1174 PROCEEDINGSIEEE, OF THE VOL. 70, NO. 10, OCTOBER 1982 Spaceborne Synthetic-Aperture Imaging Radars: Applications, Techniques, and Technology Absrrclet-ln the last four years, the f‘ii two Earthabiting, space or object being imaged. Thus thesize of the resolutionelement borne, synthetic-aperture imaging radars (SAR) were sudydevel- increaseslinearly withthe observing wavelength and sensor oped and operated. This was a major achievement in the development altitude, and is inversely proportional to the aperture size. In of spaceborne radar sensors and ground processors. The data acquired the optical and IR regions, very high resolution is achievable with these semrs extended thecapability of Earth re(ywces and ocean- surfice observation into a new region of the electromagneticspectrum. from orbit with reasonable size apertures because of the short This paper is a review of the different aspects of spaceborne imaging operating wavelength. In the microwave region, the operating radnrs. It includes a review of: 1) the unique chprncteristicr of space- wavelength is relatively large, and apertures of many hundreds borne SAR systems; 2) the state of the art in spaceborne SAR hardware and SAR optical and digital processors; 3) the different data-handling of meters to manykilometers are required to achieve high techniques; and 4) the different applicationsof spaceborne SAR data. resolution of tens of meters or less. This, of course, is imprac- tical at the present time. I. INTRODUCTION The SARsensor circumvents this limitation by using the N JUNE 1978, theSeasat satellite was put into orbit around ranging and Doppler tracking capability of coherent radars to the Earth with a synthetic-aperture imaging radar (SAR) as acquirehigh-resolution images of the surface fromorbital one of the payload sensors [ 121 . The Seasat SAR provided, altitudes. Two neighboring targets are separated by their dif- I ferential time delay and Doppler history, neither of which is a for the fit time, synoptic radar images of the Earth’s surface with a resolution of 25 m. The success of this complex sensor function of the distance to the sensor. Thus the resolution of was a major technological advance, and it opened up a new aSAR system is independent of the sensor altitude. This dimension in our capability to observe, monitor, and study the unique advantage does impose some restrictions on the sensor Earth’s surface [31],[39]. In November 1981,the second imaging swath,antenna size, and powerrequirements. The imaging radar was successfully operated from space onthe basic properties of spaceborne SAR systems are discussed in Shuttle [ 331. The Shuttle Imaging Radar-A (SIR-A) acquired Section I1 andthe technologicalaspects of the sensor are images over a variety of regions around the world with an imag- presented in Section 111. ing geometry different from the one used by the Seasat SAR. Because the SAR uses the Doppler history to achieve high Spaceborne photographybecame available in the early 1960’s resolution in one of the spatial dimensions, each pixel is gen- with the advent of the space age. This was followed in the late erated by processing a large number of successive echoes. This 1960’s and 1970’s with the acquisition of multispectral visible leads to a large number of arithmetic operations in order to and infrared (IR) imagery, thermal imagery, and passive micro- generate the image. This arithmetic complexity is beyond the wave imagery. These sensors allowed us to acquire informa- present capability of real-time processors. The development of tion about the surface by studying its emitted energy in the digital processors for spaceborne SAR data is a very active and microwave and IR regions of the spectrum and the reflected challenging research field. This aspect of the SAR system is energyin the visible and near-IR regions. All these sensors the subject of Section IV. are passive in nature, i.e., they detect the energy which is gen- A very elegant way of executinga large number of the erated by the sun or thesurface. types of calculations required for SAR data processing can be The SAR imaging sensor provides information about the sur- achieved withoptical techniques. Inactuality, the first air- faceby measuring and mapping the reflected energy in the borne SAR systems used optical processors. In Section V, we microwave region, thus extending the capability of sensing the discuss how these types of processors have been adapted to surface properties into a new dimension. In addition, because handle the unique aspects of spaceborne SAR data. it uses its own energy and operates at a relatively long wave- Once the image has been formed, a number of post-image- length,it acquiressurface imagery at all times, i.e., day or formation processing steps are used to maximize the usefulness night and through cloud cover. Thus it has the unique capa- of the information in the radar imagery. These include, among bility required for continuous monitoring of dynamic surface others, radar image registration to