The Event Horizon Telescope seems to include these radio telescopes currently:

Image by European Southern Observatory (ESO)/O. Furtak; CC-BY 4.0-licensed, see the source on Wikipedia Commons and the original image link.

See also the map and list on their official website.

It is also said, that it has a “effective diameter of the entire planet”. But it's striking, it does not include half-of the earth… It has no telescopes in Africa, Asia or Australia.


  • Why were not they included?
  • Would it increase the diameter if they would include some from there? If so, again why did not they? (😉)
  • Or, similarly asked: How can it be the diameter of the earth if only half-the-earth is included in the network?

Please try to explain it in a simple way first. I am no astronomer.

What I've noticed so far when skimming their official pages is that they write here:

As more telescopes are added to the EHT, we will be able to produce images of the emission around black holes. In general, the fidelity of images produced by an interferometric array increases as additional telescopes are added to the array.

This seems to support my question of why they did not use possibilities in Asia or so… I guess, it would also make sense to have a more or less uniform distribution across the planet (or does not this matter?), that's why a "dot" in Africa might make sense too…

  • $\begingroup$ In addition to the answer covering Asia and Australia, the case of Africa is simply that no such facilities exist there yet. One of the EHT announcments mentioned though, that they plan on expanding by bringing a station in Africa up into the network. $\endgroup$ Commented Apr 13, 2019 at 21:00
  • 1
    $\begingroup$ I think Rob's answer is much more detailed than mine, and may be a better one to accept because it addresses your question much more specifically than I could. Of course its 100% your decision. $\endgroup$
    – uhoh
    Commented Apr 16, 2019 at 6:56

3 Answers 3


Would it increase the diameter if they would include some from there?

No. Not by much, at least. The telescopes are already ~20,000 km apart, so you can't create a longer baseline that still has a simultaneous view of the target.

Don't forget: Earth is a sphere. Only one half of that sphere can observe M87 at the same time.

Telescopes in the Eastern hemisphere would enable more observations, but I don't know if that would bring any improvement over what they already had.

  • 1
    $\begingroup$ More telescopes would bring significant improvement, even if they were in the other hemisphere. The answer to the OP's question is the suitability of the site's equipment and altitude above atmospheric water. $\endgroup$
    – uhoh
    Commented Apr 14, 2019 at 0:13
  • 1
    $\begingroup$ Part of the question was "Would it increase the diameter if they would include some from there? " and "How can it be the diameter of the earth if only half-the-earth is included in the network" . The answer to that is No. Before you can launch into great big detailed answers, you have to answer the basics. $\endgroup$
    – Hobbes
    Commented Apr 15, 2019 at 6:43

tl;dr: @Hobbes' answer is demonstrably wrong; the EHT takes a large fraction of its data when the target is not visible from one of the extreme sites.

If there were sites distributed all the way around the earth, it would be tremendously advantageous to collect from all of them, and the image resolution would improve with the addition of longer duration data and more diversity in baseline vectors.

As pointed out by @AtmosphericPrisonEscape: The main requirements for a site to be used in the Event Horizon Telescope include:

  1. Low noise, stable receiving at 230 GHz (1.30 millimeters wavelength)
  2. High altitutde to minimize attenuation and dispersion at 230 GHz due to atmospheric water
  3. digitization of several 2GHz wide channels near 230 GHz for a total digitization rate of 32 Gigabits/second. Not too many sites are equipped to do all of this! See this answer.
  4. infrastructure to calibrate, monitor atmospheric water, record and store huge amounts of data, monitor GPS time signals and support on-site atomic clock of sufficient precision for time stamping of these 1.3 millimeter wavelength signals.

Not so many sites currently can support all of these requirements simultaneously and sufficiently. But this may grow in the future.

At 1.30 millimeters wavelength (230 GHz), a huge dish antenna must maintain a parabolic shape to an accuracy of order of hundreds of microns while the dish tilts up and down during observations. Dishes with surface accuracy like that are few and far between.

