3-5 Development of Airborne high-resolution multi ...€¦ · 2 Principles of Imaging Radar and Synthetic Aperture Radar 2.1 Microwave Imaging Radar Microwave imaging radar transmits

109

1 Introduction

In recent years, the global environmentissues have been drawing greater attentionworldwide. In this context, remote sensingtechnology is expected to contribute to themonitoring of desertification and observationsof volcanic eruptions, earthquakes, and forestfires. Radio waves are useful in such observa-tions in that they are not affected by weatherconditions and in that they allow observationon a 24-hour basis. Synthetic aperture radar(SAR), one type of microwave remote sensor,transmits microwaves from an airplane or asatellite, receives the echoes returned from theground surface. This radar system can attainsignificantly high resolution through the useof advanced signal processing such as aperturesynthesizing and pulse compression. Theimages provided by SAR appear quite similarto monochrome aircraft photograph. More-over, the SAR’s use of microwaves has theadvantage of rendering it capable of acquiring

images at any time of day or night, and evenunder adverse conditions like cloud covers. Inaddition, SAR can identify objects in a man-ner different from that of optical sensors, aslight and microwaves show different behav-iors in scattering and reflection. SAR hasalready been mounted on satellites and used inmilitary applications. When mounted on anairplane, SAR provides diverse observationopportunities where parameters such as reso-lution, observation range, and incident angleare optimized for each observation purpose.Although observation width becomes narrow,an airborne SAR can obtain a high-resolutionimage with 10 times the resolution as thisimage taken from a satellite-borne SAR.

Further, improved hardware will make cut-ting-edge observation functions available forapplication experiments. While a satellite-borne radar system acquires observation dataonce following virtually the same flight courseand direction, airplane-borne SAR can per-form quick observations in arbitrary directions

UMEHARA Toshihiko et al.

3-5 Development of Airborne high-resolutionmulti-parameter imaging radar SAR, Pi-SAR

UMEHARA Toshihiko, URATSUKA Seiho, KOBAYASHI Tatsuharu,

SATAKE Makoto, NADAI Akitsugu, MAENO Hideo, MASUKO Harunobu, and

SHIMADA Masanobu

The airborne X/L-band Synthetic Aperture Radar system (Polarimetric and Interfero-metric SAR, Pi-SAR) was developed by the Communications Research Laboratory andNational Space Development Agency of Japan in their joint project from 1993 to 1996. Theresolution of the X-band image is about 1.5 m and L-band is about 3.0m. Both SARs canmake fully polarimetric observations. The X-band SAR has a cross-track interferometricfunction that measures the ground height with accuracy of 2 m. These systems are installedin the airplane, GulfstreamⅡ. In this paper we describe our Pi-SAR system and the groundprocessing system. In addition, we also discuss the performance of our system by usingthe Pi-SAR datad.

Keywords Synthetic Aperture Radar(SAR), High-resolution, Polarimetry, Interferometry, Dual fre-quency

110

and, in the event of a disaster, for example, itcan provide observations every day or multi-ple times in one day.

Since 1993 the Communications ResearchLaboratory and the National Space Develop-ment Agency of Japan have been jointlydeveloping an airborne three-dimensionalhigh-resolution X- and L-band imaging radar(Pi-SAR: Polarimetric and InterferometricSAR) and carrying out a number of relatedexperiments. Designed not only for experi-mental purposes but also for practical use, thisradar system is installed in a small jet and iscapable of measuring the ground surfaceunder the left side of the plane at a width of upto 40 km, from an altitude of approximately40,000 feet (12,000 m). Pi-SAR is a multi-functional, high-performance radar system,featuring innovative interferometric andpolarimetric observation functions. Thispaper describes the Pi-SAR system, its dataprocessing system and data processing steps,and provides an evaluation of its capabilitiesbased on the obtained observation results.

2 Principles of Imaging Radar andSynthetic Aperture Radar

2.1 Microwave Imaging RadarMicrowave imaging radar transmits radio

pulses downward and slantly in a direction,normal to the advancing direction of an air-craft or satellite, and receives the echoesreturned from the ground as shown in Fig.1.Since the time a pulse takes to return home isproportional to the distance to the ground, thereceived pulses provide a scattering profile ofthe ground perpendicular to the advancingdirection of the radar. As the radar proceedsin the azimuth direction, it provides two-dimensional ground images. The image reso-lution in the range direction is determined bythe pulse width of the radar, while the azimuthresolution is determined by the antenna beamwidth in the azimuth direction. In general, theimaging radar adopts an antenna with a “fanbeam” type pattern that is very sharp in theazimuth direction and relatively wide in the

cross-track direction either to the left or rightside of the advancing direction. If the pulsewidth is narrowed, the resolution in the rangedirection is increased, but the S/N (signal-to-noise) ratio declines as the transmission powerbecomes low. Meanwhile, since the azimuthresolution is determined by the beam width,the beam width spreads in proportion to theflight height and the resolution declinesaccordingly. In order to maintain resolutionlevels in flights at high altitudes, the beamshould be narrowed in the azimuth direction;thus, a large antenna will be required. Indeed,one imaging radar we previously developedfeatured an antenna aperture of about 3.5 m [1],so as to maintain a resolution of about 60 m atan altitude of 3,000 m at an incidence angle of60 degrees.

In contrast to this approach, movement inthe azimuth direction can be used to synthe-size an equivalent array antenna in a techniquecalled “synthetic aperture radar (SAR)” asdescribed next.

