SCANSAR AND SPOTLIGHT IMAGING OPERATION STUDY FOR SAR SATELLITE MISSION

Transcription

1 SCANSAR AND SPOTLIGHT IMAGING OPERATION STUDY FOR SAR SATELLITE MISSION Bor-Han Wu, Meng-Che Wu and Ming-Hwang Shie National Space Organization, National Applied Research Laboratory, Taiwan *Corresponding author: ABSTRACT Synthetic Aperture Radar (SAR) has been widely used for remotely sensed on Earth for decades. The advantages of SAR are to provide all weather and all day capability imagery and can be complemented with optical images. In considerations of the applications of SAR, user needs, mission definitions and critical technology preparations, a development of the X-band SAR program was planned and enforced from 2015 to The goal of the program is to develop a prototype SAR payload, including electronic units, high power amplifier (HPA) modules and high-gain multibeam antenna. This prototype shall present the technology readiness for NSPO (National Space Organization) future SAR small/microsatellite mission. So far, the preliminary assessment based on the designed subsystems and components shows that signal, power amplifiers, power supply, antenna gain pattern, data sampling, compression, recording and downlink system can meet the requirements of Stripmap imaging mode. However, ScanSAR and Spotlight modes are also necessary in order to achieve the goals for the wider swath and high resolution respectively for broader applications. In this study, the imaging operations and corresponding design parameters of the ScanSAR and Spotlight modes are presented. Additional system requirements respect to Stripmap operation are also analyzed to current SAR payload design and specifications. KEYWORDS:SAR imaging mode, ScanSAR, Spotlight, System requirements. 1. INTRODUCTION The original radar system developed during World War II was designed to measure range (to a target via echo time delay) and direction (of a target via antenna directivity); while, Doppler shift were used to measure the speed. Carl Wiley proposed in 1951 that Doppler shifts could be processed to obtain fine resolution in a direction perpendicular to range. The 2D images of ground surface can be made by creating the effect of a very long antenna in signal processing stage. This method was termed Synthetic Aperture Radar (SAR). Up to now, SAR images have been applied to various research topics such as mapping, geology, forestry, agriculture, oceanography and the like. Conventional SAR imaging systems are having three fundamental imaging modes: stripmap, scansar and (staring) spotlight. By combining these three operational concepts, A1

2 various hybrid modes are developed, e.g., the mosaic, sliding spotlight, TOPS,...etc. NSPO SAR satellite mission shall, at least, be capable of acquiring data by stripmap, scansar and spotlight operations. Symbols, SAR system parameters or its expressions are listed in Table 1. Table 1. List of symbols Content Spec. λ Wavelength m h Orbital altitude ~500 km V sat Satellite speed km/s ϕ Incident angle R (slant) Range to target km B Chirped pulse bandwidth 300MHz B r Receiver bandwidth or LPF 10/30/100/300 bandwidth MHz θ az Azimuth antenna beamwidth unavailable θ el Elevation antenna beamwidth unavailable --- Overlapping of θ el unavailable G Antenna peak gain unavailable F Noise figure of receiver ~5 db T 0 Noise equivalent temperature 290 K L t System total losses except F 7 db τ p Pulse width 40~50 µs PRF Pulse repetition frequency 4000 Hz τ p PRF Pulse duty cycle 18% P t Transmitted power 2.4 KW N beam Number of beams used 2~5 N look Number of looks in scansar 2 L SA Synthetic aperture length Rθ az or R θ T A Exposure time L SA /V sat B D Doppler bandwidth 2V sat θ az /λ T G Gap duration unavailable T B Burst duration unavailable θ Synthetic aperture angle Table 5 δ gr Ground range resolution c/(2bsinϕ) δ az Azimuth resolution (1), (3), (5); Table 2, 4, 6 S Swath hθ el /cos 2 ϕ; Table 2, 4, Scene length in stripmap ~ V sat (2min.) --- Scene length in spotlight (4) σ 0 Normalized RCS, also named as Backscattering coefficient σ/(δ gr δ az ) NESZ Noise Equivalent Sigma (2); Zero, an index of system Table 2, 4, 6 radiometric sensitivity 2. OVERVIEW OF STRIPMAP MODE Range compression via a FFT (Fast Fourier Transform) operation of the received data together with transmitted chirp signal s matched filter is a common technique in radar system. After range FFT (and Inverse FFT), the signal can be shown as in Fig.1 by two different forms. Figure 1. Left is range compressed data arrangement on 2D spatial space, where (x, y) denotes (azimuth, range); Right shows Doppler Histories on time-frequency domain where x = V sat t. (Note that V sat constant, so the azimuth space/time can be interchanged in illustrations.) Left in Fig.1 shows every target corresponds to a set of data lying on one curve. The curved arrangement of data shows the range cell migration during exposure time. Echoes at the locus of cells are used to synthesize and form a single sample located at (x i, y i ). Right part shows Doppler history of each target behaves as linear frequency modulation (LFM); the concept of the subsequent azimuth compression is the same as range compression, and can be applied. 2.1 Continuous mode Continuous mode is the conventional stripmap mode. It contrasts to burst mode which is the basis of scansar mode. In continuous mode operation, all echoes in a synthetic aperture A2

