GB2104753A

GB2104753A - Radars

Info

Publication number: GB2104753A
Application number: GB08120517A
Authority: GB
Inventors: Stephen Rotheram
Original assignee: Marconi Co Ltd
Current assignee: BAE Systems Electronics Ltd
Priority date: 1981-07-02
Filing date: 1981-07-02
Publication date: 1983-03-09
Also published as: GB2104753B

Abstract

A surface imaging radar produces a two dimensional image pattern representing the earth's surface. Radar echo signals are processed to generate an image which represents the unwrapped phase for each point in the two dimensional pattern. The term unwrapped phase of a signal is its true phase value relative to some reference and includes multiples of 2??? in addition to the principal phase value. The unwrapped phase value can be obtained from the phase derivative of radar echo signals. This enables information on the nature of the surface, as well as its position, to be derived.

Description

SPECIFICATION Radar This invention relates to radar and is particularly applicable to surface imaging radars in which a two dimensional date pattern or visual display of, for example, part of the earth's surface is produced.

Radar imaging of surfaces is useful for a number of applications such as for crop monitoring, land use surveys, ocean surveillance, pollution monitoring, ship routing, iceberg tracking, storm tracking etc.

This invention seeks to provide an improved radar.

According to this invention, a radar system includes means for producing, from a radar echo signal relating to a radar target lying within a range or bearing cell, a signal representative of the unwrapped phase of the echo signal, and means utilising the signal to produce a radar image of the target.

The term target is used herein in a general sense to include anything which, when intercepted by a transmitted radar signal, returns an echo to a radar receiver. The unwrapped phase of a signal is its true phase relative to some phase reference and includes multiples of 27r in addition to the more common principal phase value which is restricted to the range 0 to 27r radians. A quantity equivalent to the unwrapped phase is the phase derivative of a signal, since from such a value the number of multiples of 27r can be deduced as well as the principal phase value.

The invention is particularly applicable to the imaging of a sea surface, and enables a two dimensional radar image to be built up from coherent phase detection of radar echoes. Previously coherent detection has been used to enable amplitude variations of echo signals to be extracted, but knowledge of the true (or unwrapped) phase has been discarded with a two dimensional image is synthesised.

The phase of the signal scattered from a rough surface contains information about the surface. In the case of land this phase variation is simply proportional to the surface elevation and is caused by the varying path length from the radar to the surface. Thus, in this case a phase image simply gives the surface elevation. In the case of sea, the fluid particles in a sea wave travel in different directions at different points along the wave. These fluid motions are known as wave orbital motions and act to Doppler shift the scattered signals. A Doppler frequency shift may be regarded as just a phase variation, so that the scattered field contains a phase variation related to the wave orbital motions. Having produced a phase image of the surface it is possible to further process the phase image to produce a true replica of the sea surface.

The invention is further described by way of example with reference to Figures 1, 2 and 3 which are explanatory diagrams, Figure 4 shows in schematic fashion a synthetic aperture radar, Figures 5, 6 and 7 show parts of a radar system in accordance with this invention, and are used for purposes of explanation and, Figure 8 shows a radar processing system which enables unwrapped phase information to be produced and used for imaging applications.

To obtain the phase it is necessary for the radar signal to be coherently detected. This means that a stable phase reference must be available and be phase locked with the carrier wave of the transmitted radar signal. In coherent detection two outputs are formed. In one the signal is mixed with the reference wave to form an in-phase or I output. In the other the signal is mixed with the reference wave after a phase shift of 900 has been imparted to the reference and forms a quadrature or Q output.

The two outputs I and Q can be conveniently represented as the real and imaginary parts of a complex output S=l+iQ. As an equivalent alternative one can represent S by amplitude A and phase 0 as S=Aeif. The phase determined from I and Q lies between 0 and 27r. This is known as the principal value of the phase, and differs from the unwrapped phase as shown in Figure 1. If one transmits a continuous radar signal from a platform which moves, for example, the complex output S will vary continuously in the complex plane and might trace out a curve such as ABCD in Figure 1. If the phase is zero at A it will be 3600 or 2z radians at B.The principal value of the phase then starts at zero again, increases to 27r at C, starts at zero again and finishes at the value 2 or 900 at D. The unwrapped phase is zero at A, 2n at B but then carries on increasing continuously to become 4# at C and 9#/2 at D. This is illustrated in Figure 2 in which the principal value and the unwrapped phase are shown varying with time.