2.2 Synthetic Aperture Radar (SAR)When focusing on a single target, for

example, with moving radar having a relative-ly broad beam in the azimuth direction, a tar-get continues to receive the beam for theentire period during which the plane flies fromposition 1 to position 3, as shown in Fig.2. Inother words, this is equivalent to using anantenna with an aperture substantially as largeas L, or using an array antenna whose ele-ments are extended from positions 1 to 3

Journal of the Communications Research Laboratory Vol.49 No.2 2002

Principle of microwave imaging radarFig.1

111

along the flight course to receive the returnedsignals.

By synthesizing the signals received fromsuch a “virtual” array antenna, the azimuthresolution can be narrowed down to half thereal antenna aperture, regardless of flight alti-tude. Namely, if the S/N ratio is not taken intoaccount, we can say that the smaller the anten-na, the higher the resolution. On the otherhand, pulse-compression technology is adopt-ed to raise the range resolution. Specifically,FM chirp compression is usually adopted inSAR. FM chirp compression is a technique bywhich FM-modulated long pulses are trans-mitted and received by a matched filter withinverted frequency-delay characteristics toconvert the original pulses into narrow ones ofa larger power (Fig.3). Looking at this sort ofaperture synthesis in a different way, it may beseen as a signal-processing method that usesan appropriate matched filter to capitalize onthe fact that the Doppler effect (caused by theplane’s movement against the ground) pro-vides the same phase change as that resultingfrom FM modulation. In this sense, theprocess of aperture synthesis is referred to asazimuth compression.

For such range compression and azimuthcompression, we must acquire signal wave-form data, including phase information as wellas signal intensity, using a receiver. As aresult, the radar system becomes quite com-plex, relative to real aperture radar, and therequired recording and processing dataincreases significantly in size. At the sametime, the system features excellent resolutionand a new function enabling the utilization ofphase information. In short, the capabilities ofinterferometric and polarimetric observationscreate a distinctive advantage when perform-ing observations of the ground surface[2][3].(1) Interferometry

When two antennas are installed in thecross-track direction, as shown in Fig.4, thephase difference between signals captured bythe two antennas is determined by the differ-ence in distance from each antenna to the tar-get, although this difference includes instabili-

ty in integral multiples of 2π. Therefore, ifthe phase difference between the signalsreceived by the two antennas is measured withhigh precision, Equation (1) provides the inci-dence angle,θ, at each point on the ground,while Equation (2) provides h[4].

whereλis wavelength, B is the baseline lengthbetween two antennas,αis the angle madebetween the baseline and horizontal line, H isthe altitude of the platform, and R is the rangedistance.(2) Polarimetry

Polarization is the trajectory of the electricfield vector, a trajectory that changes alongwith time at different points in space. Usuallysuch polarization is elliptical, but in somecases it may be circular or linear. If each ofthe transmitter and receiver antennas of theradar has the function of transmission and


Basic concept of chirp compressionFig.3

Basic concept of azimuth compressionFig.2

(1)

(2)

112

reception both horizontally and verticallypolarized orthogonal waves almost simultane-ously, a scattering matrix is provided that hasfour components combining the two (verticaland horizontal) polarization components inboth transmission and reception. When the

scattering matrix is made of complex numberscarrying the phase-relation information, thescattering (in elliptical polarization) from atarget irradiated by polarized waves can becalculated by integrating those polarizationcomponents. In other words, all of the infor-mation related to polarimetry, including thebackscattering coefficient, can be determinedfrom the relations between these components,as each features a complex amplitude,although the scattering matrix features onlyfour-fold information. This is an essential fea-ture of polarimetry.

3 Pi-SAR System

Airborne three-dimensional high-resolu-tion imaging radar (Pi-SAR) is a radar systemdeveloped jointly by the CommunicationsResearch Laboratory and the National SpaceDevelopment Agency of Japan; this systemenables observation of the ground surface


Schematic description of height meas-urement (Inferometry)

Fig.4

System parameter and performanceTable 1

113

using dual frequencies simultaneously－the X-band (with a wavelength of 3.14 cm) and theL-band (with a wavelength of 23.6 cm). Thehorizontal resolution is 1.5 m in the X-bandand 3 m in the L-band, and Pi-SAR is capableof polarimetric observations (by transmitting/receiving horizontally and vertically polarizedwaves) using either band. Pi-SAR also fea-tures an interferometric function to provideelevation profiles of the ground using twoantennas through X-band observation. Table 1shows the major specifications of Pi-SAR[5].As shown in the block diagram in Fig.5, thetransmission signals (100 MHz in the X-bandor 50 MHz in the L-band) are fed to the anten-na in the form of chirp signals and then irradi-ated to the ground. The radio waves reflectedoff of the ground are received by the antennasand subjected to A/D conversion in the receiv-er (with phase information maintained), afterappropriate adjustment of receiver gain andother parameters. This data is saved in a datarecorder along with supplemental data, includ-ing airplane attitude. The data recorded intape media is instantaneously read out by areading-head and then sent to a real-timeprocessor. The real-time processor carries outnecessary processing on the X- and L-bandpolarization data and visualizes the obtainedimage on a screen.

The system for the L-band can bedescribed with almost the same block diagramas that for the X-band, except that it has nosub-antenna and only one data recorder.These systems are deployed together on a two-engine jet, the GulfStreamII. This airplaneallows observation at high altitudes (typically12,000 m) and at speeds of 220 m/s duringobservation[6].