3 length (as in left part of Fig.1) are collected and then be processed to obtain the best azimuth focusing result (enhanced SNR and fined resolution). Image resolution, swath and the radiometric sensitivity to ground targets (NESZ) are common indices of performance of a SAR system. The pulse bandwidth determines ground range resolution in a range of the incident angles. Azimuth resolution in broadside (zero squint) case is expressed by δδ aaaa = λ RR 2LL SSSS, (1a) = VV ssssss BB DD. (1b) By setting θ az λ/l, a limiting value of half antenna length can be derived. Swath depends on elevation beamwidth of antenna and incident angle on curved Earth surface. A good estimation is based on the geometry of SAR antenna center, Earth center and target on Earth surface. In the case of a small θ el, it can be simplified as in Table 1. NESZ = σ 0 /SNR image. While substituting B r B, θ az λ/l and δ az L/2 into (2), a conventional form of SAR equation is present. Table 2 shows estimated stripmap capabilities of NSPO SAR, wherein incident angles range from 30 to 45 for planned normal operations but can also be up to 50 for special requests. Table 2. Capabilities of stripmap imaging with a resolution of δ gr = δ az = 3m. Incident angle Bandwidth (MHz) Swath (km) NESZ (db) Burst mode Contrast to continuous mode, the burst mode processes only parts of echoes during the azimuth compression. As shown in the bottom panel in Fig.2, only the signals in dashed area are collected and/or processed. NESZ is a factor of SAR equation. Based on radar equation, we introduce SNR in = P r /kt 0 B r and receiver noise figure F = SNR in /SNR out to obtain SNR at receiver output. By multiplying range compression ratio (N r = τ p B) and azimuth compression ratio (N a = PRF T A ) to SNR out, an estimation of averaged SNR on a focused SAR image is derived. Hence, NESZ = (4ππ)3 RR 3 FF kktt 0 VV ssssss 1 BB rr 1, (2) PP tt ττ pp PPPPPP λ 2 GG 2 LL tt θθ aaaa BB δδ gggg δδ aaaa where NESZ is specified as backscattering coefficient σ 0 which makes SNR = 1, i.e., Figure 2. Range compressed signal of continuous and burst modes. A full exposure time T A is used in continuous mode processing, but a shorter duration T B is used in burst mode. The area labeled by 0 denotes the data gap T G. T p = T B +T G is burst repetition period. (Adopted from Holzner and Bamler, 2002) In the bottom panel of Fig.2, every dashed area is a group of contiguous transmitted/received pulses. It is called a burst which consists of PRF T B pulses followed by a gap. Since Doppler FM rate remains constant in T A A3