There is a further consideration. The EHT needs to be a "movie camera" because the swirling accretion disk around the black hole (the thing actually images) is constantly changing. For the M87 object things change more slowly than for the black hole at the center of our Milky Way (days versus minutes) you would really want to have around-the-world telescope coverage to produce around-the-clock observations.

The results are described in four main papers (there are/will be more):

While the simple view is that the resolution is determined by the farthest baseline that the source is visible from simultaneously, the situation is more nuanced because with so few sites we really can't call this an aperture in a conventional sense. So you can still use a subset of all possible baselines to "fill-in" missing information and improve the image reconstruction.

You can see between 04h and 06h UTC the eastern-most site (PV; 30m dish in (Spain)) cuts out and the western-most sites (JCMT (Hawaii), SMA (Hawaii)) come on line.

A brief reading suggests that they stop using the data from a given site when the object drops below 20 degrees elevation above the horizon at that site.

In the second image below, you can see all sites used. The solid lines indicate baselines used to produce the image, the dashed lines indicate baselines used to collect data for calibration of the EHT network.

enter image description here

Figure 2. (of Paper III). EHT 2017 observing schedules for M87 and 3C 279 covering the four days of observations. Empty rectangles represent scans that were scheduled, but were not observed successfully due to weather, insufficient sensitivity, or technical issues. The filled rectangles represent scans corresponding to detections available in the final data set. Scan duration varies between 3 and 7 minutes, as reflected by the width of each rectangle.

enter image description here

Figure 1. (of Paper I). Eight stations of the EHT 2017 campaign over six geographic locations as viewed from the equatorial plane. Solid baselines represent mutual visibility on M87* (+12° declination). The dashed baselines were used for the calibration source 3C279 (see Papers III and IV).


Why does the Event Horizon Telescope (EHT) not include telescopes from Africa, Asia or Australia? Why were not they included?

Africa doesn't have a radio telescope in the frequency range necessary (230-450GHz) to participate in the EHT array. For Asia Wikipedia lists the "Yevpatoria RT-70 radio telescope" as capable up to 300 GHz and located in western Crimea. For Australia Wikipedia lists the "Parkes Observatory" whose frequency is also too low, with a maximum of 26 GHz. In Japan there is the "Nobeyama Millimeter Array" but it only goes up to 230 GHz.

The only radio telescopes that go up to the required frequency range (according to the Wikipedia webpage, which may not be a complete list) are:

As you can see the list is certainly incomplete. Another list of the world's radio telescopes is the website TheSkyIsNotTheLimit.org which offers this graphic:

Countries with Radio Telescopes

If you center Google Earth on the Atacama Pathfinder Experiment (APEX), the center of the EHT array, you'll notice that those other radio telescopes aren't included in the view:

Earth - Centered on Atacama Pathfinder Experiment

Would it increase the diameter if they would include some from there? If so, again why did not they?

The west coast of Africa could assist if they had a modern radio telescope on a mountain top, but they don't. Even so, such a radio telescope would be pointing sideways, through the thickness of the atmosphere over the ocean; radio telescopes work best pointing upwards and you want the ability to pan, not to be constricted to a limited range of movement. Such distant points in the array will be simultaneously active only for a short period of time, but they serve as a hand-off as one rotates into position and the other rotates out of view.

If you rotate the globe a bit you lose Hawaii but but gain western Europe (including Crimea, if they upgrade) and the west coast of Africa, including Gamsberg Mountain (in close proximity to the High Energy Stereoscopic System (H.E.S.S.)) in Namibia:

Earth - Rotated to include Gamsberg Mountain, Namibia Africa

A reason to be interested in Gamsberg is because the mountain is owned by the "Internationale Amateursternwarte" referred to as the IAS. It was established in Germany March 1999 by a group of dedicated amateur astronomers and supported by the Max Planck Institute for Astronomy. This image tells the tale:

Gamsberg Mountain Namibia Africa

There are discussions underway with respect to Gamsberg, see : "The African Millimetre Telescope" (June 2017), by Michael Backes, Cornelia Müller, John E. Conway and Roger Deane, on page 1:

"Although the EHT already constitutes an impressive VLBI network of mm-wave radio telescopes, their spacial distribution is clustered around the Americas, thus limiting its capabilities. Significant improvement can be achieved by adding a single mm-wave radio telescope in Africa, nominally called the Africa Millimetre Telescope.