3.1 Antenna SubsystemFor the polarimetric function, the X-band

main-antenna is equipped to deal with bothvertical and horizontal polarized waves. Thesub-antenna, which is required for interferom-etry, employs vertically polarized waves, asthe return power of these waves is relativelylarge. As indicated by Equation (4), the phasechange grows in proportion to the distance(the length of baseline) between the main andsub antennas and the measurement accuracy inthe ground height is improved accordingly.The two antennas should be firmly installed asfar apart as possible, as fluctuations in theantenna position caused by airplane vibrationresult in measurement errors. When thesewere installed as shown in Fig.6, antenna sep-aration was 2.3 m in this Pi-SAR configura-tion. Considering the gain and redome size,we determined the antenna dimensions as 19


Block diagram of X-band radarFig.5

114

cm (6λ) in the vertical direction and 105 cm(30λ) in the horizontal direction. The widthof the pencil-like beam in the azimuth direc-tion was about 2.5 degrees. Furthermore, weformed a square-cosecant type fan-beam elim-inating the change in received power so thatthe received power became almost constant,regardless of distance to the target in the ele-vation direction. To be specific, the receivedpower fell within the -10 dB range overapproximately 40 degrees. The antenna had astructure consisting of eight sub-array ele-ments arranged at intervals of 1/2 across thecenter so as to provide phase coherence on theradiation device during the 100-MHz sweep.The roll angle was variable between 40 and 65degrees (peak to peak) to extend the range ofincidence angle selectivity for all targetobjects and observation modes. The dual-axisvariable angle (within ±6.5 degrees) mecha-nism controlled the yaw angle so that theantennas were kept parallel to the travelingdirection, regardless of the drift angle of theairplane.

Since L-band waves are longer in wave-length than X-band waves, the antenna mustbe larger. We adopted a micro-strip antennathat enabled transmission and reception ofboth horizontal and vertical polarizationwaves with a single antenna. The patch ele-ment was square so as to cut off the cross-polar component of the linear orthogonal ver-tical/horizontal polarization waves. Themounting unit for the antenna elements and

feeder on the panel had a multi-layered struc-ture adopting a dielectric honeycomb materialto provide a lightweight, highly rigid antennasub-panel. The antenna tilt angle in the eleva-tion direction was 38 degrees so as to maintainpower within the 10-dB range over about 40(22 to 58) degrees. Fig.6 shows the externalview of the antennas installed underside of theairplane.

The redome had a sandwich structuremade of glass fabrics and epoxy resin, whichmaterials are transparent to radio waves. Thesignal attenuation caused by the redome wasaround 0.1 dB. The redome for the X-bandsystem was 45 cm in diameter, where theminor-axis aperture diameters of the verticaland horizontal antennas were 20 cm and theyaw angle was variable. The L-band systemwas mounted directly on the body of the air-plane in order to reduce air resistance.

3.2 Transmitter SubsystemAn oscillator having a basic frequency of

12.34 MHz provides eight frequencies to beused inside the transmitter. A reference signalis created with a highly stabilized quartz oscil-lator using a quartz crystal under controlledtemperatures for optimally stabilized opera-tion; indeed, a frequency stability of 10-10

rms/sec is ensured for a short term. A chirpsignal is created in the chirp generator byreading out (with the reference trigger) thewaveform data saved in EPROM (12 bit digi-tal data). Then, this is added to intermediatefrequency and transferred to the radio frequen-cy (either 9.55GHz or 1.27GHz).

The high-output amplifier for the X-bandis made of a traveling wave tube (TWT) so asto maintain stability and reduce frequency-dependent fluctuations in the output level.This amplifier can provide an output power ofabout 6 kW at the antenna end. Meanwhile, atransistor-type high-power amplifier providesan output of 3 kW for the L-band. The ampli-fied signal is supplied to either vertical or hor-izontal polarization antenna. Part of theamplified signal, branched with a coupler, isused to monitor transmission power and to


An airplane instal led SARFig.6

115

correct fluctuation.In addition to independent observation, it

is possible to use the L-band to determinewhich transmission and reception timings aresynchronized with the reference frequencyprovided by the X-band system. Low-resolu-tion (10 m and 20 m) modes are also availablefor simulation of the synthetic aperture radarsystem (ALOS/PALSAR) that will be mount-ed on a satellite scheduled to be launched in2004.

3.3 Receiver SubsystemThe receiver system has two channels (1

and 2) of the same structure so as to receivevertical waves and horizontal waves at thesame time and to enable it to handle twopolarized waves separately, to perform thepolarimetric function both in the X and Lbands. In short, Channel 1 handles the verti-cal polarization signal, while Channel 2 han-dles the horizontal polarization signal. How-ever, in order to perform interferometry andpolarimetry at the same time in X-band obser-vation, Channel 1 features a mode in which itswitches the horizontally polarized wave ofthe main antenna and the vertically polarizedwave of the sub-antenna alternately for eachpulse, as well as another mode in which ithandles only the sub-antenna signals for inter-ferometry alone. When this mode adopted byChannel 1, Channel 2 handles only the verti-cally polarized wave of the primary antenna.

Received signals in the two channels passa limiter and a bandpass filter and then enterthe first stage low-noise amplifier－a FETamplifier capable of reducing the noise levelto 3 dB or lower. After a frequency conver-sion in the gain control unit, the signals passsensitive time control (STC) and auto gaincontrol (AGC) units. Finally, the signals aredivided into two components, (I,Q), havingphase angles 90 degrees apart and detected assuch so that complex signals having phaseinformation can be obtained. Each of thesesignal components is passed through an A/Dconverter to become digital data.

