Synthetic aperture radar

Radar image (SIR-C / X-SAR) of the Teide volcano on Tenerife from the Space Shuttle Endeavor

A synthetic aperture radar (abbreviation SAR , German for " Radar with synthetic aperture ") belongs to the class of imaging radars and is used as a sensor for remote sensing. It is used like a side-looking airborne radar from airplanes or satellites and, like these, provides a two-dimensional representation of a section of the terrain by scanning the earth's surface with electromagnetic waves , but with a much higher resolution . All radar devices that do not use the SAR procedure are referred to as Real Aperture Radar ( RAR abbreviation , "Radar with real aperture").

The images generated by a SAR are relatively easy to interpret due to their similarity to photographic recordings and are used for earth exploration, mapping and reconnaissance purposes. In contrast to optical sensors, an SAR can be used in almost all weather conditions, since cloudiness in the atmosphere caused by fog, rain or snow weakens the microwave radiation far less than light rays. In addition, like any active radar sensor, an SAR can also be used at night. In this connection, one speaks of an active remote sensing system that illuminates the observed objects itself.

While the geometric resolution of a RAR deteriorates with increasing object distance due to the diverging antenna beam, a SAR can achieve a spatial resolution down to the meter and decimeter range, which is independent of the slope distance and wavelength under certain conditions.

The synthetic aperture

When a radar with a synthetic aperture is spoken of, what is usually referred to is the so-called focused SAR : additional focusing of the individual signals is achieved by the signal processor compensating for phase differences that arise from time-of-flight differences between the individual antenna positions. If no phase corrections are made to the echo signals, one speaks of unfocused SAR .

RAR images to generate a SAR image

How a SAR works

The SAR principle requires an antenna that is moved perpendicular to the direction of the beam and whose exact position is known at all times. The direction of movement is usually referred to as along track or azimuth and the transverse coordinate as cross track or range . In the literature, Along Track is also referred to as a cross range . Footprint is called the area that the real antenna currently detected Swath (dt., Swath ') the strip of ground, the footprint of the sweeps by the movement of the real antenna. The geometry corresponds to that of a simple side-looking airborne radar .

The principle of the synthetic aperture (here still unfocused) consists in replacing the snapshot of a large antenna with many images of a small, moving antenna. In the course of this movement, every object in the target area is illuminated from a varying angle and recorded accordingly. If the path of the real antenna is known with sufficient accuracy and the scenery is immobile, the aperture of a large antenna can be synthesized from the intensity and phase position of the received radar echoes, thus achieving a high spatial resolution in the direction of movement of the antenna. In practice, this can be imagined as a very large phased array antenna whose individual radiators are not connected in parallel, but whose positions are recorded one after the other by a small antenna. The radar signal processor connects the individual amplitudes and phase positions to one another as if a phased array antenna with a very large aperture had been used. The azimuthal resolving power is still dependent on the distance and is ½ ( λ ∙ R ) ^½ with R as the distance and λ as the wavelength used.

Modern computing technology enables the phase of the signal received from this location to be changed for each individual pixel being imaged. The SAR can correct the transit time differences between the individual antenna positions for each individual distance. Places that are closer to the radar, due to the trigonometric relationships , have greater differences in transit time than places that are further away. This difference in transit time is measured as the phase difference. From the recorded echo data, a separate synthetic antenna is calculated for each irradiated location, the angular resolution of which in azimuth is selected so that the geometric along-track resolution is the same for all distances considered.

The following phenomenon can be observed here: For the same angular resolution, a synthetic aperture only needs half the length of a real aperture.

A clear explanation for this is: In the case of a real aperture, changes in distance and thus measurable phase shifts of the radar echoes of an object moving in parallel from the perspective of the antenna always relate to the location of the center of the antenna. In the case of a synthetic aperture, the changes in distance and phase also have an effect as a result of the successively different positions of the real antenna along the synthetic aperture.

So that a synthetic aperture can be implemented, it is imperative that the radar system works fully coherently . This means that the phase relationship between the transmit and receive signals and from transmit pulse to transmit pulse must be known exactly. For this purpose, a highly stable frequency source is usually used, from which all required mixing and sampling frequencies and all temporally periodic processes are derived.

