the entire duration of its operating life, so the user cannot select the radar acquisition mode most suitable for the application at hand.

Another important point to be considered is that the Italian COSMO-SkyMed constellation is a joint military / civilian mission. Whenever a conflict arises between acquisition requests, commercial projects have lower priority.
 

Figure 1

Figure 2

As the satellite circumnavigates the Earth, it continuously emits millions of radar signals toward the Earth’s surface along the radar beam’s line of sight (LOS). Upon impact with the Earth’s surface, some of the signals are reflected away from the satellite, some are absorbed by vegetation or other non-reflective materials, and some reflect back to the satellite.

Signals reflected off the Earth’s surface are also referred to as backscattered signals. Processors on-board the satellite integrate the returning signals to form a strip map. Usually, the on-board memory capacity is limited so the satellite transmits the data to strategically located ground stations. These stations then compile images which can be used for data analysis.

Figure 3 is a schematic of a satellite acquiring strip map images.
 

Figure 3

Radar signals are transmitted in pulses. In any SAR system, there are three important frequencies that define its operations. The so-called Pulse Repetition Frequency (PRF) is the rate at which those pulses are transmitted and defines the resolution of the system in azimuth direction (i.e. the direction parallel to the satellite velocity), the central frequency (f0) defines the operating wavelength of the system and characterizes its propagation and penetration features, as well as the sensitivity of the system in interferometric applications. Finally, radar pulses backscattered by the Earth surface are sampled by the radar system at another frequency (fs) defining the nominal pixel size, sometimes referred to as the resolution cell size, in the range direction, related to the sensor-to-target distance.

Satellite sensors are mounted on their platforms with the direction of transmission at 90° to the flight direction. The earlier satellites (ERS, Envisat and Radarsat-1) were all right-looking satellites, meaning that microwave beam transmits and receives on the right side only of the satellite, relative to its orbital path, i.e. the system cannot rotate. Newer satellites (Radarsat-2, COSMO-SkyMed and TerraSAR-X) have both right-looking and left-looking capabilities, thus they can ‘look’ to the right or the left of the craft, but not both directions simultaneously.

The angle at which the sensor is pointed toward the earth’s surface is referred to as the off-nadir angle (or look angle) – see Figure 4. The off-nadir angle of the ERS satellites was fixed at about 23°, but all subsequent satellites were fitted with the means to vary the viewing angle of the sensors, ranging from values of 20 to 50 degrees. This ability to vary the off-nadir angle is important in that it is possible to adjust for hilly or mountainous terrain (potential impediments to InSAR) if the relationship between viewing geometry and terrain slope is not optimal.

Figure 4

Radar signals are characterized by two fundamental properties: amplitude and phase.

Amplitude is related to the energy of the backscattered signal. When a signal is emitted from the SAR sensor it is broadcast at a specific energy level. On reaching an object on the ground surface, the energy level is changed depending on a number of circumstances that relate primarily to the reflective quality of the object. Metal and hard objects (natural or artificial) have a high reflective quality and therefore the amplitude of the reflected signal is much higher than the background noise of the SAR system. Softer materials such as wood, crops, asphalt, have a lower capacity to reflect incident radar energy and so the amplitude of the reflected signal is strongly diminished.

The amplitude characteristics of signals can be visualized. An individual amplitude image will appear speckled. This is because each resolution cell is composed of many scattering elements, all reflecting incoming signals back to the satellite with different signal strengths and slightly different delays (phases), creating the spotty appearance. From one image to the next, the speckle in corresponding resolution cells can be constant or can vary. Constant levels of reflectivity, often bright spots, are indicators of stable reflections of radar signals. When the speckle varies from image to image, the pixels are decorrelated across the data set and the speckling can be minimized by averaging the amplitude of all images within the stack. The result, referred to as a Multi Image Reflectivity (MIR) map, is a means to improve the clarity of the amplitude response of the stack, highlighting those pixels that have a stable and high reflectivity in each of the images within the stack. Figure 5 represents a single amplitude image and a MIR image comprising of a stack of 60 scenes.

As well as amplitude values, radar systems record phase values – the key element in any interferometric measurement given that it is related to the sensor-to-target distance. As has already been mentioned, radar signals are characterized by a certain frequency of operation and, for interferometric applications, they can be thought of as sinusoidal waves: one complete cycle (from –π to +π) corresponding to the wavelength. It is this specific property of the radar signal, and the system’s ability to record both amplitude and phase information for each image pixel, that are used in estimating displacement.

A signal’s phase can be affected by changes in the atmosphere as it travels from the satellite to the Earth and back to the satellite. In the atmosphere, layers of moisture (cloud, fog, rain etc.) are present, through which the signal must pass. In arid areas, these layers have less effect compared to tropical and temperate areas. As a signal encounters a moisture-bearing layer in the atmosphere, the propagating speed of the signal changes. As a result, errors are introduced into the phase values recorded by the receiving sensor on the satellite. Since wavelength and signal phase have a simple and direct correlation, any change in wavelength corresponds to a change in the phase of a signal. This is an important issue in measuring ground movement given the direct relationship between ground displacement and signal phase.
 

Figure 5