3.4 Control and Signal Processing UnitThis system aims to identify the target

objects based on phase information. Thus allthe signal processing units are required towork in synchronization. For this purpose, allthe hardware is controlled by the referencesignal provided by the frequency generator.Further, X-band and L-band observations aresynchronized by supplying the X-side 1-MHzand 123-MHz clocks to the L-side.

Airplanes are subject to attitude fluctua-tions of shorter intervals than those of satel-lites. The synthetic aperture radar mounted onan aircraft, therefore, requires as much real-time aircraft attitude data as possible. For thispurpose, our system is equipped with high-precision sensor systems such as an inertialnavigation system (INS) and a GPS to provideinformation about position and attitude. Thisinformation is collected by the signal-process-ing unit and saved in a data recorder alongwith I and Q data obtained by the receiver.

One of the benefits of the synthetic aper-ture radar lies in its generation of large quanti-ties of data. Approximately three minutes of sin a flight of experiment produces about 12GB of X-band data and 6 GB of L-band data.For data storage, we have adopted two high-speed data recorders (DIR-1000) for X-banddata storage and one DIR-1000 for L-banddata storage. The data transmission rate ofthis type recorder is 256 Mbps per unit. Eachcassette has a large data capacity of 90 GB percassette; however, this is used within about 50minutes of continuous observation.

3.5 Real Time ProcessorThe airborne SAR is distinguished by its

rapid availability in the event of a disaster, andcan be used to gather information on a real-time basis during a disaster. To meet therequirements related to such applications, areal time processor was installed in the SARin 2000. This device can reproduce andprocess arbitrary X- and L-band polarimetricdata acquired in observation to provide an ini-tial overview. The number of data acquisitionpoints, however, is fixed at 1,024 in the range


116

direction, due to limitations of data size andprocessing rate, and its bandwidth changeswith resolution (max. 2.5 m). Table 2 lists theparameter settings, selectable via a switch.When reading out the data written on the tapein the recorder and transferring the readoutdata to the real time processor, it is possible tocheck to determine whether or not the writtendata is correct. The processed image data isdisplayed on-screen, and image data for eachpage can be saved to a magneto-optical (MO)disk.

4 Observation Plan and Opera-tion

4.1 Observation PlanThe airborne SAR measures the area under

the plane on its left side in a slanted directionwith a fixed width. We must determine theflight course in advance to most efficientlycover the target areas. For this purpose, thelatitude and longitude information for the tar-get areas must be known prior to observation.In addition, information about incidenceangles is required, as the backscattering cross-section of a target depends on its incidenceangle. Particularly in X-band observation, theantenna roll angle is variable and is set at avalue that will maximize the antenna gain overthe target observation area. For some targets,the observation direction (direction of emittedradio waves) is another important parameter.

Further, in mountainous areas, the flight alti-tude of the aircraft becomes substantiallylower due to the effect of ground height. Insuch cases the slant range shifts from the cor-rect value and the target may go out of theobservation range. We must take these factorsinto account when determining observationparameters. In addition, when measuringsteep mountainous areas, we must determinethe incidence angle so as not to cause fore-shortening or produce shadows on slopes. Inthis manner, a flight plan is designed to coverabout ten observation paths per day, to mostefficiently execute the required observation.The aircraft on which the airborne SAR isinstalled is capable of continuous flight at analtitude of 12,000 m at about 220 m/s for aboutfour hours. This translates into a distance ofmore than 3,000 km, which would appearlargely sufficient in terms of observation. Infact, the net observation time is approximatelyhalf of the above-mentioned maximum flighttime, as the observations require a certain leadtime for stabilization of airplane attitude(including drift angle) before entering therespective flight path, and the aircraft needstime to ascend, descend, move between flightpaths, and turn around.

4.2 OperationWhen a proper observation plan is made,

the starting observation position, the observa-tion mode, the observation distance, and theparameter settings can be saved in a parameterfile before the flight of experiment. Whenstarting actual observations, we read out thisparameter file to establish the observationmode, saving time in preparations and pre-venting erroneous parameter settings. On theday of observation, we first check the weatherconditions and the flight course and activatethe INS and GPS before takeoff. Nitrogen isfilled in the X-band wave guide under pres-sure to prevent discharge prior to takeoff.When approaching the observation site, theoperator begins observation following thesteps listed on Table 3, based mainly on theinformation provided by the pilot. When the


Selective parameters for real timeprocessing

Table 2

117

aircraft has arrived at the point at whichstraight flight will begin, the pilot stabilizesthe drift angle so as to maintain the flightcourse and notifies the operator of the timingfor commencement of observations. Theoperator for X-band observation then makestemporary settings for antenna roll angle andyaw angle to correct for drift angle. The oper-ator is then notified of the start timing andmakes the appropriate settings for the observa-tion parameters. We set aside about 10 sec-onds for averaging of the aircraft altitude anddrift angle. When the parameter setting havebeen completed, the SAR system controls therecorder and, when the tape is runningsmoothly, begins observations after three sec-onds of noise observation and 15 seconds ofreceiver calibration. Table 3 shows the rele-vant timetable (indicating only the essentialitems for such settings). Operators areabsorbed in final verification of observationparameters and hardware conditions duringobservations.

5 Data Processing Subsystem

The higher the horizontal resolutionbecomes, the greater the amount of dataacquired in SAR observation. Accordingly,data readout and data processing will takelonger. Fortunately, the data processing speedof processors and the capacity of storagemedia have improved remarkably in recentyears. We have continually upgraded the SARdata processing system in step with suchimprovements. Described below is the systemstatus as of the end of 2001.