Derivation of the geometric resolution in azimuth

The best possible resolution that can be achieved with an SAR is equal to half the length of the real antenna in azimuth or flight direction, i.e. H. at a reduction of the azimuthal antenna length L _Az (top in the figure with L respectively) of the real antenna improves the dissolving ability δ _Az in accordance with:

(1) ${\displaystyle \delta _{Az}={\frac {L_{Az}}{2}}}$

To derive this, three flight positions 1 , 2 and 3 of the antenna moved in the azimuth direction are shown in the diagram above . As with RAR, the azimuthal angular resolution at wavelength λ is : ${\displaystyle \varphi _{Az}}$

(2) ' ${\displaystyle \varphi _{Az}={\frac {\lambda }{L_{Az}}}}$

Position 2 mark the location of the minimum distance from an object at point P to the flight path. If S _{0 is} the corresponding inclined distance, then the axis d _{Az of} the irradiated area has the length:

(3) ${\displaystyle d_{Az}=\varphi _{Az}S_{0}={\frac {\lambda S_{0}}{L_{Az}}}}$

The point P is irradiated not only from the middle flight position 2 , but also from every position between 1 and 3 . The distance M of the positions 1-3 thus corresponds exactly to the diameter d _{Az of} the antenna spot at the distance S ₀ in question . The SAR uses all information received from the object at point P , which originates from all recordings in the area M = d _Az . Computationally, after all values ​​have been recorded and stored, an antenna with the azimuthal length d _{Az is} simulated, which according to the above-mentioned property of the synthetic aperture with an antenna according to Eq. (3) halved resolution:

(4) ${\displaystyle \delta _{Az}={\frac {\lambda S_{0}}{2d_{Az}}}}$

d _Az is from Eq. (3) known. Replacing d _Az in Eq. (4) leads to Eq. (1):

(5) ${\displaystyle \delta _{Az}={\frac {\lambda S_{0}}{2}}{\frac {L_{Az}}{\lambda S_{0}}}={\frac {L_{Az}}{2}}}$

This means that the resolution of the synthetic aperture is independent of the wavelength and object distance.

Alternative description of the SAR principle

Another description of the SAR principle is provided by considering the Doppler shift of the echo signals reflected by an object: When entering the beam cone of the antenna, the echoes reflected by an object are shifted towards higher frequencies due to the decreasing distance. After passing the minimum distance (miss distance, exactly in transverse position), the distance increases again and the received signals are shifted to lower frequencies.

In the receiver, the center frequency of the echo signal is brought to zero by mixing it with the center frequency of the transmitted signal ( superhet or superposition principle). The remaining deviations from zero are called the Doppler frequency or Doppler for short. The Doppler course of the echoes of an object from initially positive values ​​through zero to negative values ​​is called Doppler history.

Every object at the same distance from the flight path also has the same Doppler history, albeit shifted in time, according to the arrangement along the flight path and the flight speed.

Objects at other distances, on the other hand, either have a shorter Doppler history if they are closer or, if they are further away, a longer Doppler history with the same frequency range, which is referred to as the Doppler bandwidth.

If the radiation angle of the real antenna is not too large, the Doppler history can be viewed as a linear course of the frequency over time, i.e. This means that the echo signal of an object with a center frequency of zero, mixed to zero, represents a linear frequency-modulated signal.

This signal form, known as (down) chirp , is present as a result of the pulsed and coherent transmission signal as a sequence of complex-valued individual values. If these individual values ​​are multiplied by corresponding values ​​of a similar chirp, but with increasing frequency (up-chirp), the phase rotations on which the frequency changes are based cancel each other out. The addition of the resulting individual values ​​now provides the result of the synthetic aperture for the specifically viewed object.

This process is called correlation. The correlation function to be generated appropriately for each distance is called a replica. In the ideal case, it corresponds to the complex conjugate echo values ​​of a point target.

While an adapted correlation function results in a constructive addition of all individual contributions, a non-adapted function only results in a random addition result. In this way, the echo of the observed object, which arrives at the radar receiver at the same time as the echoes of other, likewise illuminated objects, is filtered out of the signal mixture.

