Synthetic-aperture_radar

# Synthetic-aperture radar **Synthetic-aperture radar** (SAR) is a radar technique that forms a high-resolution image of the ground by combining the echoes a single, physically modest antenna collects while moving past a scene, rather than by building a real antenna large enough to resolve that scene on its own. Carried on an aircraft or a spacecraft, the radar transmits many pulses as it flies a track past the target area; because every echo also carries the exact position along that track from which it was recorded, a processor can later combine them as though they had all arrived at once, at an antenna as long as the whole flight path the pulses spanned. The result is an antenna synthesized after the fact rather than built in metal, and a form of imaging that works through cloud and in complete darkness because it supplies its own illumination. A three.js microsim elsewhere on the site renders this synthesis directly, an antenna's own motion standing in for a far longer one it could never physically carry. Because the synthesized aperture can be made far longer than any dish an aircraft or a satellite could carry, SAR reaches an along-track resolution that ordinary, real-aperture [[Radar|radar]] can match only by growing the antenna itself or moving to a much shorter wavelength. The same coherent recording that sharpens the image also carries the raw material for measuring how a scene has shifted between two passes, and for building terrain models from the geometry of two slightly separated tracks. [[Inverse_synthetic-aperture_radar|Inverse synthetic-aperture radar]] turns the idea around, holding the antenna fixed and exploiting a target's own rotation or motion to synthesize the same kind of aperture, a technique this article returns to only in passing since it has one of its own. ## Motivation and applications A real antenna's angular resolution is set by its size measured in wavelengths, so the finer the image a designer wants, the larger the antenna needs to be; at the ranges and wavelengths radar normally works at, an antenna good enough to image the ground finely from aircraft or orbital altitude would be far too large to fly. Synthetic-aperture radar sidesteps the problem by using the platform's own travel to stand in for that missing antenna, synthesizing an aperture many times the real one's length at no cost in mass or drag. Because the radar supplies its own illumination, it images through cloud and complete darkness alike, which is what makes it useful for continuous Earth observation regardless of weather or time of day: mapping land cover and topography, tracking coastlines, glaciers and floods between passes, and measuring ground subsidence or a volcano's slow swelling before it erupts. The technique reaches beyond Earth as well, overlapping there with [[Radar_astronomy|radar astronomy]]: [[NASA]]'s Magellan orbiter used SAR through the permanently overcast atmosphere of [[Venus]] to map most of the planet's surface at a resolution no optical camera in orbit could have matched.[^magellan] ## Basic principle A stationary [[Radar|radar]]'s cross-range resolution at range R is set by its beamwidth, which for a real antenna of along-track length D and wavelength λ is of order λ/D radians, so the resolvable distance on the ground, the beamwidth multiplied by the range, worsens the farther away the target sits. SAR keeps the same small antenna but uses how long the platform takes to cross that widening beam: a point on the ground stays illuminated for a stretch of track that itself grows with range in just the way needed to offset the loss. Recording every pulse's amplitude and phase over that whole dwell, then filtering the record as though it had come from one very long antenna rather than a sequence of short looks from a small one, focuses the image. Carried through, the achievable along-track resolution settles near half the real antenna's own length, `D/2`, independent of both range and wavelength, a genuinely counterintuitive result: a smaller antenna, which dwells on each point longer and so synthesizes a longer aperture, can match a larger one's resolution rather than lose to it. This trade is the concept the three.js companion elsewhere on the site works through directly. ## Algorithm Turning raw echoes into a focused image is, at bottom, a two-dimensional matched-filtering problem. One filter compresses each pulse along the line of sight into a fine range profile, the same [[Pulse_compression|pulse-compression]] principle any [[Chirp|chirped]] radar relies on; a second filter compresses the whole set of range profiles along the flight track into a fine cross-range profile, undoing the changing-range phase history a stationary point on the ground traces out as the platform passes it. Early systems did this optically, passing exposed film through a lens system that itself acted as the matched filter; digital processors now do the same job with the [[Fast_Fourier_transform|fast Fourier transform]] and related frequency-domain methods. ## Existing spectral estimation approaches The classical way to focus the cross-range dimension is close to a Fourier transform of the recorded pulse history: fast and well understood, at the cost of sidelobes set by the aperture length and its weighting, a family of problems that also appears more generally as [[Least-squares_spectral_analysis|least-squares spectral analysis]]. Adaptive alternatives such as the Capon (minimum-variance) method, amplitude-and-phase estimation (APES), and the iterative sparse asymptotic minimum-variance (SAMV) method instead choose a data-dependent filter for every pixel, buying lower sidelobes and finer separation of close scatterers at real computational cost. Parametric subspace methods go further, modelling a resolution cell as a small, countable number of discrete scatterers: the eigenvector method and the related MUSIC algorithm split the data into a signal subspace and a noise subspace and locate the scatterers from the signal subspace alone, which can outresolve the classical Fourier limit when the assumed scatterer count is right. Backprojection instead sums the matched-filtered return from every pulse directly into every image pixel in the time domain, at heavy computational cost, but without assuming a straight, evenly sampled flight path, which is what lets it image the slow, irregular geometry of a geosynchronous-orbit synthetic-aperture radar (GEO-SAR). None of these [[Estimation_theory|estimation]] approaches dominates: the classical route stays cheapest for routine mapping, and the rest buy sharper images or geometric flexibility at real computational cost. ## Multistatic operation A monostatic radar transmits and receives from the same antenna and location; a multistatic synthetic-aperture radar instead separates the transmitter from one or more receivers, sometimes by a large distance, or drops a dedicated transmitter altogether and receives echoes of a broadcast signal already illuminating the scene, the same [[Passive_radar|passive]] strategy ordinary [[Radar|radar]] can use. Spreading transmitter and receivers apart buys extra viewing angles on the same ground, useful for building terrain height models or for seeing how a target's [[Radar_cross_section|radar cross-section]] varies from more than one direction at once, and it lets a receiver work silently, without ever radiating a signal that could reveal its own position. The cost is a harder problem to solve, since every receiver's clock and position must be known relative to the transmitter's to whatever precision the image demands. ## Scanning modes A synthetic-aperture radar can point and hold its beam relative to the flight path in three broadly different ways, trading swath width against azimuth resolution differently each time. ### Stripmap mode airborne SAR In stripmap mode the antenna holds a fixed angle to the flight path throughout the pass, illuminating a continuous strip of ground beside the track. It is the simplest mode to fly and to process, and it images an unbounded length of ground, but its azimuth resolution is fixed by the antenna's own beamwidth and cannot improve without changing the hardware. ### Spotlight mode SAR Spotlight mode instead steers the beam to keep illuminating one patch of ground well past the moment a fixed stripmap beam would have swept off it. The longer dwell synthesizes a longer aperture and so a finer azimuth resolution than stripmap gives from the same real antenna, at the cost of imaging only that one chosen patch rather than a continuous strip. ### Scan mode SAR Scan mode, or ScanSAR, goes the other way, periodically steering the beam between several sub-swaths at different look angles during a single pass to cover a much wider total swath than stripmap manages in one pass; because each sub-swath gets only a fraction of the full dwell, its azimuth resolution is correspondingly coarser. ## Special techniques Beyond forming a plain reflectivity image, several further techniques pull more out of the same recorded data or adapt the synthetic-aperture idea to other geometries. ### Polarimetry A polarimetric SAR transmits and receives combinations of horizontal and vertical polarization, since how a scatterer twists the returned polarization carries shape and material information a single-polarization image discards. An elementary system records one such combination; a full quad-polarization system records all four and combines them coherently, keeping phase to separate scattering mechanisms, or incoherently, using intensity alone. ### Interferometry Interferometric SAR (InSAR) compares the phase of two images of the same ground from slightly different positions or dates, since that phase difference records the extra distance travelled to within a small fraction of a wavelength. A spatial baseline turns the difference into a terrain height model; a temporal baseline between passes months or years apart turns it into a map of how the ground itself moved, a differential-interferometry (DInSAR) measurement fine enough to track subsidence or a volcano's slow inflation. Tomo-SAR extends the idea to many passes at once, separating scatterers stacked at different heights that a single interferometric pair would blend into one resolution cell. ### Ultra-wideband SAR An ultra-wideband SAR trades a large fraction of its centre frequency for bandwidth, sharpening range resolution directly and, at the lower end of the radio spectrum, letting some energy penetrate foliage or dry soil a narrower-band system would simply reflect from the surface of. ### Doppler-beam sharpening Doppler-beam sharpening applies the principle behind SAR's azimuth focusing, that targets at different angles within the beam carry slightly different [[Doppler_effect|Doppler]] shifts as the platform moves, without forming a full synthetic aperture; it narrows the effective beam somewhat, useful in some airborne mapping and weather-avoidance radars, but stops well short of a properly focused SAR image's resolution. ### Chirped (pulse-compressed) radars Almost every synthetic-aperture radar transmits a [[Chirp|chirp]], a pulse whose frequency sweeps linearly over its duration, because a chirp compresses on receive, through a [[Matched_filter|matched filter]], into a pulse far shorter than the one transmitted, giving the range resolution of a short pulse while keeping the energy only a long one can carry. ## Typical operation Two coupled resolution budgets govern a single acquisition, and the same pulses that set them then have to be turned into a usable image. ### Data collection Range resolution comes from the transmitted pulse's bandwidth after compression: more bandwidth compresses to a shorter effective pulse and so separates two targets at nearly the same range. Azimuth resolution comes from the real antenna's beamwidth acting through the synthesis described above, so a narrower beam yields a longer synthetic aperture and a finer along-track resolution at a given range. Meeting both means transmitting a steady train of pulses at a chosen [[Pulse-repetition_frequency|pulse-repetition frequency]] as the platform advances, with each pulse's full amplitude and phase stored for later processing; as with any radar receiver, how faint an echo can still be used is set by the receiver's own noise floor and the resulting [[Signal-to-noise_ratio|signal-to-noise ratio]], which coherent combination across the whole aperture, not any one pulse, overcomes.[^noisefloor] ### Data processing The simplest product is an amplitude image, whose pixel brightness records how strongly a patch of ground scattered the illuminating microwaves back. The phase recorded alongside every pixel is discarded in a plain amplitude image but is exactly what interferometry and polarimetric decomposition need. Because many small, randomly placed scatterers within one resolution cell interfere with one another, the same coherent recording that focuses the image also studs it with a grainy, salt-and-pepper texture called speckle, a genuine consequence of coherent imaging rather than added noise, usually tamed by averaging several looks of the same ground at some cost in resolution. The chain is often likened to optical holography, since both reconstruct an image from a recorded wavefront's amplitude and phase only afterward, rather than focusing it directly the way a lens does. ## Image appearance A focused SAR image reads like an aerial photograph at a glance, but its geometry and its brightness both follow rules a camera does not share. ### Range, cross-range, and angles The image's two axes are slant range, the round-trip distance actually measured, and azimuth, the along-track position; turning slant range into a true ground distance, correcting for the terrain's own height, is a separate geometric step before the image lines up with an ordinary map. ### Visibility Because the radar looks sideways at a shallow grazing angle rather than straight down, how much of a slope is visible, and how bright it appears, depends strongly on the angle between that slope and the beam rather than on the slope's own reflectivity alone. ### Mirroring artefacts and shadows A slope tilted toward the radar is compressed in the image, foreshortened, and once it tilts past vertical to the beam its base and its top can arrive in the wrong order, laying the slope over on top of the ground in front of it; a slope tilted away receives no illumination and appears as a blank radar shadow. ### Objects in motion A moving target's echo carries a different Doppler history from a stationary one's, so the filter matched to a fixed point smears it, displaces it from its true position, or both; the same mismatch is exploited deliberately in [[Moving_target_indication|moving-target indication]] to flag vehicles against a stationary background, though it remains a nuisance for ordinary mapping. ## Commercial industry Synthetic-aperture radar began as a government and military pursuit and remains one, but it has also grown into a commercial data industry, with privately operated satellite constellations selling imagery, change detection and analytics built on SAR's ability to see through cloud and in darkness to insurers, financial analysts, agriculture and shipping alongside its older civil and defence customers. Because the raw imagery is difficult for a non-specialist to interpret directly, much of the commercial value sits in processed products, automated flood maps, ship detections and ground-deformation alerts, rather than in the pixels themselves. ## History Side-looking airborne radar, which points its beam out to the side of the aircraft rather than ahead, supplied the reconnaissance imagery synthetic-aperture radar grew from, and the technique itself is generally credited to Carl Wiley at the Goodyear Aircraft Corporation in 1951, who recognised that the changing Doppler shift of ground returns as an aircraft flew past them could sharpen along-track resolution far beyond what the real antenna beam allowed.[^wiley] It matured through military reconnaissance programmes over the following decades before reaching civilian Earth observation: [[NASA]]'s Seasat, launched in 1978, carried the first spaceborne synthetic-aperture radar flown for civilian oceanographic and land remote sensing, and though the mission itself operated only briefly, it proved spaceborne SAR practical and shaped the missions that followed it.[^seasat] The Magellan mapping of Venus described above, and the long series of civil and military SAR satellites flown since Seasat, carried the technique from a wartime reconnaissance tool into a routine instrument of orbital remote sensing. ## Relationship to phased arrays A [[Phased_array|phased array]] and a synthetic-aperture radar solve the same problem by different means: both form a narrow, steerable response by combining many coherent samples with the right relative phase, the same [[Beamforming|beamforming]] principle either way. A phased array does this instantly, with many physical elements existing at once in space; a SAR does it after the fact, with one small antenna occupying each of those positions in turn as the platform flies past, the combination happening in a processor rather than a feed network. Because the two are mathematically the same operation carried out in space or in time, they can be combined: an electronically scanned array steered during a synthetic-aperture pass gains the array's fast, agile pointing together with the synthetic aperture's fine along-track resolution. ## Microsims This article carries no p5.js sketch of its own; a three.js microsim elsewhere on the site renders the aperture-synthesis concept above, an antenna's own motion standing in for a much longer one, in three dimensions. Two neighbouring Wikitube sketches make good companions to it. The Radar sketch models the pulse-timing fundamentals every synthetic-aperture radar still relies on for its range dimension: a pulse-repetition frequency that sets an unambiguous range, and a pulse width that sets how finely two close targets separate, the range-side counterpart to the azimuth-side synthesis described above. The Doppler effect sketch shows the changing frequency a moving source produces ahead of and behind itself, the same geometric fact SAR's azimuth processing exploits to turn a target's changing angle into a changing frequency. A related idea appears underwater: the Sonar article describes synthetic-aperture sonar, which synthesizes an aperture from a moving platform's pings by the same reasoning, at the speed of sound rather than of light. *Try:* in the [[Radar]] sketch, drag the pulse width narrower and watch the zoomed inset separate two close targets that a wider pulse had merged into one, the range-axis counterpart to azimuth resolution. *Try:* in the [[Doppler_effect]] sketch, raise the source speed and watch the frequency ahead of it climb while the frequency behind it falls, the same sign change a ground point's Doppler shift goes through as a SAR platform passes it. *Try:* in the [[Sonar]] sketch, watch the same round-trip echo timing this article's antenna-motion picture assumes, played out with sound instead of radio.  **Microsim — three.js (Wikitube framework), pending deploy:** *Synthetic-aperture radar: a long walk becomes a long antenna* will play here once `https://wikitube-3d-microsims.netlify.app/radar/Synthetic-aperture_radar.html` is live.  *Built from `MICROSIM_GUIDE/specs/sims/Synthetic-aperture_radar.json`; part of the [[PORTAL_Radar|Radar]] set.*  ## Wikipedia : Wikitube **Strict pair:** [Wikipedia](https://en.wikipedia.org/wiki/Synthetic-aperture_radar) : [Wikitube](https://en.wikitube.io/wiki/Synthetic-aperture_radar) Skeleton mirrored at revision 1374355414. Prose, emphasis and the microsims are Wikitube's own. ## See also - [[Inverse_synthetic-aperture_radar]] - [[Radar_astronomy]] - [[Radar]] - [[Pulse_compression]] - [[Beamforming]] - [[Phased_array]] - [[Doppler_effect]] - [[Sonar]] - [[Matched_filter]] ## References Standard SAR theory — the real-aperture resolution limit, the D/2 azimuth-resolution result, and the matched-filtering basis of range and azimuth compression — is textbook material used throughout radar engineering and is not separately footnoted here, per the Wikitube style guide's §6.1. Page numbers below are PDF pages of the open edition linked in Bibliography. [^magellan]: Citation needed: NASA's Magellan mission fact sheet or press kit would confirm the exact mapping dates and the percentage of Venus's surface imaged. [^wiley]: Citation needed: Carl Wiley's 1951 Goodyear Aircraft Corporation report or patent describing Doppler-based azimuth sharpening would confirm the date and the precise claim of priority. [^seasat]: Citation needed: NASA's Seasat mission summary would confirm the 1978 launch date, the instrument complement, and the cause and date of the mission's early end. [^noisefloor]: Ellingson, S. *Radio Systems Engineering, Revised First Edition*. 2023, pp. 96-101 (PDF pages). Open Textbook Library: https://open.umn.edu/opentextbooks/textbooks/radio-systems-engineering . CC BY-NC. ## Bibliography - Christian Tiberius; Max Mulder. *Engineering Signal Analysis: From Fourier to filtering: Theory*. 2026. Open Textbook Library: https://open.umn.edu/opentextbooks/textbooks/engineering-signal-analysis-from-fourier-to-filtering-theory . CC BY. - Don Johnson. *Fundamentals of Electrical Engineering I*. 2014. Open Textbook Library: https://open.umn.edu/opentextbooks/textbooks/fundamentals-of-electrical-engineering-1 . CC BY. - Michael Stiber; Bilin Stiber; Eric Larson. *Signal Computing: Digital Signals in the Software Domain*. 2020. Open Textbook Library: https://open.umn.edu/opentextbooks/textbooks/signal-computing-digital-signals-in-the-software-domain . CC BY-SA. - Allen Downey. *Think DSP: Digital Signal Processing in Python*. 2012. Open Textbook Library: https://open.umn.edu/opentextbooks/textbooks/think-dsp-digital-signal-processing-in-python . CC BY-NC. - Steven Ellingson. *Radio Systems Engineering, Revised First Edition*. 2023. Open Textbook Library: https://open.umn.edu/opentextbooks/textbooks/radio-systems-engineering . CC BY-NC. - John Dyer; Chad Davis. *Measurement and Instrumentation: An Introduction to Concepts and Methods, 1st Edition*. 2020. Open Textbook Library: https://open.umn.edu/opentextbooks/textbooks/measurement-and-instrumentation-an-introduction-to-concepts-and-methods . CC BY-NC-SA. ## External links This article carries no p5.js sketch of its own; its three.js companion is embedded in Microsims, above.