5.1 HardwareFig.7 presents configuration of the SAR

processing system.In this figure, SAR data processing com-

puter and SAR application processing comput-er are separated only for convenience; they donot each feature customized programs. Thuseither computer may be used as backup for theother, and their connections are configured toenable both primary and backup functions.Raw data is read out to a D1 tape device con-nected to the primary calculator via SCSIinterface and is subject to Level-0 processingto arrive at a predetermined data format.Since approximately five minutes of observa-tion in the X-band provides 19 GB of data (atmaximum), a 180-GB HDD can only serve astemporary storage. Thus we save data that hasbeen subject to Level-0 processing to a 180-GB DTF tape. This basic data-processing sys-tem is duplicated; thus, parallel processing ispossible.

After producing the Level-0 data, wecheck the hardware and conduct browse pro-cessing (10 m per pixel) for catalog prepara-tion purposes. This processing is carried outon data acquired from all flight paths, andobtained data will be added to the cataloginformation. Later, SAR images areprocessed. The image size is 5 km×5 km, andapproximately 1.5 hours is required for dataprocessing of one scene of single polarimetricdata. Data that has gone through basic pro-


Operation flow (per observationpath)

Table 3

118

cessing such as single-look slant-range com-plex (SSC) data and multi-look ground-rangeamplitude (MGA) data is saved to a hard diskin the file server and sent through the network,

at which point it is subject to higher-level dataprocessing, as necessary.

5.2 Software for data processingSAR software can be divided into three

categories, depending on function, as follows:(1) Level-0 data production;(2) Image production; and(3) Distributed data production

The contents of each process are describedbelow.(1) Level-0 data production

Raw data recorded to the data tape duringthe experiment flight is read out for the pro-duction of Level-0 data. More specifically,the main process consists of the separation ofreceived signals based on the types of polar-ized waves and observation modes involved(i.e., calibration, noise data, or actual observa-tions). At the same time, a file is made offlight information for each data frame; thesefiles will be required when performing motioncompensation and when referring to radar set-tings during observation.(2) Image production

The image production process is dividedinto the following four steps, which featuredifferent data formats for output. Appendix-1lists the formats for individual products, inaddition to other relevant items.(a) Single-look slant-range complex imageproduction

In this step, as the central element ofimage reproduction, range compression andazimuth compression are performed, with cor-rection of radar parameters and motion com-pensation of the aircraft based on the receivedsignals. The received signals are visualized inthis process. This step will be explained inmore detail later.(b) Multi-look ground-range amplitude imageproduction

The power of SSC data is calculated andsubjected to multi-look processing (4-look forPi-SAR) to perform conversion from slant-range data to ground-range data.(c) Multi-look ground-range polarimetricimage production


Confinguration of SAR processing systemFig.7

119

SSC data is converted into the Mullermatrix and subject to multi-look processing.After calculation of the components in thecomplex scattering matrix, slant-range toground-range conversion is performed.(d) Interferometric processing

In this calculation, phase difference isderived from the phase difference obtainedwith the main and sub antennas by removingphase difference from a plane. Further, coher-ence between data sets is calculated to providemulti-look slant-range interferometric imagesand multi-look slant-range coherence imagesas output.

In addition, the “browse image produc-tion” process is performed to check observa-tion conditions, confirm the observation range,and prepare catalog data. The size of browseimage is 20 km long in the range direction and50 km long in the azimuth direction with onepixel size being 10 m.(3) Preparation of materials for users

In principle, the obtained data is com-pressed and recorded in CD-ROM and othermedia formats for easy distribution in thisstep. We have adopted JPL’s SIR-C “SLC”quad-pol data method for compression. Thismethod enables the compression of an original2,048 MB of data [for 4 ch-SSC images (8bytes×4) for each 5 km×5 km area] to 10bytes per pixel, to fit on a 640 MB CD.

5.3 SSC Image ProductionThis section explains single-look slant-

range complex image production, which rep-resents the core of the actual reproductionprocess.(1) Range compression

I and Q data saved in the recorder consistsof power data sets containing information onphases that are separated by 90 degrees. Thisdata is converted into complex numbers. Wecorrect the DC bias (zero level adjustment)and gain for both the I and Q data sets in orderto average fluctuations in the characteristics ofA/D converters. Gain/phase changes in thereceiver amplifiers and attenuators are alsocorrected. The data is then subject to the

range compression described in Section 2.2,in a matched-filter process. In fact, this repre-sents a convolution calculation, using therange reference function in the given frequen-cy region. With Sref (fτ) representing theresults of Fourier transform of the range refer-ence function, Srev (fτ) representing the Fouriertransform of received waves, and W (fτ) repre-senting the window function in the current fre-quency region,Src (fτ) = W (fτ) Srcv (fτ) S* ref (fτ) where * represents the complex conjugatenumber.The range compression signal is obtained byinverse Fourier transform of Src (fτ).(2) Radiometric correction

In addition to corrections related to thereceiver, the range-compressed signal under-goes transmitter level correction, antenna ele-vation correction, and attenuation correction(dependent on the propagation distance of theradio waves).(3) Aircraft motion compensation

Motions in height and attitude of the causeshifting and phase rotation in slant-range pix-els. Thus the influence of these motionsshould be removed prior to azimuth compres-sion. If the motion-induced shifts in positionof the transmission antenna [y: cross-trackdirection (+: left), z: vertical direction (+:upward)] relative to the reference course ofthe aircraft are expressed byΔytrans (y), Δztrans

