Very Long Baseline Interferometry works on the principle that radio waves arriving at two widely separated antennas from the same source will arrive at slightly different times — the exact time difference depends on the source's position and the distance between the antennas. By recording the precise arrival time and waveform at each antenna simultaneously, scientists can later cross-correlate the recordings and extract information about the angular structure of the source.
Step 1: Simultaneous observation with atomic clock precision
All participating observatories point at the same target simultaneously during coordinated observation windows, typically 10 days each April when both M87* and Sgr A* are accessible from most sites. Each antenna records incoming 1.3mm radio waves at extremely high data rates — up to 64 Gbps in recent campaigns — storing data on arrays of hard drives. The recordings must be timestamped to within one ten-trillionth of a second using hydrogen maser atomic clocks at each site, among the most precise timekeeping instruments available.
Step 2: Petabyte data shipping ("sneakernet")
After observations, the accumulated petabytes of hard drives from each site are physically shipped to correlation centres at MIT Haystack Observatory (Massachusetts) and the Max Planck Institut für Radioastronomie (Germany). The South Pole Telescope data cannot be shipped until the Antarctic winter ends in October — the station is physically inaccessible for months. At the correlators, supercomputers cross-correlate the signals from every possible pairing of telescopes (351 unique pairs from 27 stations), extracting the interference patterns that encode the sky structure.
Step 3: Image reconstruction from "visibilities"
The raw correlation products — called "visibilities" — are complex numbers representing the amplitude and phase of interference between each baseline pair. Converting these into an actual image requires computational algorithms: CLEAN-family algorithms developed over decades of radio astronomy, and more recently regularised maximum likelihood (RML) imaging and machine learning approaches. Multiple independent imaging teams use different algorithms and compare results — agreement between methods is required before results are published, providing a rigorous check against reconstruction artefacts. The final images represent months of computational work and scientific cross-checking.
Why 1.3mm (230 GHz) wavelength?
The EHT observes at 1.3mm for two reasons: (1) interstellar scattering — the smearing of radio waves by turbulent ionised gas between us and the galactic centre — is less severe at shorter wavelengths, allowing the black hole shadow to be resolved; (2) shorter wavelengths cannot penetrate Earth's atmosphere without severe water vapour absorption. At 1.3mm, observations are possible from high-altitude, dry sites. This is why all EHT sites are at high elevation and/or in desert regions. A planned expansion to 0.87mm (345 GHz) will further improve resolution.
