Far From Radio Interference, the Square Kilometre Array Takes Root in South Africa and the Australian Outback
The telescope’s first phase, SKA1, blazes the path to radio astronomy’s future discovery machine
Standing Tall: Dish antennas stand out against the sky at the Murchison Radio-astronomy Observatory, in Western Australia. They’re part of the Australian Square Kilometre Array Pathfinder telescope (ASKAP), which is equipped with special “phased array feeds”—sets of 188 individual receivers that pick up radio signals reflected off the dishes, giving the telescope a wide field of view.Photo: CSIRO
Even in early winter, the sun is harsh in Western Australia’s Murchison shire. In this land of unpaved roads, kangaroo tracks, and low, scrubby vegetation, visitors can and sometimes do get lost. Nevertheless, here I am, a few hundred kilometers from the coast, standing on rusty red dirt, hiding under my sun hat. I am visiting a future site of one of the most ambitious telescopes ever conceived.
With just a hundred or so residents in an area bigger than the Netherlands, this piece of the Australian outback is something precious in a world swamped by wireless signals: an island of unusual calm, a clear window onto the cosmos. Back in the cool of our four-wheel-drive vehicle, one of my guides, Antony Schinckel, is emphatic about the location’s merits. “We really found this one of the best areas on the planet,” he says.
Schinckel, a telescope director with the Australian government’s Commonwealth Scientific and Industrial Research Organisation, and his colleagues have already braved hostile conditions to turn a small portion of this vast territory into one of the world’s leading radio astronomy facilities. Heavy rains can cut deep ruts in roads to the site, making them all but impassible. At one point early on, Schinckel recalls, his car suffered four flat tires in the same day as it ran over one acacia tree stump after another.
But the hard work is now starting to pay off. Over the last eight years, astronomers and engineers have transformed land where cattle once grazed into a kind of astronomical garden: the Murchison Radio-astronomy Observatory (MRO). Dozens of gleaming-white 12-meter-wide radio dishes, tailor-made for cataloging galaxies, now dot the landscape. They’re joined by thousands of spiderlike antennas, which form a state-of-the-art array capable of picking up electromagnetic waves dating back almost to the start of the universe.
These different kinds of antennas have been used to create two telescopes at MRO that are stretching the capabilities of radio astronomy. The telescopes are also the prelude to a much more ambitious project: the Square Kilometre Array (SKA). Already more than 25 years in the making, the SKA promises to be a radio telescope of immense sensitivity, by virtue of a collecting area equivalent to more than (you guessed it) a square kilometer. When the project is complete, sometime in the early 2030s, it could encompass more than two thousand dishes in Africa and half a million or so antennas in Western Australia, dwarfing the telescopes at the MRO and other such facilities. In the process, the SKA—a collaboration among 10 member countries involving more than 500 engineers—will test the limits not only of telescope design but also of data processing pipelines, international coordination, and the infrastructure of big-science projects.
“Nobody’s ever built anything on the scale we’re attempting,” acknowledges SKA director-general Philip Diamond. But he and many other astronomers think the effort will be well worth it. SKA’s sensitivity, resolution, and ability to scan large areas of the sky quickly will let it probe some of the universe’s most pressing mysteries. By cataloging vast numbers of galaxies through their hydrogen emissions, for example, the SKA is expected to help pin down the identity of dark energy, which is driving the universe to expand at an accelerating rate. The telescope will also be able to measure an unprecedented number of pulsars—spinning stellar remnants that beam electromagnetic radiation out along their magnetic poles. When these cosmic beacons wind up in tight orbits around black holes, they can be used to hunt for evidence of new physics that might finally allow physicists to develop a unified theory of quantum mechanics and gravity.
Given its staggering scale, the SKA is proceeding in stages—beginning with a smaller incarnation called SKA1. Although just a fraction of the size of the SKA, this first iteration will still be the largest radio telescope in the world, Diamond says. Part of it will be built in South Africa and the other portion here in Western Australia, and the two sites will operate—as the full SKA will—as two independent telescopes. The South African component of SKA1 (known as SKA1-mid) will encompass 197 radio dishes with diameters of 13.5 and 15 meters. Data from those dishes will be combined to study a range of targets, including pulsars and radio emissions from hydrogen that sits relatively close to our own Milky Way galaxy.