(y), and if similarly, those of the receptionantenna are expressed byΔyrcv (y), Δzrcv (y),the look angle isθ(τ), and the wavelength ofthe adopted microwave isλ, then the slant-range shiftΔr (t,τ), and phase rotationΔφ(t,τ) are expressed by:

Among these correction quantities, thepositional information (derived by integratingthe INS data that was recorded with thereceived signals) is embedded in the inde-pendent GPS positioning data. The attitudedata is calculated by integrating the angularacceleration detected by a gyro installed in the


120

INS. Transmission pulses must be resampledwith reference to this positional information.Thus, based on this aircraft attitude data,resampling is carried out in order to maintainphase information using a finite impulseresponse (FIR) filter with linear phase charac-teristics. The filter length and sampling inter-val are variable.(4) Estimate of center Doppler frequency

The center Doppler frequency is easilyestimated, provided that the information aboutaircraft attitude and speed are preciselyknown. However, in most cases, it is difficultto measure the aircraft attitude precisely.Therefore, the Pi-SAR processing system isusing the positional information provided bythe above-mentioned aircraft motion compen-sation, instead of estimating the centerDoppler frequency.

Nevertheless, it is also possible to conductcorrection using the estimated center Dopplerfrequency, as follows. First, the azimuth-linesignal previously subject to range compressionis converted into the Doppler frequency regionto allow estimation of its power. We examinethe correlation between the estimated powerand round-trip azimuth antenna pattern corre-sponding to the Doppler frequency, and deter-mine the frequency that shows the peak ofsuch correlation. This frequency, however,will feature uncertainty equivalent to an inte-gral multiple of the pulse-repetition frequency.This instability is corrected based on the cen-ter Doppler frequency provided by the aircraftattitude correction. This value is estimated inseveral different range positions and is calcu-lated by least-square interpolation.(5) Correction of range migration

Considering the airplane’s trajectory rela-tive to a point on the ground, we can expressthe slant-range distance by a quadratic equa-tion that provides the shortest distance to thetarget, i.e., when it is located to the immediateleft of the plane. As a preliminary processprior to azimuth compression, we have to cor-rect the range distance (or arrange the dis-tances in a different sequence) within theadopted bandwidth. This correction process is

called “migration correction.” We convert thetime-axis data into frequency-axis data andcalculate the movement at the same Dopplerfrequency so that all of the target informationis arrayed in the azimuth line.(6) Azimuth compression

This is a data processing operation bywhich the reference azimuth wave is createdfrom the center Doppler frequency and aver-age speed of the aircraft followed by correla-tion with signals that have experienced rangemigration. This process is also a correlationcalculation in the adopted frequency region, asin the case of range compression. If Sref-az (f,τ)is the Fourier transform of the referenceazimuth wave, S'rc (f,τ) is that of the signalsubject to range migration and W (f) is thewindow function, then azimuth compressionsignal can be provided by inverse Fouriertransform of a correlation calculation Sac (f,τ)= W (f) S'rc (f,τ) S*ref-az (f,τ) where * repre-sents the complex conjugate number.

6 Evaluation of the Pi-SAR System

(1) ResolutionIn general, the resolution of a synthetic

aperture radar is expressed byΔr = c/2B in theslant-range direction, as mentioned earlier,where B is transmitter bandwidth. Theazimuth resolution (single-look) is given byΔAz = D/2, where D is the real antenna aperturelength (usually around 1.2 times the realantenna length).

The process used in general analysis isalso used to perform multi-look so that theazimuth-direction resolution may become ashigh as the range-direction resolution andthereby reduce speckle noise.

In X-band resolution, when the transmitterbandwidth is 100 MHz and the antenna lengthis 1.05 m, thenΔr = 1.5 m andΔAz = 0.63 m,while in L-band resolution, when the transmit-ter bandwidth is 50 MHz and the antennalength is 1.55 m,Δr = 3.0 m andΔAz = 0.93 mare expected. When evaluating the resolutionof SAR, we need a point target in a locationwhere the background scattering level is low.


121

For this purpose, CRL conducted a calibrationexperiment by deploying corner reflectors atTottori dune. Fig.8 shows the obtained cali-bration data. This figure indicates that therange resolution is approximately 1.5 m andthat azimuth resolution is 0.83 m in X-bandresolution, while the range resolution isapproximately 2.9 m and azimuth resolution is

1.2 m in L-band resolution; these results are innear agreement with the design specifications.(2) Minimum receiving sensitivity

The SAR system conducts internal calibra-tion and noise data before and after eachobservation. The minimum receiving sensitiv-ity can be estimated from this data. The leveldiagram indicates that the noise level, which ismeasured based on the power at the output ter-minal of the receiver, lies between -77 dB and-87 dB. Fig.9 presents sample calibrationdata. This indicates that the input level isalmost linear in the region from approximately-88 dB. Thus the minimum receiving sensitiv-ity is nearly equivalent to the noise level (-87

dB), which is well beyond the specified levelof -80dB or less.(3) Ground projection accuracy

In Pi-SAR data processing for data projec-tion in the ground-range direction, projectionon the ground surface is carried out based onthe slant-range information about flight heightand incidence angle at each observation site,assuming that the ground elevation is uniformin the area under observation. In order toexamine the pixel position in the ground-rangedirection, we deployed several corner reflec-tors over about 10 km in the range direction inOhgata village in Akita prefecture. We thencompared the determined positions of thesereflectors with the position data obtained byGPS. We found that the ground-projectionaccuracy declined in proportion to the magni-tude of errors in flight height. When groundprojection was carried out using the GPS sys-tem, the altitude error resulted in poor projec-tion accuracy. However, Pi-SAR can estimatethe altitude of the aircraft provided by interfer-ometric processing. We found that groundprojection based on this estimated heightinformation would provide positioning accura-cy as high as approximately one pixel.