Suppose the unwrapped phase is (t) varying with time t and one requires 0(t) from t=O to t=T with 0(0)=0. An entirely equivalent representation is the phase derivative b'(t) from t=0 to t=T as 6(t) can be found by integration;

Thus > (t) and 0'(t) can be considered equivalent. As radar echo signals will normally be sampled, the relation between (t) and sfi'(t) becomes numerical integration and is approximate, but a phase derivative image can be considered to be equivalent to an unwrapped phase image.

Let the signal S(t)=l(t)+iQ(t) and its derivative dS(t)/dt=S'(t)=l'(t)+iO'(t) be known from t=0 to t=T. The phase derivative is now given by I(t)Q'(t)-Q(t)I'(t) #'(t)=-- I(t)+Q(t) Thus a knowledge of a signal and its derivative allows the phase derivative to be found.

Often the Fourier transforms s(w), or some equivalent quantity, of the output S(t) is known.

Indeed S(t) may have been obtained from s(w). For example pulse compression and SAR (synthetic aperture radar) processing can be interpreted as the weighting of a set of samples giving s(w) followed by Fourier transformation to give S(t). These are related by

In such a case the derivative of the output is also simply found as the Fourier transform of -iws(w)

Figure 3 shows in a simplified manner one way of converting a set of samples of s(w) into s(t). In this Figure s(# and -i#s(#) are Fourier transformed to give S(t) and S'(t). From these the phase derivative 0'(t) is found. This is integrated to give the unwrapped phase b(t).

Figure 4 shows a synthetic aperture radar which is suitable for implementing the invention.

Imaging radars utilise real aperture or synthetic aperture. Real aperture systems do not need a moving platform, but a common type is the side looking airborne radar, as in Figure 4. Synthetic aperture radars need a moving platform and are commonly implemented as a side looking airborne radar as in Figure 4 or as a spaceborne system. The antenna is on the side of an aircraft 30 which flies along a path 31 and radiates a beam 32 which illuminates an area of the surface called the radar footprint 33. As the aircraft 30 moves along its path 31 the radar footprint 33 sweeps out a strip 34 called the illuminated swath. It is this swath that is to be imaged. The flight direction, which is the x direction in Figure 4, is called the along-track or azimuth direction whilst the orthogonal direction which is the y direction is called the cross-track or range direction.Resolution in the range direction may be achieved by timing a short pulse from transmission to reception. If the pulse is of duration T then the resolution in the range or y direction will be cz/2 sin 0 where c is the velocity of light and 9 is the angle to the vertical as in Figure 4. In practice pulse compression is often used, in which case a long frequency modulated (chirp) pulse is transmitted to cover the entire swath. The chirp pulse has length T and bandwidth B with Bz much larger than unity. This excess bandwidth may be used to correlate the received pulse with a reference waveform and produce a compressed pulse of length I/B.If BT is greater than n, the received pulse can be compressed into n independent short pulses corresponding to n ranges y1, y2...., yn. Suppose the aircraft is at x=xi when the pulse is transmitted. As the aircraft velocity v is much less than the velocity of light c, the aircraft position is essentially unchanged during the transmission and reception of a single pulse. Suppose the received voltage is P(xj, t). This is fed into n range gates and correlated with n different reference waveforms to produce n output signals S(xj, y1), S(xp y2),.. . S(xl, y,). The time within a pulse is related to range y and a simple flow chart illustrating the generation of S(xp y,) for n range cells is shown in Figure 5.

In a real aperture system (as opposed to a synthetic aperture radar) resolution in azimuth or x direction is given by the width of the beam. If the beamwidth is P then the width of the beam on the surface is RP where R is shown in Figure 4. If the interpulse period is At then the aircraft travels a distance vAt between pulses. Thus the aircraft transmits pulses at x1, x2,... x1... and a distance vAt separates each position. Thus the outputs S(xj, yk) with i=1, 2... and k=1, 2... n form a set from which an image can be formed and this is conventionally formed in known imaging radars as an amplitude pattern from the squared modulus 15(Xj, yh)l2.

The correlation of the reference waveforms with the received voltage P(xj, t) may be written S(xj, yk)=#dtP(xj, t)Ak(tk-t) where Ak is the kth reference waveform.