Australia’s part, known as SKA1-low, aims to pick up lower-frequency radio waves, including ones that originated from a time, billions of years ago, when astronomical objects like stars first lit up the universe. To receive these waves, the telescope won’t use dishes. Instead, it will use many simple, fixed antennas designed to pick up signals over a very wide range of frequencies, including ones in the TV and FM bands that happen to coincide with the frequencies of some of the universe’s oldest light. To yoke those antennas together into a single powerful telescope will require state-of-the-art amplification and signal processing (more on that later).
At the MRO, astronomers are already hard at work testing prototype SKA antennas. A patch of antennas was incomplete during my visit in June, but it already looked crowded. Dozens of the spindly 2-meter-tall structures, which resemble little fir trees, were packed together in a messy steel miniforest.
By 2024, the SKA team expects to install more than 131,000 of these treelike antennas, grouped in clusters and extending into the desert for tens of kilometers along three spiral arms. The result won’t be much more photogenic than the test patch was. But if all goes well, the array could produce truly stunning results: the first detailed images of a universe as it was transforming from a murky sea of neutral hydrogen into something we’d recognize today—a black ocean of space studded with shining stars and galaxies.
Low-Frequency Forest: Australia’s part of the Square Kilometre Array radio telescope, known as SKA1-low, will consist of many fixed triangular-shaped antennas. Prototypes of the antennas, seen here, are now being tested at the Murchison Radio-astronomy Observatory. Seeing Back in Time: Technicians connect the prototype fixed antennas being tested at the Murchison site. The system funnels signals from as many as 256 antennas into one optical fiber, for transmission to the next processing stage. Eventually, the site will have more than 100,000 of the low-frequency antennas, which will collectively peer back at a time when the universe was a mere billion years old.Photos: ICRAR/Curtin University
The icons of radio astronomy are its dishes. New Mexico’s Very Large Array, for example, with its orderly lines of 25-meter-diameter dishes, has shown up in several motion pictures, most notably 1997’s Contact.
The dishes of the Very Large Array work much like an optical telescope does, by focusing incoming radio waves onto receptors. But the radio band is wide, and a telescope design that works well in one swath of frequencies isn’t necessarily the best choice for another. Basic physics dictates that the longer the wavelength to be picked up, the bigger the dish needed to maintain the same resolution. The upshot is that beyond a certain wavelength, common sense suggests a move to antennas that can directly receive the radio waves.
The idea of doing radio astronomy with such antennas is not new. In the early 1930s, the technology enabled Karl Jansky to make the first detection of radio waves from beyond the solar system. Pulsars were discovered serendipitously in 1967, when their clockwork-like signals were detected using an array of dipole antennas outside Cambridge, in the United Kingdom.
But at some point, says astronomer Randall Wayth, the longer wavelengths—the sort that are ideal for such fixed antenna arrays—fell out of vogue in radio astronomy. More recently, renewed interest in that part of the radio band is being spurred by astronomers’ desire to peer far back into the universe’s past. And, conveniently, they can now lean on a range of advances in digital electronics, signal processing, and computing to create a new generation of arrays.
“It’s definitely a renaissance,” says Wayth, an associate professor at Curtin University, in Perth, and a senior research fellow at Western Australia’s International Centre for Radio Astronomy Research. The general approach bears more than a passing resemblance to phased-array radar systems and to the antenna arrays being developed for 5G cellular networks.
Wayth directs one of the telescopes at the leading edge of this revival: the Murchison Widefield Array, or MWA. As one of the official “precursor” telescopes for SKA, the MWA is helping to work out the kinks in combining many passive antennas into a single state-of-the-art telescope.
As with SKA1-low, the Murchison Widefield Array’s antennas are designed to pick up radio waves at the lower end of frequencies used for radio astronomy. SKA1-low’s design calls for antennas that are sensitive from 50 to 350 megahertz. MWA’s antennas detect signals in a somewhat narrower range, from 80 to 300 MHz. In contrast with SKA1-low’s fir-tree antennas, those of the MWA call to mind sunbathing knee-high spiders. They’re on metal grids designed to reflect incoming radio waves back up to them.
Although their antennas look different, the Murchison Widefield Array and SKA1-low share the same basic approach as well as a big scientific ambition: gazing into a still-murky period in the early universe called the Epoch of Reionization. The name refers to a time, roughly 13 billion years ago and about a billion years after the big bang, when early stars and other objects heated neutral hydrogen atoms enough to knock their electrons off, transforming the cosmos from an opaque sea of neutral hydrogen into the transparent universe we see today. Remarkably, it’s still possible to detect radio waves emitted by those neutral hydrogen atoms. The waves were emitted with a wavelength of 21 centimeters, but by the time they reach Earth, billions of years of cosmic expansion will have stretched them to a couple of meters.