Alternative derivation of the geometric resolution in azimuth

The radar antenna moves uniformly and without acceleration at the speed v ₀ . If the constantly changing azimuthal angle that the direction to the object at P forms with the antenna axis, the Doppler shift of the echo signal of this object assigned to the angle is given by: ${\displaystyle \varphi }$

(6) ${\displaystyle f_{D}(\varphi )=2{\frac {v_{0}}{\lambda }}\sin \varphi \approx 2{\frac {v_{0}}{\lambda }}\varphi }$

The approximation is valid for angular openings of the real antenna that are not too large. The total Doppler bandwidth B _{D of} the echo signal is obtained by inserting the maximum azimuth angle used and subtracting the values ​​from one another:

(7) ${\displaystyle B_{D}=f_{D}\left({\frac {\varphi _{Az}}{2}}\right)-f_{D}\left({\frac {-\varphi _{Az}}{2}}\right)\approx {\frac {v_{0}}{L_{Az}}}-{\frac {-v_{0}}{L_{Az}}}={\frac {2v_{0}}{L_{Az}}}}$

The frequency of a signal of the duration T can be, at best, with a frequency resolution δ f  = 1 / T are determined. Applied to the SAR signal, this means that the best possible frequency resolution is determined by the available observation time. However, this is equal to the time that the radar needs to cross the distance M  =  d _Az :

(8th) ${\displaystyle T_{SAR}={\frac {d_{Az}}{v_{0}}}={\frac {\lambda S_{0}}{v_{0}L_{Az}}}}$

It is called the aperture time. Hence the frequency resolution is given by:

(9) ${\displaystyle \delta f_{D}={\frac {1}{T_{SAR}}}={\frac {v_{0}L_{Az}}{\lambda S_{0}}}}$

limited. According to Eq. (6) this Doppler frequency resolution corresponds to a spatial angular resolution of:

(10) ${\displaystyle \delta \varphi \approx {\frac {\lambda }{2v_{0}}}\delta f_{D}={\frac {L_{Az}}{2S_{0}}}}$

This corresponds to a spatial distance in azimuth of:

(11) ${\displaystyle \delta _{Az}=S_{0}\delta \varphi ={\frac {L_{Az}}{2}}}$

Hence this is the best possible resolution of a SAR in azimuth.

For the formation of the synthetic aperture one can imagine a filter bank in which:

(12) ${\displaystyle K={\frac {B_{D}}{\delta f_{D}}}=2{\frac {\lambda S_{0}}{L_{Az}^{2}}}=T_{SAR}B_{D}}$

Filters in a row cover the entire Doppler bandwidth. The echoes of an object appear one after the other at the output of each filter according to their current Doppler shift. If you capture these signals and add them in the correct time and phase, the result will have an amplitude K times higher than a signal at the output of a filter. The energy of this useful signal increases to the K ²-fold value, the energy of undesired signal components such as noise or echoes from neighboring objects, however, due to the random nature of the additions, only to the K- fold. This improves the signal-to-noise ratio (SNR = signal-to-noise ratio) - that is the ratio of useful energy to interference energy - also by a factor of K.

The value K = T _SAR B _D is called the time bandwidth product. As can be easily calculated, the resolution is equal to the synthetic aperture length divided by the time-bandwidth product and the airspeed divided by the Doppler bandwidth:

(13) ${\displaystyle \delta _{Az}={\frac {L_{SAR}}{T_{SAR}B_{D}}}={\frac {v_{0}T_{SAR}}{T_{SAR}B_{D}}}={\frac {v_{0}}{B_{D}}}}$

SAR example

Origin of the phase difference when viewed from different positions

In order to achieve an azimuth resolution of 1 m at a distance of 10 km, an aperture length of 10 km / 1 m = 10,000 wavelengths is required when using a real antenna. At 10 GHz transmission frequency, corresponding to a wavelength of 3 cm, this is around 300 m, which is a practically unrealizable quantity. As mentioned above, a corresponding synthetic aperture only needs to be half as long. The same resolution is thus achieved with echo data recorded along a distance of 5,000 wavelengths or 150 m. However, the real antenna must ensure that the object in question can be illuminated the entire way. This requires a real aperture length in azimuth of 10 km / 5,000 = 2 m.