In other words, when the ground heightchanges within the produced image, a groundprojection error occurs from elevation change,unless the ground height data for each pixelposition is taken [7].(4) Interferometry

The error of interferometry is estimated byequation (1), in which the altitude error is 3.2m where a phase-measurement error is 5degrees at an altitude of 12,000 m and at anincidence angle of 45 degrees. In order toevaluate the accuracy of altitude data in realmeasurements, we conducted interferometryprocessing on the observation data for Ohgatavillage, Akita prefecture, and calculated alti-tude for purposes of comparison. Ohgata vil-lage, which was built on reclaimed land, fea-tured almost no variations in ground elevation.In addition, since the unit size of each paddywas very large (90 m×150 m), much largerthan Pi-SAR’s minimum area of resolution,


Response from CR (X-band)Fig.8

Input/output characteristics of thereceiver

Fig.9

122

we were able to gather a sufficient quantity ofdata samples from a single paddy. Then wecalculated the average and standard deviationof the data for several measurement points

featuring different incidence angles. The cal-culation results are listed in Table 4.

Note that at the time of interferometry pro-cessing 2×2 averaging was performed onboth the azimuth and range directions, andpixel spacing was 2.5 m. Approximately 600pixels were used in this statistical treatment.The obtained results show that shallow inci-dence angles result in higher sensitivity tophase change and thus produce less error, andthat some averaging improves the groundheight accuracy. The required accuracy can beobtained by averaging over (5 m)2 - (10m)2.(5) Polarimetry

Polarization is expressed by a scatteringmatrix composed of four complex numbers, asdescribed earlier. It is necessary to measure ascorrectly as possible the absolute value ofthese elements, and the relation between ele-ments. Thus external calibration, usingknown targets, becomes necessary. In general,we are required to estimate inter-channelimbalance (with respect to phase and power)as well as to calculate the absolute values ofthe respective channels using corner reflectors(CR), and active radar calibrators (ARC). Inthe case of Pi-SAR, we had conducted calibra-tions using CR and ARC installed in Tottoridune and on ice in Saroma Lake[8]. Gainimbalances in the X- and L-band systems hasbeen made public in the form of calibrationcoefficients (see the handout CD of Pi-SAR

data and the CRL Web page). The level ofcross-talk in both the X- and L-band ones is solow that it may be ignored. Since a patchantenna is adopted in L-band observation, theantenna position does not change between dif-ferent polarization waves. On the other hand,in X-band observation, antennas in differentpositions are used for each polarization waveand the roll angle is variable. Thus a correc-tion related to antenna position becomes nec-essary to gain accurate phase information[9].Furthermore, for precise polarimetric calibra-tion, it would be helpful to conduct a calibra-tion experiment in which we installed a num-ber of such reference targets at different inci-dence angles in the imaging area of interest.Such an experiment, however, is not practicalto perform. Van Zyl et al. conducted calibra-tions[10] for JPL’s AIRSAR, using natural tar-gets as well as artificial ones. However, it isdifficult to find a natural scattering target inJapan that is uniform over a wide range; thuswe have been developing an alternativemethod of calibration.

7 Summary

The observation result obtained until nowshows that the development target was almostattained about the system by which the air-borne synthetic aperture radar together devel-oped by Communications Research Laborato-ry and National Space Development Agencyof Japan includes a data-processing function.Calibration coefficients for the radar have alsobeen clarified through experiments employingcorner reflectors deployed at Tottori dune. Aswe have added a real time processor unit tothe basic hardware system, it can now beadopted quickly and effectively－in the eventof emergency, for example. The system hasbeen upgraded, whenever necessary, toaccommodate massive data processing tasks,and modifications of the software have thusreached a satisfactory stage.

The future challenges regarding hardwareinclude an exploration of the possibilities ofalong-track interferometry and real-time


Mean height and S.D. by inci-dence angles in Ohgata village

Table 4

123

downlink of data. In order to use the obtainedPi-SAR and other observation data moreeffectively, we anticipate moving forward incollaboration with research institutions thatare extensively involved in related applica-tions. In addition, we have distributed CD-ROMs to potential users that contain Pi-SAR

data for various areas such as forests, farms,and oceans. We expect that more and moreresearchers will become interested in the Pi-SAR program and that synthetic aperture radarwill be subject to further and more extensiveinvestigation in the future.