Here the ranges y, and times tk are related by ctk yk= +constant 2 sin 9 It follows that

It follows that correlating the received voltage with the time derivative of the reference waveform produces an output proportional to the range derivative of the signal (range and time are virtually interchangeable in this context). The signal and its range derivative may now be combined to give the range derivative of the phase.

with S(xj, yk)=I(xj, yk)+iQ(xyr yk). This arrangement is illustrated in schematic form in Figure 6. It is functionally the same as Figure 3 is one interprets Fourier transformation with correlation and shows how the phase derivative for a real aperture system can be formed.

The azimuthal resolution of a real aperture system is Rss, and the beamwidth of an antenna is approximately ss-=#/D where # is the radar wavelength and D is the length of the antenna in azimuth.

Thus the resolution is RAID. This can be made smaller by using smaller wavelengths, longer antennas or shorter ranges but there are physical limits, for example one cannot carry an antenna more than a few metres long on an aircraft. This becomes a severe limitation for spaceborne systems for then R is very large and resolution can be very poor. This is overcome using the synthetic aperture radar. If phase coherence in the system can be maintained over a time of order Rss/v, which is the time that a point on the surface stays in a beam, then the outputs from successive pulses at a given range yk, S(x1, yk), S(x2, yk)... can be thought of as the radar echo returns that could be received by an array of antennas at xl, x2 etc.In this way a very long antenna can be synthesised of length about Rss and its azimuthal resolution may be shown to be D/2. Suppose a synthetic aperture is to be formed from 2m+1 successive outputs S(xj-m, yk), S(xj-m+1, yk),... S(xj, yk),... S(xj+m, yk) centred on S(xj, yk). These are then weighted and added as in an antenna array to give an output V(xj, yk). This amounts to correlating the samples of S with an azimuthal reference waveform. The process is illustrated in Figure 7.The outputs from the system in Figure 5 go into a store from which the outputs S(xj+1, y, with I going from -m to +m, are removed and correlated with the azimuth reference waveform to give the output V(xj, Yk).

The unwrapped phase of V(xj, yk) is denoted by #(xj, yk) to distinguish it from the phase #(xj, yk) of S(xj, yk). This can be found from either of the phase derivatives ##/#x or ##/#y or, ideally, from both. A radar arrangement in which this process is implemented is shown in Figure 8. A received pulse P(xj, t) enters n range gates. In each gate it is correlated with a range reference waveform and its derivative to produce outputs S(xjr yk) and #S(xj, yk)/#y for k=1 to n, which are stored.From the store S(xj+1, yk), for l=-m to +m, are correlated with an azimuth reference wave form and its derivative to produce outputs V(xi, yk) and #V(xj, yk)/#x. Similarly #S(xi+1, yk)/#y for I=-m to +m are correlated with the same azimuth reference waveform to give the output DV(xl, Yk)/DY. If V has the in-phase and quadrature components V=l+iQ, then the phase derivatives are

These are calculated and stored with V. These phase derivatives or the amplitude image tVI2 can then be displayed. The phase can be unwrapped using numerical integration orTribolets method (described in general terms in an article by J. M. Tribolet in IEEE Transactions on Accoustics, Speech and Signal processing, Volume ASSP-25 April 1977) and the unwrapped phase image displayed. In Figure 8 the azimuthal operations are shown following the range operations. As they are of the same type they can be done in the reverse order. In practice a 2-dimensional range and azimuth correlation can be done at the same time. The unwrapped values clr(xi, y,) can be displayed in a two-dimensional pattern to give a very satisfactory radar image, which, particularly when sea surfaces are imaged, can be more acceptable than a conventional amplitude display.

Claims

1. A radar system including means for producing from a radar echo signal relating to a radar target lying within a range or bearing cell, a signal representative of the unwrapped phase of the echo signal, and means utilising the signal to produce a radar image of the target.

2. A radar as claimed in Claim 1, and wherein the motion of the radar system is used to constitute a synthetic aperture radar, and a two dimensional data pattern or display is generated having an azimuth axis corresponding to the direction of motion, and a range axis corresponding to a direction transverse to the motion.

3. A radar as claimed in Claim 2 and wherein reference signals which are coherent with the transmission of radar signals are used to correlate returned radar echoes relating to different azimuth and range cells.

4. A radar as claimed in Claim 2 or 3 and wherein the derivative of phase angle is obtained both with respect to range and azimuth, and from which the unwrapped phase is determined for each range/azimuth cell in the two dimensional array.

5. A radar substantially as illustrated in and described with referenca to Figure 8 of the accompanying drawings.