From the length of the synthetic aperture (here in the example L  = 150 m), a virtual near and far field of the synthetic aperture of the antenna can be calculated. The border between the two regions is r _fern  ≈ 2 ·  L ²  / λ and here at about 1500 km. Only then would the electromagnetic waves from the individual source locations form a flat wave front. Most satellites have their orbit within this distance, so they are in the near field of the synthetic aperture. The distance to the target differs between the positions of the platform. If the target is on the central axis of the real aperture, the distance is less than if the real antenna has to squint from an edge position towards the target. This is expressed in a phase difference Δ φ . This means that a simple summation of the real components of the individual diagrams cannot be carried out, but the imaginary component must also be taken into account, as is necessary in the near field. This means that a phase correction has to be made in the image processing software for each individual pulse period in order to generate a sharp image, which leads to the term “focused SAR”.

According to Eq. (12) then 2 × 3 cm × 10 km / (2 m × 2 m) = 150, as it is according to Eq. (13) also has to be. At a flight speed of 100 m / s, the Doppler bandwidth is 100 Hz, the aperture time 1.5 s and the best possible frequency resolution is 0.67 Hz.

Resolution in range

As with RAR (also: Side-Looking-Airborne-Radar , SLAR), the image coordinate perpendicular to the flight direction (range) is generated by distance measurement. This is done by evaluating the different signal transit times of the echoes from objects at different distances. Such a measurement can only take place in the radial direction (= direction of propagation of the transmitted signal). So that a floor surface can be mapped in the transverse direction by a distance measurement, the antenna viewing direction must have a lateral component. This means that the flight path of an SAR (projected onto the ground) is always parallel to the near edge of the swath at a certain distance.

The resolution in the radial direction (slant range) is basically determined by the signal bandwidth of the transmitted signal used. In the case of steep angles of incidence, the achievable range resolution in the plane (ground range resolution) deteriorates according to the projection of the radial resolution path onto the plane ground. At an angle of incidence of 45 °, it is therefore 1.4 times worse than in the radial direction. In the case of normal incidence, a distance resolution in the plane is no longer defined.

Essential elements of a SAR

Pulse compression

→ Main article : Pulse compression method

In order to produce a pictorial representation of the flown area, it makes sense to select the ground range resolution comparable to the azimuth resolution. The decisive factor for the slant range resolution is initially the bandwidth of the radar signal sent: ${\displaystyle B_{\mathrm {R} }}$

(14) ${\displaystyle \delta _{\mathrm {Sl} }={\frac {c}{2B_{\mathrm {R} }}}}$

c is the speed of light . A signal bandwidth of 150 MHz is required for 1 m resolution.

Compared to the slant range resolution, the ground range resolution is all the more reduced as a result of the projection, the steeper the grazing angle ε of the incident beam is compared to the ground:

(15) ${\displaystyle \delta _{\mathrm {Gr} }={\frac {c}{2B_{\mathrm {R} }\cos \epsilon }}}$

For this reason, the range resolution is often selected to be correspondingly finer than the azimuth resolution (at 45 ° that is about 70% of the azimuth value).

In the first decades of radar development, unmodulated pulses were used; H. Signals, for example from a continuous signal (CW of Engl. Continuous Wave were 'cut') by a short high-buttons of the transmitter tube. Such a signal has a bandwidth that is inversely proportional to its duration:

(16) ${\displaystyle B_{\mathrm {Rect} }={\frac {1}{T_{\mathrm {Rect} }}}}$

Increasing resolution requirements therefore led to ever shorter pulses; Attempts were made to compensate for the reduced energy content by increasing the transmission power. Depending on the frequency range, 10 MW or higher pulse powers could be achieved. Increasing the pulse repetition frequency (PRF, Pulse Repetition Frequency) to improve the energy balance are often other considerations, such as a. the distance uniqueness, contrary.