4 (so-called LFM), the azimuth resolution (= V sat / f B ) is T A /T B times that expressed by (1b). Burst mode was first used in the Magellan mission to Venus in 1990 to conserve transmit power and data downlink capacity. In addition, radar system can use the gaps to collect additional information such as different polarization (switching H/V) or rang swaths (switching antenna beams); however, lower spatial resolution images are obtained. 3. SCANSAR MODE Based on the concept of burst mode operation, a multi-beam SAR antenna can illuminate different sub-swath through different beam. By stitching these sub-swath images together, a wide swath image is obtained. ScanSAR technique was first used on SIR-C and SIR-X shuttle missions in ScanSAR Operations SS5) are embedded in the gap (labeled as scan interval) of SS1 burst imaging. The operation procedure is briefed as following, (a) Turn on beam-1, (b) Transmit/Receive N 1 pulses through beam-1, (N k = PRF k T B,k for beam-k) (c) Wait (~5ms) for residual echoes from sub-swath 1 being received by beam-1, (d) Disable beam-1 and turn on beam-2, (e) Operation of beam-2 is the same as the step (b) to (d), and so as beam-3, beam-4 and beam-5; (f) Disable beam-5 and turn on beam-1, (g) Repeat step (b) to (f), until data from the demanded area has been collected. 3.2 Burst Design For a scansar operation with using 5 antenna beams, the gap duration T G in each sub-swath must contain at least 4T B for operating the other 4 beams. That is the exposure time T A 5T B in this single look case. Usually, T A = T B + T G + T B = 6T B is selected where additional T B is used to guarantee all targets can be completely illuminated by bursts. Figure 3. Concept of ScanSAR imaging by using 5 beams. Left panel shows the variations of burst/gap structure of 5 sub-swaths in Envisat mission. (Adopted from Envisat mission on Web) Fig.3 illustrates an example of scandsar operation using 5-beam steering. It shows that 4 additional burst imaging (sub-swath SS2 to Conventionally, global SAR satellite missions use multi-looking with N look = 2~4 to reduce radiometric discontinuities resulted from the inadequate Doppler centroid estimation for scalloping compensation in scansar processing (e.g., by SPECAN algorithm). Selection of N look is limited by beam switching time which is, for our reflector antenna, determined by the architecture of RF network. Examples of the operation parameters for 2-look scansar imaging in a range of the incident angles are shown in Tables 3. Note A4

5 that slant range, synthetic aperture length and full exposure time remained as in stripmap mode. Table 3. Estimated parameters for scansar operations with N look = 2 and N beam = 5. Incident angle Slant range (km) Synthetic aperture length (km) (full) Exposure time, T A (sec) Burst duration, T B (sec) Pulses per burst (no.) Capability of NSPO SAR System Comparing to (1b), the azimuth resolution of the scansar can be pre-estimated by δδ aaaa = VV ssssss (TT BB TT AA )BB DD. (3) NESZ can be obtained by substituting the azimuth compression ratio N a = PRF T B in derivation of (2). Since azimuth resolution in scansar is coarser than other imaging mode, in order to retain the reliable NESZ for ocean applications, higher range resolution is unnecessary. Roughly setting δ gr = δ az (i.e., squared pixels), the higher sensitive NESZ is able to derive as in Table 4. Table 4 shows that some of NESZ are the same when using different N beam, for instance, N beam = 2 and 3 at 30 incident angle. Considering B = c/(2δ gr sinϕ) and δ gr = δ az, NESZ is controlled by receiver bandwidth B r. NESZ presents only two values at fixed incident angles, either B r = 10MHz or 30MHz will be used in NSPO scansar imaging. The high sensitive NESZ and wide swath in scansar mode powerfully supports the ocean observations. The median/low resolution data can be efficiently processed by SPECAN algorithm. Another Full-Aperture processing can preserve phase information even for multi-look imaging. Consequently, the scansar data can also be applied to SAR interferometry. 4. SPOTLIGHT MODE In the spotlight imaging, SAR antenna always illuminates to a small ground area. The longer exposure time (and synthetic aperture length) the higher azimuth compression ratio and finer resolution. 4.1 Spotlight Operations Table 4. Achievable swath and NESZ in 2-look scansar imaging. Swath (km) / NESZ (db) N beam = 2 (~15m resolution) N beam = 3 (~21m resolution) N beam = 4 (~27m resolution) N beam = 5 (~33m resolution) km / -26.8dB 34.5km / -26.8dB 45.0km / -31.6dB 55.3km / -31.6dB 35.4km / -23.0dB 50.6km / -27.7dB 66.0km / -27.7dB 81.2km / -27.7dB 42.3km / -21.5dB 60.5km / -26.3dB 78.7km / -26.3dB 97.0km / -26.3dB Figure 4. Steering of SAR antenna and the resulted synthetic aperture length. (Adopted from Carrara et al., 1995) Fig.4 shows a symmetric imaging case of spotlight mode. In its left panel, a SAR sensor A5