The Murchison Widefield Array is racing to be the first telescope to detect those elongated echoes of the far-distant past. Astronomers hope the study of this radiation will help reveal more about how reionization altered and shaped the early universe—for example, how structures like galaxies formed and changed in that pivotal epoch. “It’s one of the major phases during the evolution of the universe which is completely unknown,” says Benedetta Ciardi, an astrophysicist at the Max Planck Institute for Astrophysics in Garching, Germany, and a staffer at one of MWA’s competitors, the LOFAR telescope, based in the Netherlands.
Spider Town: Randall Wayth, an associate professor at Curtin University and director of the Murchison Widefield Array, crouches next to Tile 107 of the telescope (above). An aerial photo of the MWA (bottom) shows more of the telescope’s tiles, each of which contains 16 of the spiderlike antennas.Photos: Top: Ben Scandrett/Department of Industry; Bottom: ICRAR/Curtin University
To hunt for signals from this epoch—or to perform any of its observations, actually—the Murchison Widefield Array sops up radio waves from many directions at once. Incoming signals are boosted at the center of each spidery antenna by a pair of low-noise amplifiers and then sent to a nearby “beamformer.” There, waveguides of various lengths, printed on circuit boards, impart delays to the antenna signals. With the right selection of delays, the beamformers virtually “tilt” the array, so that radio waves arriving from a particular patch of the sky all seem to reach the antennas at the same time—as they would if they were being received by a single large antenna. SKA1-low will do this entire process digitally, without the waveguides. That approach will enable it to construct multiple beams—as if the array were pointing to multiple spots in the sky simultaneously.
The MWA antennas are divided into groups. Signals from each group are sent to a single receiver that distributes the signals among various frequency channels and then sends them on to the observatory’s central building over fiber. There, a set of field-programmable gate arrays and graphics processing units correlate the data, multiplying the signals from each receiver with those of every other one and integrating over time. This number crunching is the heart of interferometry [PDF], a process that combines the signals from multiple dishes or antennas to create a single strong signal, as though it came from one telescope.
Much like a single dish, the resolution of such a virtual telescope is inversely proportional to its physical size. Bigger is, of course, better. In particular, for a virtual telescope consisting of a set of dishes or fixed antennas, the telescope’s maximum resolution is set by its longest baseline, or distance between a pair of elements. The longer that distance is, the finer the resolution.
Astronomers have used this property to construct virtual telescopes that reach across continents, enabling resolutions so fine that they have been used to home in on the area around the supermassive black hole at the center of the Milky Way. But size isn’t the only consideration. A single pair of antennas, however far apart, will give you only a small piece of information about the light emitted from an object. To construct pictures, astronomers must fill out the array. More fixed antennas or dishes yield a combinatorial explosion of different baselines, which can then be used to create a telescope-like image through a process called aperture synthesis. The imaging capabilities of such an array thus depend on several factors, including the total number of antennas, the span of the array, and the details of how the antennas are placed relative to one another.
At the Murchison Widefield Array, the output from the observatory’s servers is sent down hundreds of kilometers of fiber, first to the coastal city of Geraldton and then on for another 400 kilometers or so to a supercomputing center in Perth.
The MWA can ship more than 25 terabytes of data a day to the Perth facility. But in the coming years, that data rate will be dwarfed by the output of SKA1-low. The array’s 131,000 antennas will collectively produce upwards of a terabyte of data every second, says Keith Grainge, an astrophysicist at the University of Manchester, in England, who leads the SKA working group dedicated to signal and data movement. “It’s about an eighth of an Internet that we’ve got to transport,” Grainge says.
Once the data reaches Perth, it must be further processed in order to transform it into sky maps and other scientific products that astronomers can use. This is an exascale problem, says Andreas Wicenec, a professor at the University of Western Australia who is studying the computational needs of the project. Wicenec estimates that SKA1-low will need a supercomputer at least as fast as the current world-record holder, China’s Sunway TaihuLight. The only hitch is that this supercomputer must be significantly cheaper and consume just a fifth as much power as the Sunway TaihuLight, which can eat up 15 megawatts performing computations.
Wicenec isn’t fazed. “If we don’t get such a machine,” he says, “we will still be able to do amazing science, just not the most challenging projects initially. That’s the advantalge of radio astronomy,” he adds. “You can easily scale up and down and just do what’s currently affordable. A few years later we can then ramp up.”