multispectral visible/IR phenomena. images, automatictextural analysis, andspeckle statistical The imaging resolution of passive sensors is equal to their analysis. These techniques are addressed in Section VI. angularresolution (i.e., observing wavelength over aperture Section VI1 addresses interpretation techniques and applica- size) multiplied by the range between the sensor and the area tions of spaceborne SAR data. We review the techniques and image features that an interpreter uses in-extracting informa- tion about the surface,and we presenta variety of specific Manuscriptreceived February 1, 1982; revisedJuly 29,1982. The researchdescribed in this paper was performedunder Contract with examples in the different areas of the Earth Sciences-geology, the National Aeronautics and Space Administration. The submission of oceanography, glaciology, and agriculture. this paper was encouraged after the review of an advance proposal. The authors are with the Jet Propulsion Laboratory, California Insti- At the end of each section, we briefly discuss the develop- tute of Technology, Pasadena, CA 91 109. ments expected in the near future. In Section VIII, we present 0018-9219/82/1000-1174t00.75 0 1982 IEEE ELACHI et al.: SPACEBORNE SAR’s 1175 \I Fig. 2. Imaging radar viewing geometry: (a) in the range plane, (b) in the azimuth plane. over a certain region, the received echoes contain a complete Doppler history and range-change history for each point on Fx. 1. Coordinate system for SAR image formation. A set of circles the surface that is being illuminated. These complete histones and hyperbolasdefme the equirange and equi-Doppler lines, are then processed to identify uniquely each point on thesur- respectively. face and to generate the image. This is why a very large num- ber of operations are required to generate each pixel in the our opinion of the major challenges in the field of SAR remote image. Such is not the case with optical sensors. A simplified sensing forthe 1980’s and briefly review the development comparison is that the radar sensor generates the equivalent being planned for spaceborneSAR systems during this decade. of a hologram of the surface, and furtherprocessing is required to obtain the image. This processing can be done either opti- II. SPACEBORNESAR PRINCIPLE cally or digitally. In the synthetic-aperture technique, the Doppler information One unique featureof the synthetic-aperture imaging radar is in thereturned echo is usedsimultaneously withthe time- that its resolution capability is independent of the platform delay information to generate a high-resolution image of the altitude. This is a result of the fact that the image is formed surface being illuminatedby radar [ 161, [701, [261, [441, by using the Doppler history and the differential time delays, [ 521. The radar usually “looks” to one side of the moving none of which is a function of the range from the radarto the platform-to eliminate right-left ambiguities-and perpendicu- surface. This unique capability allows the acquisition of high- lar to its line of motion. It transmits a short pulse of coherent resolution images from satellite altitude as long as the received electromagnetic energy toward the surface. Points equidistant echo has sufficient strength above the noise level. from the radar are located on successive concentric spheres. In this section, we will discuss the main features of space- The intersection of thesespheres with a flat surface gives a borne SAR systems. These include: azimuth and range ambi- series of concentric circles centeredat the nadir point (see guities, range walk, the effects of Earth’s rotation and orbital Fig. 1). The backscatter echoes from objects along a certain characteristics, surface interactions, SAR image characteristics, circle will have a well-defined time delay but differentDoppler key tradeoffparameters influencing the sensor design, and characteristics. nonconventional SAR systems. Points distributed on coaxial cones, with the flight line as the axis and the radar as the apex, provide identical Doppler A. Ambiguities shifts of the returned echo but different delays. The intersec- There will be ambiguity in the response if the PRFis so high tion of these cones with a flat surface gives a family of hyper- that return signals from two successive transmitted pulses arrive bolas(Fig. 1). Objects on aspecific hyperbola will provide simultaneously at the receiver. This is called range ambiguity. equi-Doppler returns.Thus if the time delay and Doppler Conversely, if the PRF is so low that the returnis not sampled information in the returned echoes are processed simultane- at the Nyquist rate, there will be Doppler azimuth ambiguity. ously, the surface can be divided into a coordinate system of The upper limit of the PRFis fixed by the range or elevation concentric circles andcoaxial hyperbolas (Fig. l), and each beamwidth of the SAR antenna. A view of the beam geometry point on the surface can be uniquely identified by a specific in the range plane is shown in Fig.