3. The Africa Millimetre Telescope

Adding the Africa Millimetre Telescope (AMT)$^{12}$, a single mm-wave radio telescope on the African continent, to the EHT network will significantly increase the coverage in the $u$-$v$-plane (see fig.2). This will greatly improve the imaging capabilities of the EHT and, hence, advance its capabilities to directly image the ‘shadow’ of Sgr A∗. In particular, it will enable the ‘Eastern sub-array’, including the IRAM 30 m telescope, NOEMA, SPT, ALMA, and AMT to perform imaging observations, thus adding a significant amount of daily observing time of Sgr A∗ to the EHT (see fig. 1 on the right). As shown in fig. 1, the AMT will have common baselines for observations of Sgr A∗ with all the high-sensitivity telescopes (the IRAM 30 m telescope on Pico Veleta, NOEMA, ALMA, and the LMT) as well as with all the other EHT telescopes. The $u$$v$-plane coverage of the current EHT configuration is presented in figure 2; included is the improvement in coverage by adding the AMT.

Gamsberg Mountain - Namibia Africa

Figure 1: Baselines of the current EHT VLBI network (in yellow) and additional baselines provided by the AMT (in red). Note: the Combined Array for Research in Millimeter-wave Astronomy (CARMA) ceased observations in 2015, IRAM PV denotes the IRAM 30 m telescope on Pico del Veleta, Plateau de Bure denotes NOEMA, and Gamsberg denotes the AMT.


Potential sites on the African continent where Sgr A∗ can be observed at elevation angles ≥ 40° and that are at sufficiently high altitude to ensure an average precipitable water vapour column of less than 6 mm include sites on Mt. Kilimanjaro (≥ 4,300 m a.s.l.) in Tanzania and the Sani Pass in the Drakensberg mountains in Lesotho (≥ 3,050 m a.s.l.). However, primarily because of temporal overlap in visibility of Sgr A∗ with the mm-wave telescopes in the Americas, the western-most site, Mt. Gamsberg (2,347 m a.s.l.) in Namibia has been chosen as the primary site for further investigations. Additional supporting arguments for this site are that the land is owned by the Max-Planck Society, and that the government of Namibia is very encouraging of astronomy development.

The addition of Gamsberg will significantly improve the coverage, but construction hasn't been approved yet.

In general, the fidelity of images produced by an interferometric array increases as additional telescopes are added to the array.

Yes, but while adding one anywhere does something adding one or more in perfect locations maximizes the return on the enormous investment.

In "Synthesis Imaging in Radio Astronomy II", A Collection of Lectures from the Sixth NRAO/NMIMT Synthesis Imaging Summer School. Edited by G. B. Taylor, C. L. Carilli, and R. A. Perley. ASP Conference Series, Vol. 180, 1999 (.PDF - Warning: 43 MB):

This is a collection of papers, on page 537:

27. Interferometric Array Design

M.A. Holdaway & Tamara T. Helfer
National Radio Astronomy Observatory, Tucson, AZ 85721, U.S.A.

Abstract. We investigate some of the principles which lead to the design of radio interferometric arrays and array configurations, including both abstract issues such as sensitivity and Fourier plane coverage, and practical issues such as moving antennas and site topographical constraints. We draw on the design and history of existing arrays and also give a glimpse of what ideas and algorithms are helping design new instruments such as the Submillimeter Array (SMA) and the Millimeter Array (MMA).