Because the pulse power cannot be increased at will for technical reasons (dielectric strength of the components), the pulse compression method was increasingly used in the 1960s . For this purpose, a comparatively long pulse is changed in frequency during transmission. Most often, a linear frequency modulation (LFM) is used, in which the transmission frequency changes linearly from a lower limit to an upper limit (up-chirp) or vice versa (down-chirp). The term chirp comes from the fact that an acoustic LFM signal sounds like chirping. By the way, bats use this signal form in the ultrasonic range.

At the receiver end, this signal is converted into a short pulse corresponding to the bandwidth using suitable methods.

Pulse compression with a SAW filter

Was used at the start of analog SAW components (SAW = Surface Acoustic Wave , dt. Surface acoustic wave ) to the pulse expansion and compression. A short pulse excites a surface acoustic wave that runs over a substrate with dispersive properties. At the other end of the substrate, the different frequency components arrive at different times and thus form the desired LFM pulse. A similar SAW component with complementary characteristics is used for the compression, and the stretched pulse is compressed again over time to its original length while maintaining its bandwidth.

Since the mid-1980s when digital technology advanced into frequency ranges beyond 100 MHz, only digital signal processors have been used in practice. These use fast digital-to-analog converters , which synthetically generate the signal from previously calculated data - possibly in several frequency segments to be combined. Upon receipt, the echo is digitized uncompressed and the pulse compression is carried out using a correlation process in the computer. The advantage of digital technology is that the replica for compression can be obtained directly from the transmitted signal by looping it into the receiver. The deviations from the ideal shape contained in the transmission signal, for example due to distortions in the high-frequency transmission amplifier (HPA from High Power Amplifier ), are thus recorded immediately. The replica is generated by forming the conjugate complex function from the sampled data. A compression with this reference function corresponding to a filtering with a matched filter (engl. Matched Filter ) which, in a white background noise , the output with the highest possible signal / noise ratio (engl. Signal / Noise - shortly S / N ratio ) supplies.

The properties of the pulse compression correspond to those of the SAR signal in the Doppler range. Here too, the time bandwidth product (often greater than 1000) gives the shortening factor for the chirp signal as well as the gain in signal-to-noise ratio.

Finally, it should be noted that the bandwidth required for a specific resolution can be distributed over several pulses ( frequency step method). This reduces the costly bandwidth requirements for the radar components. At the same time, however, the complexity of the radar-internal control and the SAR processor increases .

antenna

Of the many known antenna types, only three are used in SAR applications:

Reflector antenna

This type of antenna is similar to the widely used satellite TV receiving antennas. The properties such as size, bundling ability, side lobe behavior u. a. are fixed at design time. A mechanical rotating device and / or several feed elements must be provided for a pivoting in space (for example with elevation or azimuth). The advantage of this type of antenna lies in its suitability for large bandwidths and a cost-effective implementation. The reflector antenna requires an HF power amplifier (HPA, High Power Amplifier ) as the source for the transmission signal. The practically required RF power ranging from about 1 to 10 kW can currently only tube amps, mostly traveling wave tubes ( Traveling wave tube amplifier , short TWTA ) are provided.

Passive array antenna

A phased array antenna is made up of many individual radiators that are arranged on a flat surface in a regular grid. Each of these radiators or a group of radiators is connected to a feed network via a phase shifter. The direction of view of the antenna can be electronically swiveled over a wide range (for fixed installations up to ± 60 °) by changing the phase shifter settings. The advantage is the practically instantaneous beam control, as is often required with multimode radar devices and special SAR modes. The disadvantage is the high costs compared to the reflector antenna. Large swivel angles and high signal bandwidths require special feed networks with a transit time that can be controlled in real time ( True Time Delay , or TTD for short ) in order to counter the dispersion of the signals. The passive array antenna also requires a central power source in the form of an HPA.

Active array antenna

This antenna, which has only recently been implemented, is an array antenna in which each radiator or small groups of radiators each have their own transmitter amplifier and receiver ( active electronically scanned array ). The agility of this antenna type corresponds to that of the passive array antenna, with an additional degree of freedom due to the selective switch-off possibility of individual transmission amplifiers. The high effort is justified by some advantages. The distributed generation of the transmission energy enables semiconductor amplifiers with a low operating voltage to be used. In addition, the failure of individual amplifiers does not render the entire system unusable ( redundancy ).

