The legal shield around South Africa’s most valuable scientific instrument does not reach the objects most likely to interfere with it.
The Karoo Central Astronomy Advantage Area (KCAAA), established under the country’s Astronomy Geographic Advantage Act, protects MeerKAT and the Square Kilometre Array-Mid (SKA-Mid) telescope now under construction from ground-based interference. Satellites sit outside it.
“Satellites are physically geographically located outside the KCAAA while their transmissions enter the protected area,” the South African Radio Astronomy Observatory (Sarao) said in a written response to questions from TechCentral.
The act does allow the minister to act against spectrum users that “transmit or broadcast into” the zone, including forcing a move to another frequency. But large moving constellations were not what those rules were written for.
That is what Sarao has asked Icasa to fix.
In submissions to the communications regulator’s public hearings this month on the draft radio frequency spectrum regulations, it called for enforceable coordination between satellite operators and radio astronomy facilities, and for compatibility assessments covering aggregate interference from whole constellations rather than single satellites.
“A single satellite transmitting into the main beam can overwhelm faint cosmic signals,” Sarao said. Interference can be orders of magnitude stronger than the signal astronomers are trying to detect. The larger worry is cumulative: thousands of satellites transmitting at once, compounded by unintended electromagnetic radiation leaking from satellite electronics.
The window is now
There is no satellite headcount at which radio astronomy tips over, Sarao said, because it depends on frequencies, orbits, power and beam configurations. It pointed to ITU-R studies of the 2 690-2 700MHz band that “demonstrated potentially severe interference to radio astronomy” from constellations far smaller than those now proposed.
South Africa is not yet routinely measuring satellite interference at MeerKAT. Monitoring “has primarily focused on terrestrial services” and is being extended to space-based sources. The clearest evidence so far comes from elsewhere: SKA Obversatory-supported observations by the Lofar telescope detected unintended emissions from Starlink satellites between 110MHz and 188MHz, inside a protected radio-astronomy band.
Read: South Africa’s giant SKA telescope clears major technical hurdle
One mitigation under international discussion stops short of switching satellites off across the whole reserve. Through the International Telecommunication Union, the international community is “engaging with operators to develop technical and regulatory mitigation measures,”, said Adrian Tiplady, Sarao deputy MD for strategy and partnerships. “One of the options is to create nulls (holes) in their coverage around radio telescopes.” Those holes would be far smaller than the protected area, focused on the dishes rather than the Karoo as a whole.
The window is now, Sarao said, because satellites have long operational lives and their numbers are climbing. “It is much easier to establish effective coordination and mitigation measures as the satellite environment is being developed than to try to retrofit them after millions of devices are already in orbit.”
Sutherland faces the optical version. A recent European Southern Observatory study in Astronomy & Astrophysics, by ESO astronomer Olivier Hainaut, models the effect of the more than 1.7 million satellites covered by announced or proposed low-Earth orbit constellation filings. Roughly 14 000 satellites are in orbit today, or about 32 000 counting dead craft and debris. Most of the 1.7 million comes from a single February filing by SpaceX for a million satellites to host AI data centres.
Nicolas Erasmus, instrumentation scientist and astronomer at the South African Astronomical Observatory, was careful about the headline numbers when asked about them by TechCentral. The 1.7 million refers to systems announced or proposed, not approved and certain to fly. The much-quoted finding that up to 28% of a telescope’s field of view could be lost is, he said, “the peak modelled fraction of pixels affected in a particular 300-second VLT/Fors2 (Very Large Telescope, in Chile) exposure scenario, not 28% of all observations or observing time, and not a measurement at Sutherland”.
“If even a substantial fraction were deployed, satellites would cease to be an occasional nuisance and become part of the normal observing environment,” he warned.
What the Sutherland data shows
The picture on the ground varies sharply by instrument. MeerLicht, the wide-field telescope that observes in tandem with MeerKAT, is the most exposed. It images about 1.7 by 1.7 degrees of sky at once, an area that would hold a three-by-three grid of full moons. A peer-reviewed study that ran a streak-detection system over roughly 200 000 MeerLicht images found 9 107 satellite or debris trails. A trail in a frame is not a lost exposure, Erasmus said: affected pixels can often be masked, or several exposures combined to remove the streak.
At the other end, a UCT and SAAO student, Enzo Afonso, analysed about 30 000 acquisition images from the robotic 1m Lesedi telescope and found 90 frames, about 0.3%, carrying a satellite streak. “So currently it doesn’t seem that detrimental to daily operations,” Erasmus said, “but with the exponential increase in numbers of proposed constellations this might eventually become a serious issue.” Afonso’s analysis has not yet been written up.
The Southern African Large Telescope sits in between. Its science field of view is about eight arcminutes across, under 1% of MeerLicht’s by area, and much of its work is spectroscopy, so its odds of catching a satellite are low. But as a 10m-class telescope its far greater light-collecting area makes even CubeSat-sized objects detectable, and a satellite crossing a spectrograph slit can contaminate a spectrum without leaving an obvious streak.
Mirrors
The proposal that worries Erasmus most is not a conventional satellite. Reflect Orbital has floated a constellation of 50 000 mirror satellites by 2035, designed to bounce sunlight onto ground targets such as solar farms. The US Federal Communications Commission approved a single demonstration satellite, Earendil-1, in July, carrying an 18m by 18m reflector – and ruled that the light pollution and telescope interference objections raised by astronomers fell outside its authority to consider.
If any target zones sat near Sutherland the effect “would be particularly bad”, Erasmus said, because the site is rated among the three darkest observatories in the world. Space mirrors used near it “would ruin a first-class dark site”.
South Africa’s Sutherland astronomy regulations control artificial outdoor lighting inside a declared Astronomy Advantage Area, but they “help greatly, mainly against activities on the ground”, Erasmus said. “Those regulations cannot prevent a satellite licensed in another country from passing overhead.”
…article continues below…
The most effective mitigation is at source: licensing authorities requiring cumulative impact assessments, brightness limits and accurate public orbital data. The ESO study recommends the total satellite population stay below about 100 000, all faint enough to be invisible to the naked eye. Operators have tried. SpaceX darkened an experimental Starlink, DarkSat, and later added sun visors, but a peer-reviewed study found the coating roughly halved its brightness, not enough for large telescopes.
None of this is an argument against satellite internet, which is extending broadband to places fibre and cellular towers never reached. The ESO researchers stressed that those benefits should not be minimised, and that the task is balancing them against scientific discovery. – © 2026 NewsCentral Media