1. Introduction

Array design can include a variety of topics: how many antennas should the telescope have, and how big should they be? Are there astronomical requirements which dictate an aspect of the array layout? How many antenna configurations will there be, and how will the different configurations work together? How should we design each individual configuration? But the central topic of array design deals with how to efficiently sample the Fourier plane. Each interferometer, or pair of antennas, at a given moment in time samples a single point in the Fourier plane, and we need to arrange the antennas in such a way that the set of sampled points permits us to make high quality, high sensitivity images. Since most antennas require a fair amount of infrastructure with not insubstantial capital costs on the ground beneath their bases (called antenna pads), it is important to design a good set of antenna configurations which adequately sample the Fourier plane before the array is built.

The above applies to compact arrays (all the antennae involved are interconnected and local to each other) and VLBI where data is recorded and combined at a later date. When the wavelength being measured is less than a millimeter any unwanted movement (or movement not accounted for) of even a millimeter introduces serious errors, each of which needs to be calculated and removed; leaving multiple sources of very tiny errors (which randomly add and subtract from each other, introducing noise).

Read on to page 547:

4.3. VLA-Y and GMRT-Y

The main benefit of the VLA's "Y" configuration is that it is a convenient 2-D arrangement of antennas which gives reasonable 2-D snapshot Fourier plane coverage. The BIMA and OVRO "T" arrays are similar in concept. The bad things about a "Y" or a "T" are that the regularity in the antenna directions along the arms will lead to a sort of grating response in the point spread function, and that it will take several hours of earth rotation synthesis for the Fourier samples to overcome this deficit.

The "Y" and "T" configurations are compromises: they seek to maintain the convenience of a 1-D array in reconfiguring the antennas, but would also like to get good Fourier plane coverage. As such, they are like arrays of fractal dimension 1.5: better than 1-D, not so good as fully 2-D arrays. Examples for the VLA's coverage and beams for a snapshot and a full track are shown in Figures 27-3 and 27-4.

The Giant Metrewave Radio Telescope (GMRT) in India has 14 inner antennas in a 1 km configuration and 16 in an outer configuration in the shape of an irregular "Y" (see Figure 27-5). The two configurations will often be used separately (the 30 m antennas were not designed for reconfiguration). The irregularity of the "Y" mainly stems from where land could be obtained, but it also produces a snapshot beam with 16 antennas which has much lower sidelobes than the VLA snapshot beam has with 27 antennas.

The wye (Y-shaped) configuration uses the least antenna groups to obtain the best result by filling in enough points that it's convenient to fill in the Fourier plane using an optimized algorithm, it is also convenient to position such a shape on existing continents. A five armed star provides better results, again with fewer antennas, but is obviously more expensive than the wye configuration. The best configurations, such as the randomly filled circle, don't fit multiple continents and their intervening oceans.

Notice the Y-shaped configuration of the current EHT array sites in user rugk's question. More sites in a narrow slice of the Earth's circumference allow increased capabilities but if a particular sector is competently (capability) covered it's a duplication of resources and an increase of processing time compared to a perfect location that fills in a hole. A site that is near the edge is useful to increase the observation window, but one too far away to be able to simultaneously observe with a sufficient number of other site isn't anywhere near as useful. Wherever the site is it must be capable of operating at the necessary frequencies and have enough sensitivity to gather useful data at that distance. It's a tall order.

  • $\begingroup$ For clarification: SPT has been part of the array since the beginning (2017) and is on the map in the original question. The mount carries an additional 230 GHz receiver that is enabled during the campaign window. It uses the primary mirror, but has drop-in secondary/tertiary mirrors to pick off the beam and separate cryogenic/readout packages. There are several advantages to its inclusion - it forms some of the largest baselines and the weather is probably the most consistently good in the array because there is so little precipitable water in the atmosphere there. $\endgroup$
    – Josh
    Commented Apr 7 at 14:45

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .