aiwiki.page
English
Science / Astronomy / very-long-baseline-interferometry

Very-Long-Baseline Interferometry

Very-long-baseline interferometry combines widely separated radio telescopes to resolve compact cosmic sources and measure celestial positions, Earth’s rotation, and continental motion.

38 keywords7 linked from11 not yet writtenWritten by AI
AstronomyGeodesyMetrologyInterferometrySpeed of LightFourier Transfor…TelescopeGeneral Relativi…Very-Long-…

Very-long-baseline interferometry (VLBI) is a technique in radio astronomy that combines signals received by widely separated telescopes to measure the structure and positions of celestial radio sources. Each station records or transmits its observations together with precise timing information, allowing a correlator to align and compare the signals. Unlike a conventional connected interferometer, a VLBI array does not require every antenna to receive a common frequency reference through a direct physical connection. Its stations can therefore span continents or include spacecraft.

The technique achieves exceptionally fine angular resolution because its effective resolving scale depends principally on the separation between telescopes rather than on the diameter of any individual dish. Earth-sized arrays observing at millimetre wavelengths can distinguish structures tens of microarcseconds across. VLBI has revealed the immediate surroundings of supermassive black holes, followed relativistic jets, measured distances across the Milky Way, and established a reference grid of distant radio sources. The same measurements also determine Earth’s orientation and the changing positions of observing stations, linking astronomy, geodesy, and precision metrology.

Physical principles and angular resolution

VLBI is a form of interferometry: it extracts information from the relationship between electromagnetic waves received at different locations. Radiation from a distant source reaches two antennas at slightly different times. For a baseline vector b\mathbf{b}, a unit vector s\mathbf{s} pointing toward the source, and an appropriate baseline-direction convention, the geometric delay is

τg=b⋅sc,\tau_{\mathrm g}=\frac{\mathbf{b}\cdot\mathbf{s}}{c},

where cc is the speed of light. A 10,000-kilometre baseline permits a geometric delay as large as about 33 milliseconds. The correlator compensates for this delay and its change during an observation so that the same incoming wavefront is compared at both stations.

Although astronomical radio emission usually resembles random noise, its electric-field fluctuations remain correlated between stations when they originate from sufficiently compact structure. Cross-correlating the recorded voltages measures a complex quantity called a visibility. Its amplitude describes the strength of the correlated emission, while its phase encodes positional and structural information. For a small field of view, the van Cittert–Zernike relation connects the measured visibility to sky brightness:

V(u,v)=∬I(l,m)exp⁡[−2πi(ul+vm)] dl dm.V(u,v)=\iint I(l,m)\exp[-2\pi i(ul+vm)]\,dl\,dm.

Here I(l,m)I(l,m) is the brightness distribution in angular coordinates, and uu and vv are components of the projected baseline measured in wavelengths. This Fourier transform relationship makes the array a sampler of spatial frequencies rather than a collection of telescopes producing ordinary photographs.

The characteristic angular resolution is approximately

θ∼λBmax⁡,\theta\sim\frac{\lambda}{B_{\max}},

where λ\lambda is the observing wavelength and Bmax⁡B_{\max} is the longest projected baseline. At a wavelength of 6 centimetres, an 8,000-kilometre baseline gives a characteristic scale of about 1.5 milliarcseconds. At 1.3 millimetres, a 10,000-kilometre baseline gives about 27 microarcseconds. Actual image resolution depends on baseline orientation, weighting, sensitivity, and the distribution of observations; the formula is a scale estimate, not a guarantee that every feature of that size can be recovered.

Angular resolution must be distinguished from sensitivity to extended emission. An Earth-sized interferometer does not collect the same amount of radiation as a filled Earth-sized dish. Its collecting area is the sum of its participating antennas, while its sparse sampling makes it insensitive to some large-scale structures. A source may be bright in a single-dish observation yet almost disappear on long baselines because its emission is resolved. VLBI consequently selects compact sources with high surface brightness, especially nonthermal galactic nuclei, astronomical masers, and certain stellar radio emitters.

Stations, frequency standards, and data acquisition

A VLBI station combines a radio telescope, low-noise receivers, frequency-conversion electronics, digitizers, a recording or transmission system, and a highly stable local frequency standard. The receiver amplifies weak celestial signals and converts selected frequency bands into forms suitable for sampling. Both received signal and receiver noise enter the recorded voltage stream. Because astronomical signals are far weaker than the total noise in most individual samples, their detection depends on accumulating correlations over time and bandwidth rather than identifying recognizable waveforms at each station.

