How separated radio dishes act like one giant telescope

When the Event Horizon Telescope released its image of the environment around the black hole in M87 in 2019, no single camera had taken that picture. Radio observatories in different parts of the world had recorded signals from the same target. Their recordings were brought together, compared and turned into an image. MIT Haystack describes the observations and processing.

How can instruments separated by kilometres, or even continents, behave like one telescope? They do not photograph separate pieces of the object and stitch the pieces together. They measure how the same arriving radio waves differ from place to place. Those differences carry information about where the radiation came from.

The same wave arrives at different times

A radio wave is an electromagnetic wave, like visible light but with a longer distance between neighbouring peaks. That distance is its wavelength. A wavefront joins places at the same stage of the wave's oscillation. For a sufficiently distant source, a small part of an arriving wavefront can be treated as a flat sheet crossing the array.

If the sheet approaches two dishes obliquely, it reaches one before the other. Change the source's direction and the arrival-time difference changes. The line joining the antennas is called a baseline. Its length and orientation together determine which directional information the pair can measure.

Imagine two microphones recording the same distant clap. Sliding one recording in time can align the shared sound. Radio interferometry also compares a shared signal, but the astronomical signal usually resembles random noise, not an isolated clap. The instruments must preserve its rapid fluctuations, including their relative timing. The analogy stops there: the array uses electromagnetic signals, specialised receivers and phase-sensitive calculations rather than human-scale sound. NRAO introduces the arrival-time principle.

Preserve the wave, not just its brightness

Each dish concentrates radiation onto a receiver. In a common receiver design, electronics mix the signal with a stable reference called a local oscillator, shifting it into a lower frequency range that later stages can handle. Digitisers then sample it, producing streams of numbers. The frequency conversion must preserve the relationships needed for comparison; measuring only each dish's average brightness would discard crucial information. NRAO explains this receiver chain.

Connected arrays can distribute reference signals over fibre. ALMA's system includes corrections for changes in the reference path. NRAO describes the photonic reference. Widely separated observatories instead use very stable local frequency standards and timestamp their recordings. In very-long-baseline interferometry, hydrogen-maser atomic clocks help keep the recordings comparable. The Event Horizon Telescope explains this timing requirement.

Before comparing a pair, the processing compensates for the predicted arrival-time difference. In the simplest picture, it delays the earlier stream until corresponding parts meet. Real systems also correct fractional-sample timing and phase, updating the corrections as Earth turns. The alignment is calculated for a chosen reference direction, not every point in the sky simultaneously. NRAO's technical notes explain delay and phase correction.

What the comparison actually measures

A specialised processor called a correlator multiplies corresponding signal values and averages the products. A shared astronomical contribution can remain in that average, while unrelated receiver noise tends towards zero. Noise does not disappear completely: a finite observation still has uncertainty.

Within a narrow frequency channel, phase describes where a wave lies in its cycle. A change in relative phase changes the correlation. As source direction or baseline orientation changes, the response passes through a repeating pattern called fringes. Radio arrays measure this electronically; they do not need visible stripes between the dishes.

The pair's result, called a visibility, records an amplitude and a phase. It is one measurement of the source's structure, not one image pixel. NRAO derives the two-antenna measurement.

Each baseline asks a different question

Picture describing an image using stripes: broad stripes for large-scale changes, fine stripes for small-scale changes, and different stripe directions for different orientations. This is a mathematical description, not a claim that real astronomical objects are striped.

A baseline measures a particular spacing and direction of these patterns. The quantity is called a spatial frequency: how rapidly brightness changes across angle on the sky. Longer projected baselines measure finer angular patterns; shorter ones supply information about broader structure. “Projected” means the separation as viewed from the source, rather than simply the distance measured along the ground. ALMA explains baselines and spatial frequencies.

Many antenna pairs therefore provide many different constraints. Earth's rotation changes their projected orientations and lengths while they track a source, adding measurements without requiring the dishes to move across the ground. Combining such observations is Earth-rotation aperture synthesis. It assumes the source can be treated as sufficiently steady over the measurements being combined. ESO explains why baseline direction and Earth's rotation matter.

A sharper view is not a bigger collecting surface

The characteristic angular scale becomes smaller when the baseline grows or the wavelength shrinks. In the WonderLab, an illustrative wavelength of 30 millimetres and projected baseline of 1 kilometre give about 6.19 arcseconds. One arcsecond is 1/3,600 of a degree. Doubling that baseline halves the scale to about 3.09 arcseconds.

Spreading the dishes apart does not add collecting surface between them. Collecting area comes from the antennas themselves; sensitivity also depends on receiver noise, bandwidth and observing time. An array can approach a huge instrument's resolving scale without matching the sensitivity or image response of a filled dish that large. ESO explains the missing collecting surface.

The image still has to be earned

The atmosphere and electronics can alter signal strength and phase. Astronomers observe reference sources and solve for these changes before interpreting the target. Calibration must track effects quickly enough that the corrections remain useful. NRAO describes the calibration work.

The calibrated measurements are then transformed into an image and processed to reduce artefacts from incomplete sampling. Because many spatial frequencies were never measured, reconstruction is not unique. Nor can long baselines replace missing short ones: broad, smooth emission can be poorly recovered without compact-array or single-dish measurements. ALMA explains reconstruction and missing short spacings.

The giant telescope is therefore real as a coordinated measurement system, not as an invisible metal dish. Its power comes from preserving relationships: between arrival times, between antenna pairs, and between the measured patterns that constrain a picture of the sky.