SAR processor

At the beginning of SAR technology in the 1950s to 1960s, there was only analog signal processing. SAW techniques were used for pulse compression and optical processors in the form of conical and cylindrical ground lenses were used for SAR focusing. The disadvantage: the lenses could only be used for a precisely defined geometry in terms of height and side distance. With this method it was possible to achieve resolutions in the meter range, but the lack of motion compensation only led to optimal results in exceptional cases.

It was only with the introduction of faster computers and analog / digital converters at the beginning of the 1980s that the SAR principle experienced the hoped-for breakthrough. Due to the initially meager computing power, there was a search for computing time-saving algorithms for SAR processing. The come first to the application principle was the range-Doppler processor in which the focusing mainly in the two coordinates by the fast Fourier transform (FFT of Engl. Fast Fourier Transformation was done). These processors still worked off-line with data recording and only provided the results after the respective flight. Other algorithms ( chirp scaling, frequency scaling and many others) are now available. These now allow real-time SAR processing, even for very special SAR modes (see below).

SAR focusing only produces a good result if the location of the antenna deviates less than approximately λ / 16 from the ideal flight path. At 10 GHz transmission frequency this is less than 2 mm! One of the most important tasks of an SAR processor for systems in air use is therefore motion compensation today. For this, the position and movement data is highly sensitive one hand, GPS -assisted gyro platforms recorded and evaluated and additionally applied autofocus calculation method to the inevitable deviations to recognize from an ideal trajectory and eliminate them. Autofocus methods require SAR image sections to be calculated multiple times in order to determine the motion errors. Therefore, the required computing capacity for real-time requirements is considerably higher than for systems without autofocus capability.

Special SAR procedures

Geometric SAR modes: a) stripmap SAR (standard); b) spotlight SAR; c) Scan SAR

Stripmap SAR

This procedure is the standard procedure: The antenna diagram is not panned in the cross track or range area. The swath lies parallel to the ground track (projection of the flight path onto the surface of the earth).

Squinted SAR

A SAR image can also be generated if the viewing direction of the antenna is not directed across, but obliquely forwards or backwards. The processing requires additional algorithms for the correction of the range and Doppler coordinates that do not intersect at right angles. In the pre-alignment and below the flight path, the SAR principle fails for reasons of principle.

Spotlight SAR

With this SAR method, the azimuth resolution is improved compared to the limit specified in (1) in that the antenna remains firmly directed at a specific target area (spot) for a longer period of time; is rotated accordingly in azimuth. This increases the time bandwidth product and consequently the achievable resolution improves. However, this is done at the expense of the total area that can be displayed, because the next spot can only be targeted at a distance determined by the observation time and the flight speed.

Scan SAR

Here one makes use of the agility of a passive or active array antenna by operating several strips at different distances and squint angles almost at the same time according to a sophisticated schedule.

Interferometric SAR

Due to the coherent SAR signal processing, the SAR is also suitable for three-dimensional images. For this purpose, a second antenna with a complete receiving line is installed at a low height above the SAR antenna. The complex SAR images coming from both receivers differ in phase due to the different echo paths. This phase difference can be used to determine the object heights and thus to create a three-dimensional terrain model. If a fixed reference point is available, differential interferometry can be used to determine exact heights with accuracies down to the mm range.
The method also works in systems with only one receiving antenna. For this purpose, one evaluates the recording data of two parallel flight paths interferometrically (two-pass interferometry). However, because of the time interval between the two recordings, moving elements are not recorded.

Polarimetric SAR

A polarimetric radar is able to send and receive waves of different polarization. From the polarization of the waves received or the change in polarization, further information about the mapped area can be obtained, which, for example, makes it easier to distinguish between forest and houses.

Inverse SAR (ISAR)

The ISAR is a reversal of the classic SAR principle: the radar antenna is fixed and the observed object is moving. The achievable resolution is determined by the time-bandwidth product of the echo signals. The method is used, for example, to image satellites. Furthermore, a ship moving in rough seas or on its own can be mapped by ISAR in such a way that the type of ship can be identified.