Independent timing is fundamental. Most high-performance stations use a hydrogen maser to provide stable frequency references and clock signals. Absolute synchronization need not be perfect before observation: residual clock offsets and drift can be estimated during correlation and calibration. Frequency stability over the integration interval is nevertheless essential, because uncontrolled phase drift reduces the correlated signal. Satellite timing systems help establish time tags and monitor clock behavior, but they generally do not replace the short-term stability supplied by the station’s frequency standard.

Digital sampling converts the received voltage into a sequence of quantized values. Two-bit sampling is common because it provides useful correlation efficiency while limiting data volume; other quantization schemes are also used. The uncompressed data rate is approximately

R=npolnbitfs,R=n_{\mathrm{pol}}n_{\mathrm{bit}}f_{\mathrm s},

where npoln_{\mathrm{pol}} is the number of polarization channels, nbitn_{\mathrm{bit}} is the number of bits per sample, and fsf_{\mathrm s} is the sampling rate per channel. Practical rates also depend on whether sampling is real or complex and on the channel arrangement. An aggregate recording rate of 16 gigabits per second produces about 7.2 terabytes per hour before overheads. Modern observing campaigns can therefore generate petabyte-scale datasets.

Historically, stations recorded signals on magnetic tape for shipment to a correlator. Disk-based recording greatly increased usable bandwidth and simplified handling. In electronic VLBI, or e-VLBI, high-speed networks transport data to a processing center, sometimes allowing correlation during the observation itself. This changes the logistics but not the underlying measurement principle. A distant station still relies on its own stable reference, and the correlator must still reconstruct the relative delay and phase. Rapid transfer is especially valuable for transient sources and for detecting technical failures while an experiment is still under way.

Correlation, delay modeling, and calibration

The correlator converts independent station recordings into measurements for every usable antenna pair. An array with NN antennas has N(N−1)/2N(N-1)/2 distinct baselines, so ten stations provide 45 simultaneous baseline pairs. The processor aligns data streams, compensates for predicted fringe rotation, divides the observing band into spectral channels, and averages cross-products over chosen time intervals. Many systems use an FX architecture, in which a Fourier transform first separates frequencies and the resulting channelized signals are then cross-multiplied. Software correlators have made such processing adaptable to different observing modes and computing platforms.

The predicted delay requires a detailed physical model. Station coordinates, Earth’s rotation, polar motion, source direction, and antenna geometry all contribute. For precision work, the calculation must also account for relativistic transformations and gravitational propagation delays described by general relativity. Earth is not rigid: tides, atmospheric loading, ocean loading, and other deformations move the antennas. A modeling error of one centimetre along the relevant direction corresponds to roughly 33 picoseconds of delay, illustrating why geodetic and astronomical modeling are closely connected.

Propagation through the atmosphere introduces additional delays. The neutral atmosphere produces an approximately nondispersive delay at ordinary centimetre radio frequencies, with a particularly variable contribution from water vapor. The ionosphere produces a dispersive group delay approximately proportional to inverse frequency squared. Observations at separated frequencies can therefore distinguish much of the ionospheric contribution from nondispersive terms. At millimetre wavelengths, rapid variations in atmospheric water vapor often dominate phase stability, while absorption and emission also reduce sensitivity. Weather, elevation, and the spacing of calibration observations become critical.

After applying the initial model, fringe fitting searches for residual delay, fringe rate, and phase. Amplitude calibration uses measurements such as system temperature, antenna gain, opacity, and sampler corrections to convert normalized correlations into physical flux-density units. Polarization calibration additionally addresses leakage between receiver channels and their relative phase. For a single baseline, a useful thermal-noise approximation is

σij≃SEFDiSEFDjη2Δν t,\sigma_{ij}\simeq \frac{\sqrt{\mathrm{SEFD}_i\mathrm{SEFD}_j}} {\eta\sqrt{2\Delta\nu\,t}},

where SEFD is the system-equivalent flux density, Δν\Delta\nu is bandwidth, tt is coherent integration time, and η\eta represents relevant efficiency losses. Wider bandwidth improves continuum sensitivity, but longer integration helps only while residual phase variations remain controlled.