124 Journal of the Communications Research Laboratory Vol.49 No.2 2002

Appendix-1. Formats of the products(1) Data products

Single-look Slant-range Complex: SSCProcessingRange compressionSingle-look azimuth compression

4-byte + 4-byte complex

1: Theoretical value without any window function2: Nominal value (depends on airplane speed and PRF)

Multi-look Ground-range Amplitude: MGAProcessingMulti-look processing of SSC dataSlant-range to ground-range conversionCalculation to Amplitude

2-byte integer

Multi-look Ground-range Polarimetric: MGPProcessingCalculate complex scattering matrix from SSC dataTransform to Muller matrix, multi-look processing and transform to scattering matrix Slant-range to ground-range conversion

4-byte + 4-byte complex

SSC

X-SAR

L-SAR

Bandwidth

100MHz50MHz50MHz

Resolution [m] Pixel [m]Az 1

0.3750.3750.750

Sr 1

1.53.03.0

Az 2

0.2610.2610.522

Sr1.2142.4282.428

MGA/MGP

X-SAR

L-SAR

Bandwidth

100MHz50MHz50MHz

Resolution [m]Gr 1 2

1.6-17.23.2-34.43.5-8.8

Az 1

1.53.03.0

Az1.252.502.50

Gr1.252.502.50

Pixel [m]Look number

4look8look4look

1: Theoretical value without any window function2: Incidence angle of 20 to 60 degrees

Browse image1-byte integerImage swath 50 km (azimuth)×20 km (range)Resolution of about 10 m (range)×10 m (azimuth)

(2) Data-file structureThe decompressed file has no header or the like. Instead, data sets are repeated as many times

as there are azimuth-direction data points along a single azimuth line.Coordinates in the image

125UMEHARA Toshihiko et al.

Nx: Data-point number in the azimuth directionNy: Data-point number in the range direction

(a) Data arrangement in each file of decompressed MGP data

Decompressed data is of the following size:

(a) nnnn* (8-byte complex number data*mmmm)*4 polarized waves(b) nnnn* (4-byte real number data*mmmm)

If nnnn = 4,000 and mmmm = 3,000, then the sizes are given by:(a)4,000* (8*3,000)*4 = 384 MB, (b)4,000*4*3,000 = 48 MB.

(3) Reference informationThe distributed data includes a reference information file that has "_a" at the end of the file name.This includes observation information about altitude and airplane speed, latitude/longitude of theprocessing site, incidence angle, pixel number, pixel size, and window function processing. You canrefer to this information file together with the JPEG data in this document. It may prove helpful forreference when analyzing reproduced images.

(b) Data arrangement in each file of decompressed power (VV-mode) data

Azimuth direction( 1 , 1 )

:( 1 , Ny )

.........

.........

( Nx , 1 ):

( Nx , Ny )Range direction

Re ( 1 , 1 ):

Re ( 1 , Ny )4-byte real

Im ( 1 , 1 ):

Im ( 1 , Ny )4-byte real

.........

.........

Amp ( 1 , 1 ):

Amp ( 1 , Ny )4-byte real

.........

.........

Amp ( Nx , 1 ):

Amp ( Nx , Ny )

Re ( Nx , 1 ):

Re ( Nx , Ny )

Im ( Nx , 1 ):

Im ( Nx , Ny )

References1 T.Kozu, et al., "Researh on Compact and High-performance Airborne Imaging Radar for Wide-area

Monitoring of Oil Pollution ", Environment Preservation Reports, pp.84-1 to 84-12,1989.

2 F.T.Ulaby, R.K.Moore and A.K.Fung, "Microwave Remote Sensing active and passive ", Artech

House,Inc. MA,Vol.2,1982.

3 Curlander,J.C. and R. MacDonough, "Synthetic Aperture Radar: Systems and Signal Processing ",

John Willy & Sons, New York,1991.

4 Allen, C.T., "Interferometric Synthetic Aperture Radar ", IEEE Newsletter, Sep. 1995.

5 Development of Three-dimensional Airborne SAR at CRL, Tech.Rep.IEICE, SEAN95-99,33-38,1995.

6 T. Kobayashi, et al., "Airborne Dual-Frequency Polarimetric and Interferometric SAR", IEICE Trans.

Commun., Vol.E83-B, No.9, pp.1945-1954, 2000.

7 T.Umehara, et al. , "Accuracy validation of ground projection Pi-SAR images ", RSSJ, No31pp.267-

268,2001.

126 Journal of the Communications Research Laboratory Vol.49 No.2 2002

8 M. Satake, et al., "Calibration of an X-band Airborne Synthetic Aperture Radar with Active Radar

Calibrators and Corner Reflectors ", CEOS 1999 SAR Workshop, Toulouse, France, 1999.

9 M. Satake, et al., "Development and Experiment of Polarization Selective Corner Reflectors for

Polarimatric SAR Calibration ", 2001 CEOS SAR Workshop, Tokyo, Japan, 2001.

10 J.J.van Zyl, "Calibration of polarimetric radar images using only image parameters and trihedral cor-

ner reflectors ", IEEE Tans. Geoscience Remote Sensing, Vol.28, No3, pp.337-348,1990.

UMEHARA Toshihiko

Senior Researcher, Environment Infor-mation Technology Group, AppliedResearch and Standards Division

Global Environment Remote Sensing

URATSUKA Seiho, Dr. Eng.

Leader, Environment InformationTechnology Group, Applied Researchand Standards Division

Microwave Remote Sensing

KOBAYASHI Tatsuharu, Dr. Sci.


Remote Sensing by Microwave

SATAKE Makoto



MASUKO Harunobu, Dr. Sci.

Executive Director, Applied Researchand Standards Division


SHIMADA Masanobu, Ph. D.

Senior Engineer, Earth ObservationResearch Center, National SpaceDevelopment Agency of Japan

Global Environment Remote Sensing

NADAI Akitsugu


Physical Oceanography, Ocean RemoteSensing

MAENO Hideo


Pi-SAR, Remote Sensing

Documents

3-5 Development of Airborne high-resolution multi ...€¦ · 2 Principles of Imaging Radar and Synthetic Aperture Radar 2.1 Microwave Imaging Radar Microwave imaging radar transmits