Bi- and multistatic SAR

With bistatic or multistatic SAR, the transmitter and receiver are mounted on two or more carrier platforms. Thus more information about the backscatter properties can be obtained with more flexible angles of incidence and emergence. A technical difficulty is the synchronization of the oscillators. New methods must also be used in the processing.

Special features of the mapping by SAR

Foreshortening: the front flank of a mountain appears shortened

The images obtained by means of SAR have some special features that must be taken into account during the evaluation:

Shortening (Engl. Foreshortening )

Foreshortening is a shortened representation of actual distances (compression of distances). Imagine a mountain being scanned by the radar beams of a SAR. The base of the mountain (in the picture: a) first reflects the radar beams, then the summit (in the picture: b). If the two points in time of the reflection are very close together, the actual distance (a - b) between the base and summit of the mountain is shown compressed (a '- b'). This effect makes the interpretation of a mountain landscape difficult.

Overlay (English lay-over )

With a tall object, such as a tower, the top of the tower is closer to the radar than the base. The top of the tower is shown earlier, i.e. closer. This creates the impression of an overhang in a radar image: point b 'would be shown before point a'. As with foreshortening, this can lead to difficulties in interpreting images of mountainous terrain.

Shadows

Due to the illumination by means of a "light source" carried along, the images show shadows, i.e. places without reflected echoes. As with optical imaging, these arise where areas are shaded by higher objects from the radar beam. The effect is more pronounced the flatter the grazing angle and the higher the object casting the shadow. On the other hand, the shadows also allow a good interpretation of the three-dimensional images. A grazing angle of 5 ° is the lower limit for easily evaluable SAR images.

Moving target shift

A moving object is shown in the wrong place. This is done because the Doppler offset of the moving object is added to or subtracted from the Doppler history of a fixed object. However, this corresponds to the history of an object arranged later or earlier. An object moving as seen from the satellite appears closer in the azimuth direction. : The SAR photo on the right was taken by a satellite that flew to the north, i.e. from the lower to the upper edge of the image, and pointed its SAR sensor to the east, i.e. to the right. Ships can be seen as bright reflections. Oil secretions on their way dampen surface waves. From there, radar radiation is only slightly reflected, and the lane appears black. Pronounced bow waves can be seen. Ships moving towards the satellite from right to left appear offset upwards in the direction of flight of the satellite. The ships are above their lane. Accordingly, the ships moving to the right appear in the lower part of the image below the dark navigation line.

Speckle

Speckle is the peculiarity of a coherent image that flat objects such as cultivated fields can take on completely different values ​​from pixel to pixel due to the random composition of the echoes from individual contributions. Pictures with speckle therefore appear torn and grainy. Speckle can be reduced by using the multilook method, at the expense of dissolution. To do this, several poorly resolved SAR images are calculated from different Doppler areas and then added incoherently (in terms of energy). The random distribution of the values ​​of an area pixel ensures a reduction in the speckle.

SAR applications

Due to its versatile application possibilities, especially in remote sensing, SAR has achieved global importance, so that the establishment of a separate conference specifically focused on SAR seemed necessary. The Eusar is mainly dedicated to the radar sensor, its technologies including image-generating signal processing and image processing, but also offers a forum for users of SAR data. To this day, EUSAR is the only conference in the world that specializes in SAR.

Airplane SAR

Airborne SAR systems are mainly used for military reconnaissance due to their all-weather capability. The currently (2005) technically achievable geometric resolution is less than 15 cm, which requires an RF bandwidth of more than 1 GHz. Since reconnaissance radar has several operating modes, this system always works with electronically pivotable passive or increasingly active array antennas with lengths of 1–4 m in azimuth. Great importance is attached to motion compensation and real-time capability, i. In other words, the systems generate high-resolution images on board and transmit them to the evaluation points on the ground. The computing capacity required for this requires the majority of the resources available on board, both in terms of installation volume and primary energy. See also SOSTAR-X .