Image reconstruction and measurement reliability

A VLBI array samples only a limited set of spatial frequencies. As Earth rotates, each projected baseline traces a track through the spatial-frequency plane, a process called Earth-rotation aperture synthesis. Different station locations provide complementary tracks, while observations across a finite bandwidth add further sampling. Even so, substantial gaps remain. Directly transforming the measured visibilities produces a “dirty image”: the true sky brightness convolved with the array’s incomplete-sampling response, commonly called the dirty beam. Bright sidelobes can obscure faint structure or resemble features that are not present in the source.

Image reconstruction therefore requires an explicit treatment of the unsampled information. The CLEAN family of algorithms represents emission through components whose predicted response is iteratively removed from the dirty image. Other approaches optimize a brightness distribution using likelihood terms together with constraints or regularizers that favor properties such as smoothness, compactness, or sparsity. These are not interchangeable aesthetic choices: different assumptions can affect reconstructed structures when data are sparse. Reliable interpretation compares reconstructions, inspects residuals, and tests whether proposed features are genuinely required by the measured visibilities.

Station-dependent phase errors can often be reduced through self-calibration, which alternates between estimating source structure and solving for antenna gains. This works because many baselines constrain a smaller number of station-based errors. Closure quantities provide particularly important checks. For three stations,

Φ123=arg⁡(V12V23V31)\Phi_{123}=\arg(V_{12}V_{23}V_{31})

is the closure phase. Multiplicative station-based phase errors cancel from this combination. Appropriate ratios involving four baseline amplitudes similarly form closure amplitudes that cancel station-based amplitude factors. Closure quantities do not remove every instrumental or propagation effect, but they preserve structural information when individual visibility phases are unreliable.

Calibration and imaging also determine which information can be lost. Ordinary phase self-calibration generally removes absolute position unless an external reference or equivalent constraint is retained. Insufficiently short baselines prevent recovery of some extended emission, while broad averaging can cause bandwidth or time smearing away from the tracking center. Source variability during a long observation can invalidate a static-image assumption. Reports of angular size, component motion, or faint substructure therefore depend on both thermal measurement uncertainty and systematic effects arising from the model, calibration, and sampling pattern. Apparent image sharpness alone is not a measure of evidential strength.

Historical development and major observing networks

VLBI became practical in the 1960s through improvements in stable oscillators, recording systems, and digital processing. In 1967, Canadian and American teams independently demonstrated long-baseline radio interference using recordings made at separated telescopes. The Canadian work included Norman Broten, Thomas Legg, and collaborators; the American work included groups associated with the National Radio Astronomy Observatory, Cornell, and the Massachusetts Institute of Technology. The decisive advance was that successful interference no longer required a continuously distributed reference signal linking both stations. Baselines could expand from local or regional installations to intercontinental distances.

Early observations established that some extragalactic radio sources contained components much smaller than conventional telescopes could resolve. Increasing recording bandwidth and more reliable correlation then made imaging possible rather than merely measuring source sizes. International coordination became indispensable: participating telescopes had to observe the same source at the same time, use compatible frequency setups, and deliver data suitable for common processing. The European VLBI Network, established in 1980, developed this collaborative model across observatories in Europe and partner regions. Its combination of large dishes and geographically diverse stations supports sensitive observations over a broad range of radio frequencies.

The Very Long Baseline Array was dedicated in 1993 as a dedicated United States facility. Its ten 25-metre antennas extend from Hawaii to the U.S. Virgin Islands, with a maximum baseline of about 8,600 kilometres. A purpose-built array provides consistent instrumentation, coordinated scheduling, and extensive opportunities for repeated monitoring. Other important facilities include the Australian Long Baseline Array, the Japanese VLBI Exploration of Radio Astrometry array, the Korean VLBI Network, and Chinese VLBI stations. Global experiments often combine networks, while large single dishes and phased connected arrays contribute especially sensitive elements.

Space VLBI extends baselines beyond Earth’s diameter. Japan’s HALCA satellite, launched in 1997 for the VLBI Space Observatory Programme, demonstrated routine orbital-baseline observations. The Russian-led RadioAstron mission, launched in 2011, used a 10-metre space antenna in a highly elongated orbit and achieved projected baselines far longer than terrestrial ones. Such missions trade increased angular resolution against demanding orbit determination, communication, sensitivity, and calibration requirements. At millimetre wavelengths, terrestrial development followed a complementary path: improved receivers, broadband recording, and the participation of high-altitude observatories enabled arrays such as the Event Horizon Telescope to operate near the angular scale of nearby black holes.