6 flies from right to left. Spotlight imaging begins at a forward looking; the antenna footprint (dashed area) is determined by a solid angle of θ el θ az. Elevation beamwidth θ el forms the imaging swath as in stripmap and scansar modes. Azimuth beamwidth θ az limits a scene length as Scene length RRθ aaaa. (4) Azimuth FOVs of SAR antenna at beginning and ending sides form synthetic aperture angles θ 1 and θ 2 at near and far edges of imaging swath. For a small θ az, the geometry can be simplified as shown in right panel. The antenna squint angles are θ/2 at beginning and - θ/2 at ending. It can be obtained synthetic aperture length as L SA R θ, because that θ is also small as shown later. Substitute L SA to (1a), azimuth resolution expresses as δδ aaaa = λ 2 θ. (5) In fact, θ is based on the requirement of δ az. Examples of operation parameters for 1m resolution spotlight imaging at various incident angles are shown in Tables 5. Table 5. Estimated parameters for spotlight operations with δ az = 1m. Incident angle Synthetic aperture angle (deg.) Synthetic aperture length (km) Exposure time (sec) Azimuth scene size (km) Averaged steering rate (deg/s) Comparing with Table 3, synthetic aperture length and exposure time are about double of those in stripmap/scansar mode. Although the scene length (corresponding to the flying direction) seems shorter, the SAR antenna can be steered back for the next acquisition. The key point is that SAR satellite with a reflector antenna must continually carry out the in-flight yaw (and little pitch) steering to aim its antenna at imaged scene center. A steering rate of ~0.76 per second is necessary for spotlight operations. This value (= θ/t A = θ/(l SA /V sat ) = V sat /R) depends only on satellite altitude and imaging incident angle. 4.2 Capability NESZ is obtained by substituting the zimuth compression ratio N a = Rθ AZ /(λ/(2 θ)) in derivation of (2), where Rθ AZ means resolution of real aperture radar. Achievable NESZ for 0.5m, 0.7m and 1m resolution cases are shown in Table 6. Table 6. Achievable swath and NESZ in spotlight imaging. Swath (km) / NESZ (db) 0.5m resolution ( θ = 1.82 ) 0.7m resolution ( θ = 1.30 ) 1m resolution ( θ = 0.91 ) km / -12.3dB 13.8km / -13.5dB 13.8km / -15.7dB 20.2km / -8.4dB 20.2km / -9.6dB 20.2km / -11.4dB 24.2km / -7.0dB 24.2km / -8.2dB 24.2km / -10.0dB Three receiver bandwidth B r = 600, 430 and 300 MHz are used respect to 0.5m, 0.7m and 1m resolution. However, higher spatial resolution, lower NESZ. 5. SUMMARY The scansar and spotlight operations have been studied and the imaging capabilities based on specifications of the developing A6

7 NSPO SAR system has been briefly examined. Stripmap imaging provides disaster monitoring, land surveillance, agriculture and forestry monitoring, and so on. ScanSAR imaging is having the key advantages of much wider observation swath and higher sensitive NESZ, which satisfies the requirements of ocean observation, therefore, coarser spatial resolution is negligible. The scansar data can be applied to ship detection, maritime monitoring and military application including the internal wave observation etc. The higher spatial resolution spotlight imaging having a worse NESZ, however, it is sufficient to monitor detailed facilities and the buildings in disaster-affected area. Geoscience and Remote Sensing IEEE International Symposium IGARSS 95 conference, Vol.3, pp Cumming, I. G. and F. H. Wong, Digital processing of Synthetic Aperture Radar Data: Algorithm and Implementation. Artech House, pp Holzner, J. and R. Bamler, Burst-Mode and ScanSAR Interferometry, IEEE Transaction on Geoscience and Remote Sensing, Vol. 40, No. 9, pp Some issues are still under investigation, such as azimuth ambiguity analysis for scansar and spotlight modes, precise PRF respect to different sub-swath in scansar mode, and PRF in spotlight. ACKNOWLEDGEMENT This research is supported by the SAR project in National Space Organization (NSPO) and a grant from National Applied Research Laboratory (NARL) in Taiwan. REFERENCES Carrara, W. G., R. S. Goodman, and R. M. Majewski, Spotlight synthetic aperture radar: Signal processing algorithms. Artech House, pp Cheng, T., K. Leung, M. Jin, and E. Chu, ScanSAR and precision processor implementation at the Alaska SAR facility, A7