Another class are mini-SARs for use on board cruise missiles (drones). Here, the smallest possible construction volume with high resolution (<1 m) and moderate strip width (1–3 km) is required. In the meantime, the required processor capacity can also be installed on board for these applications, so that only final results that have already been processed need to be transmitted to the ground via telemetry . At high carrier speeds, the measures required for motion compensation are low, so that the processor on board is only loaded relatively little by this sub-task.

In civilian terms, the SAR is used almost exclusively for mapping purposes, almost always in the form of the interferometric SAR on board turboprop aircraft. Their geometric resolution is usually in the range 0.5–2 m. The Jet Propulsion Laboratory uses a Gulfstream III u. a. to research the consequences of the oil spill in the Gulf of Mexico .

Satellite SAR

SAR image of the Indian Ocean.

Seasat, the first SAR satellite for earth observation

Initially, satellite SAR was implemented as a pure research project; it is currently entering the phase of increasing military and civil use. The military requires the reconnaissance of every point on earth within given times with resolutions in the range below 1 m. This requires several satellites with the same equipment and coordinated flight paths. In order to keep costs under control, compromises in terms of equipment are inevitable. The active array antenna, which was still required years ago in practical systems ( e.g. SAR-Lupe ), has long since given way to a simple reflector antenna .

On the civil side, the research-oriented SAR systems of the past are gradually being replaced by commercial offers for mapping customer-specific areas. Here, too, the need to reduce costs leads to the preference for systems that are as simple as possible.

The photo on the right shows a sample SAR image. The small white dots are oil stations, the black areas are thin oil films. The long-period water waves above are so-called inner waves, the small waves with wavelengths around 100 m, see lower arrow, are surface waves generated by the wind. The detail resolution of the 25 × 34 km² area is better than 100 m.

The planet Venus was mapped by the space probes Venera 15 , 16 and Magellan using the SAR method . The Cassini-Huygens space probe uses the SAR method to map Saturn's moon Titan .

Further information on satellite SAR systems that have already been implemented or are being developed can be found on the pages listed here:

In earth orbits

• Seasat
• European Remote Sensing Satellite 1 & 2
• Envisat
• Shuttle Radar Topography Mission
• RADARSAT-1
• RADARSAT-2
• TerraSAR-X
• TanDEM-X
• SAR magnifying glass
• COSMO-Skymed
• Active Electronically Scanned Array
• Lacrosse (satellite)
• YaoGan WeiXing-1
• SAOCOM 1A

To other heavenly bodies

• Venera 15 and 16
• Magellan
• Cassini-Huygens

Web links

• Synthetic Aperture Radar on Radartutorial.eu

Bibliography

A clear presentation of the topic can be found at:

• Shahan A. Hovanessian: Introduction to Synthetic Array and Imaging Radars. Artech House, Dedham MA 1980, ISBN 0-89006-082-7 .
• George W. Stimson: Introduction to Airborne Radar. Hughes Aircraft Co., El Segundo CA 1983.

The SAR principle is dealt with in the books of the Radar "Pope" Skolnik in individual chapters with compilations of radar applications and technologies:

• Merrill I. Skolnik: Introduction to Radar Systems. 2nd edition. McGraw-Hill Kogakusha Ltd., New York NY et al. 1980, ISBN 0-07-057909-1 .
• Merrill I. Skolnik (Ed.): Radar Handbook. 3rd edition. McGraw-Hill, New York NY 2008, ISBN 978-0-07-148547-0 , (Chapter 17).

More papers based on signal processing theory can be found at:

• David K. Barton: Radars. Volume 3: Pulse Compression. Artech House, Dedham MA 1975, ISBN 0-89006-032-0 .
• Donald R. Wehner: High Resolution Radar. Artech House, Norwood MA 1987, ISBN 0-89006-194-7 .

Individual evidence

1. ↑ Louis J. Cutrona: Synthetic aperture radar. In: Merill I. Skolnik (Ed.): Radar Handbook. 2nd edition. McGraw-Hill, New York NY 1990.
2. ↑ What is UAVSAR? , uavsar.jpl.nasa.gov, accessed April 19, 2020.

This version was added to the list of articles worth reading on December 1, 2005 .