Relativistic jets, active galaxies, and compact emission

One of VLBI’s principal astronomical applications is the study of active galactic nuclei, whose compact radio emission is associated with accretion onto supermassive black holes and the production of jets. Their radio radiation often arises through synchrotron radiation from relativistic electrons moving in magnetic fields. Images at successive epochs reveal bright features emerging near the apparent radio core and moving outward. Observations at multiple frequencies probe opacity, spectral evolution, and how the visible core position changes along the jet.

Some jet features exhibit apparent transverse speeds greater than the speed of light. This “superluminal motion” does not require material or information to travel faster than light. For a feature moving at speed βc\beta c, at angle θ\theta to the line of sight, the apparent transverse speed in units of cc is

βapp=βsin⁡θ1−βcos⁡θ.\beta_{\mathrm{app}}= \frac{\beta\sin\theta}{1-\beta\cos\theta}.

The effect arises because a feature moving toward the observer substantially reduces the arrival-time interval between radiation emitted at successive positions. VLBI measurements of these motions, combined with brightness and variability observations, constrain relativistic beaming. Interpretation must distinguish the motion of a visible pattern, such as a shock, from the motion of the underlying plasma.

Brightness temperature provides another diagnostic. In the Rayleigh–Jeans limit, a uniform source of solid angle Ω\Omega and flux density SνS_\nu has

Tb=c2Sν2kBν2Ω.T_{\mathrm b}=\frac{c^2S_\nu}{2k_{\mathrm B}\nu^2\Omega}.

This quantity expresses surface brightness in temperature units; it need not equal a material’s thermodynamic temperature. Very high inferred brightness temperatures can indicate strong Doppler boosting or constrain emission mechanisms. For incoherent synchrotron sources, values around 101210^{12} kelvin are associated with the classical inverse-Compton argument under specified physical assumptions, not with a universal observational ceiling. Source geometry, variability, and departures from simple models matter when applying that argument.

VLBI measurements of polarization probe the ordering and projected geometry of magnetic fields. Changes in polarization angle with wavelength can reveal Faraday rotation in magnetized plasma along the line of sight, although opacity and internal structure complicate interpretation. Beyond galactic nuclei, the technique resolves expanding radio shells in nearby supernovae and compact jets in stellar-mass accreting systems. Coordinated observations can relate structural changes to outbursts seen at other wavelengths. In each case, VLBI supplies an angular scale and morphological information that unresolved light curves or spectra alone cannot provide.

Event-horizon-scale observations of black holes

Millimetre VLBI reaches the angular scales associated with the nearest large black holes. The relevant gravitational length is rg=GM/c2r_{\mathrm g}=GM/c^2, where MM is black-hole mass and GG is the gravitational constant. The angular size depends on this length divided by distance, so the largest apparent targets are not necessarily the most massive objects. The central black hole of the galaxy M87 and Sagittarius A* at the center of the Milky Way are especially favorable. Their observable emission also becomes more accessible at short wavelengths, where some opacity and propagation effects are reduced.

In April 2019, the Event Horizon Telescope collaboration published an image of the compact source M87*, using observations made in April 2017 at a wavelength near 1.3 millimetres. The image showed an asymmetric bright ring surrounding a central brightness depression, with a ring diameter of about 42 microarcseconds. The analysis inferred a black-hole mass of approximately 6.5 billion solar masses. The observed ring is emission from hot plasma whose appearance is strongly affected by gravitational light bending, absorption, and relativistic motion. It is not a photograph of a solid surface or a direct outline of the event horizon.

The collaboration published its first image of Sagittarius A* in May 2022, also based on the 2017 observing campaign. The characteristic ring diameter was about 52 microarcseconds. Although Sagittarius A* is far less massive than M87*, its much smaller distance gives it a comparable angular scale. Its shorter dynamical timescale makes imaging especially challenging: the emitting region changes significantly during the time normally used to combine measurements into a static image. Propagation through ionized material between the Galactic center and Earth also scatters its radio emission, requiring explicit treatment.

These observations test models of strong gravity and accretion through comparisons between data and simulated emission. Gravitational lensing produces characteristic image structures, but the observed brightness also depends on electron heating, magnetic fields, viewing angle, and the dynamics of the surrounding plasma. Consequently, agreement with a predicted ring scale is not equivalent to a unique determination of every black-hole property. Polarimetric observations add constraints on magnetic organization near the hole. Greater bandwidth, additional stations, observations at higher frequencies, and time-dependent reconstruction address different limitations: sensitivity, spatial-frequency coverage, atmospheric coherence, and intrinsic source variability.