The SKA’s success will depend in part on making sure Murchison’s radio window stays as clear as possible. Cellular signals, electric motors, TV transmitters, arc welding, and many other sources of RF can interfere with observations.
The site itself is protected as much as it can be from outside noise. A “radio-quiet zone” extends out to 260 km around the Murchison Radio-astronomy Observatory. Mining companies and others that want a new license to operate any sort of transmitter within the zone must first consult with the Commonwealth Scientific and Industrial Research Organisation. “The intention is to protect the site from 70 MHz to 25.25 gigahertz,” says Carol Wilson, spectrum manager for the MRO and a senior member of the IEEE. Recently, she says, a team using the Australian Square Kilometre Array Pathfinder telescope, or ASKAP, which sits not far from the Murchison Widefield Array, found a gas cloud by observing a slight dip in the signal from a galaxy that emitted its light roughly 5 billion years ago. The dip in the signal was over a range of frequencies completely covered by a cellphone band. “It would have been impossible to do that in a more populated area,” Wilson says.
The designers of the Murchison Radio-astronomy Observatory have gone to great lengths to make sure the telescopes and associated equipment themselves don’t add to the problem. The observatory’s central building—a one-story structure housing workshops, desks, racks of processors, and a maser used to distribute clock signals to the telescopes—is completely wrapped in a continuous steel sheet. Fiber, power lines, and air pass into this Faraday cage through openings that are too narrow and long for a range of radio waves to traverse. The main entrance boasts a double-doored vestibule that acts like an airlock for radio waves; anyone hoping to enter must lever one door closed before opening the other one.
Such precautions help make the site extraordinarily quiet. And the Murchison Widefield Array is taking advantage of that silence to hunt for the first indications of the Epoch of Reionization, which should show up in subtle changes in how neutral hydrogen is distributed over the sky. When SKA1-low arrives, it will be able to map this transition in greater detail, giving astronomers a glimpse of how ancient stars and galaxies brought the universe out of its dark ages and helped shape the cosmos we see today.
Important challenges still must be overcome before workers even begin building SKA1-low, which could start in 2019 and continue for five years. A team of engineers is validating the antenna design, using a combination of simulation and measurements, in preparation for a key review next year. Antennas can interfere with one another, notes MWA director Wayth, and in some cases they can cause signals to cancel out, creating blind spots. “Right now we’re making sure we fully understand the electromagnetics of how the stations work just to make sure nothing unexpected pops up,” says Wayth, who is also part of the design team for the Australian SKA array.
But Wayth says most of the technical challenges have already been worked out. At this point, the biggest issues are logistics and infrastructure, he says. “It’s in the middle of nowhere, [and] it needs to have power and communications and timing and everything distributed to it.”
In South Africa, some of the dishes will be able to tap into local grid power, but in Australia, all the energy must be produced on site. Both observatories aim to keep their carbon footprint to a minimum. Until this year, the MRO had run on diesel generators. It now has 1.6 MW worth of solar panels, and vast packs of lithium-ion batteries that can store 2.6 megawatt-hours—more than half of what SKA1-low will need when it is fully operational. Some of the more far-flung patches of the array will likely get their own solar panels.
Maintenance is another challenge. These days, the Murchison Widefield Array gets hit by lightning about once a year, which might knock out a few sets of antennas and a receiver. But MWA’s antennas are all within a few kilometers of one another. The sheer number of SKA1-low antennas—along with the fact that they will be 2 meters tall and extend tens of kilometers from the center—could tax those maintaining the array.
There is also the matter of money. At the moment, the budget for building SKA1 in South Africa and Australia is capped at about €675 million (about US $800 million, or about a billion Australian dollars), an amount set by the project’s 10 member countries: Australia, Canada, China, India, Italy, the Netherlands, New Zealand, South Africa, Sweden, and the United Kingdom. But that funding won’t cover the entire cost of SKA1 with the specifications that astronomers are hoping for. “Even though we’re looking at spending effectively a billion dollars on SKA phase 1,” says director-general Diamond, “you can’t do everything within even a billion.” He’s trying to recruit more countries into the partnership, which could boost funding.
But if all goes well, Wayth says, SKA1-low should give radio astronomers a factor-of-10 boost in telescope sensitivity and other capabilities. “It will be a sensational telescope,” he says. “My old boss once said radio astronomers don’t usually get out of bed for anything less than an order of magnitude. And in this case, there’s lots to get out of bed for.”
This article appears in the December 2017 print issue as “Engineering the World’s Biggest Radio Telescope.”