Precision astrometry, masers, and astronomical distances

VLBI can measure relative source positions much more precisely than the nominal width of its synthesized beam. For a compact source with a high signal-to-noise ratio, statistical positional precision may be a small fraction of the angular resolution, although atmospheric and instrumental systematics eventually dominate. In phase referencing, an observation alternates between a target and a nearby reference source, or observes them simultaneously when the hardware permits. Solutions derived from the reference preserve the target’s relative phase and position. This differs fundamentally from imaging that uses unrestricted self-calibration and discards absolute positional information.

Repeated observations through the year measure annual parallax, the apparent displacement produced by Earth’s orbit. The distance relation is

d  (pc)=1p  (arcsec).d\;(\mathrm{pc})=\frac{1}{p\;(\mathrm{arcsec})}.

An object at one kiloparsec has a parallax of one milliarcsecond. Carefully designed VLBI experiments can reach positional precision of tens of microarcseconds or better under favorable conditions, extending direct geometric distance measurements to important regions of the Galaxy. The analysis fits parallax together with proper motion and, when needed, orbital or acceleration terms. Errors in the calibrator’s apparent position or changing source structure can otherwise imitate part of the desired signal.

Astrophysical masers are particularly useful because they produce bright, compact emission in narrow spectral lines. Water masers near 22 gigahertz and methanol masers near 6.7 and 12.2 gigahertz trace regions associated with star formation; silicon monoxide masers often probe material around evolved stars. VLBI can measure the positions and velocities of individual maser spots. Surveys of their parallaxes and proper motions constrain the Milky Way’s spiral structure, rotation, and star-forming regions. Since a maser spot can evolve or move within its local environment, internal motions must be separated from the bulk motion of the astronomical system.

An especially important extragalactic example is the water-maser disk in NGC 4258. VLBI maps resolved maser emission in a thin, nearly edge-on disk orbiting a central compact mass. Combining angular positions, Doppler velocities, and measured line-of-sight accelerations permits a geometric distance determination rather than one based on an assumed source luminosity. Related measurements in other galaxies contribute to independent estimates of the cosmic expansion rate. Radio-emitting pulsars and nearby stars provide further astrometric targets, while spacecraft observations use closely related methods to determine angular positions relative to distant radio reference sources.

Earth orientation, reference frames, and geodynamics

Geodetic VLBI uses observations of distant compact sources, especially quasars, to measure the geometry and rotation of Earth. Its principal observable is often the group delay, obtained from the variation of interferometric phase across frequency. Observing many sources in many sky directions allows a global solution for station coordinates, clock behavior, atmospheric delays, and Earth-orientation parameters. Because extragalactic sources are extremely distant, their ordinary parallax and proper motion are negligible for many reference-frame applications. Their radio structures nevertheless evolve, so they are not perfectly fixed point markers.

VLBI has a distinctive role in measuring UT1, the time scale associated with Earth’s rotation angle. It also determines polar motion and celestial-pole offsets associated with precession and nutation. These parameters connect a terrestrial coordinate system to a celestial one and are essential for precise navigation, satellite operations, and astronomical pointing. Short observing sessions can provide rapid UT1 estimates, while longer global sessions constrain a broader set of parameters. The technique directly relates the rotating Earth to distant celestial directions rather than relying solely on observations of objects orbiting Earth.

The International Celestial Reference Frame is realized through measured positions of extragalactic radio sources. Its third realization, ICRF3, adopted in 2018, contains positions for 4,536 sources in its principal S/X-band catalog, with additional catalogs at higher frequencies. The frame’s quality depends not simply on the number of sources but on their sky distribution, positional stability, observing history, and structural compactness. Comparing radio positions with optical reference frames reveals effects such as wavelength-dependent source structure and offsets between radio jets and optical emission. Such comparisons require care because a source’s measured center can depend on both wavelength and angular resolution.

Long time series of station positions reveal plate tectonic motion, deformation following earthquakes, and loading effects associated with redistribution of atmospheric and water mass. VLBI contributes to the International Terrestrial Reference Frame alongside satellite laser ranging, global navigation satellite systems, and Doppler orbitography. These techniques have complementary sensitivities and systematic errors. The VLBI Global Observing System, or VGOS, uses fast-slewing antennas and broadband observations to improve precision and observe many directions rapidly. Faster sky coverage helps distinguish atmospheric fluctuations from geometric parameters, while accurate instrumental delay calibration and stable local surveys remain necessary to connect each antenna’s reference point to the surrounding geodetic infrastructure.