Contact Us
2 September 2021
OPINION - The Universe in the cloud: Data-intensive astronomy and its benefits
Whether astronomers study phenomena in space that vary over time, images captured of astronomical objects, or simulate the whole Universe and what telescopes can see, they share a few things that make possible their quest for understanding. One of these is computers. Astronomers use computers a lot – from specialised computers at the back of telescopes to the laptops on which they work every day. 

The second aspect of astronomical research is code. Indeed, astronomers spend more time programming computers than actually staring at beautiful images of the sky. In fact, those images are the result of both the work of telescopes and scientific codes and software applied to those observations. Observations or simulations – data – are also an element that all astronomers need. Data is the term we use to describe any sort of digital information that we use to extract scientific knowledge.

Astronomical data is generated either by what telescopes detect, or by simulations. Like our five senses, telescopes detect a lot of things. When, for example, we focus on watching a movie, we would rather not have other noises bother us. In astronomy we use the same word – “noise” – to describe what our telescopes pick up but are not of interest. For example, if you want to look at a galaxy far far away, our own galaxy may be in the way. The signals picked up by a telescope from the Milky Way may then be considered noise. 

But we also have a saying in astronomy: One astronomer’s noise is another astronomer’s data. Indeed, nearly everything is interesting in astronomy. We even need to understand what human-made signals, such as those coming from satellites, look like when seen from our telescopes, because we need to remove those. This is how we make sure the data we keep really comes from what we are trying to observe. And with ever more powerful telescopes, the sheer amount of data itself becomes “astronomical”. That’s why we say that astronomy is a data-intensive field of science. 

In summary, telescopes, computers, code and data are the tools of the trade. The Square Kilometre Array (SKA) is being built in South Africa, starting with MeerKAT, which is already producing unequalled observations of the Universe. There is a need to make sure that the data from these telescopes is available to South African researchers and students – and that they have the skills required to analyse the data. 

This led to the establishment of the Inter-University Institute for Data Intensive Astronomy (IDIA) in 2015. IDIA is currently a partnership between the University of the Western Cape, the University of Cape Town (UCT) and the University of Pretoria, as well as the South African Radio Astronomy Observatory (SARAO) which built and runs MeerKAT and manages the SKA site in the Karoo. IDIA at UWC is closely linked to the UWC Astro Group. Prof Russ Taylor (pictured) is the founding Director of IDIA, and an NRF A-rated scientist who holds a research chair jointly at UCT and UWC and splits his time between Canada and South Africa. Prof Taylor was the founding international SKA Project Scientist and co-authored the first SKA Science Case. The Associate Director at IDIA is Prof Carolina Ödman, who leads the development and outreach programme.

All this data requires some new technological tools. One of the key technologies we use is cloud computing. The data and the codes are on computers “in the cloud”. This simply means that it doesn’t matter where the computers are located. As long as you have an internet connection you can create codes and apply them to the data wherever it is stored, and all you need to see gets served to you through the web. In South Africa, the ilifu research cloud was built by IDIA. Ilifu also has partners at the Cape Peninsula University of Technology (CPUT), Stellenbosch and Sol Plaatje universities, and also serves the bioinformatics community, because genetics and bioinformatics are also data-intensive areas of science and South Africa is a leader in those fields on the continent. With cloud computing, researchers don’t need to download the data or the codes, just the results – and codes tend to be much, much smaller than the data itself. A part of turning data into science in astronomy is even called data reduction. 

By means of cloud computing, astronomers are also able to collaborate on the data and share codes and results, which also helps to make sure that our research results are reproducible. It is a validation process whereby other scientists need to be able to carry out the same analysis and obtain the same results. Nowadays, it is not enough to describe the method used. The data and the code are part of a scientific analysis. They are digital assets that are part of scientific publications. Considering this, there is therefore also a need for a new approach to research data management. Prof Mattia Vaccari is a researcher in astronomy and Director of eResearch at UWC. In this position, he is making sure that the university and its researchers are equipped for this new paradigm.

Many of the codes and the software written by astronomers are common to many projects and the programmes themselves are the fruit of large software collaborations. One of IDIA’s areas of interest is data visualisation and one of the big software packages that IDIA scientists and staff are working on is called CARTA, the Cube Analysis and Rendering Tool for Astronomy, which enables the creation of images from the large file sizes obtained from radio telescopes such as MeerKAT. Without software like this, it is very difficult to see what’s going on in the data.

Prof Vaccari, when focusing on astronomy, is tackling yet another challenge for astronomers: how to combine different observations. Imagine that you see some galaxies in infra-red light using one telescope, then in radio waves using another telescope. If you can combine what you see from those galaxies in both infra-red and radio, you can learn a lot more about them. But it is not as easy as it seems. Some objects appear point-like in one type of light, but look like spread-out smudges in other types of light. How do you make sure you can superimpose those observations accurately? That is the focus of Prof Vaccari’s project HIPPO (Help-IDIA Panchromatic Project). It aims to create a cloud-based environment where radio data can be combined with observations in other wavelengths to study the evolution of galaxies, using machine learning algorithms to sift through large amounts of data in an automated manner.

Postdoctoral researchers Dr Matt Prescott and Dr Fangxia An (pictured) work with Profs Vaccari and Taylor on MeerKAT data from the MIGHTEE project for galaxy evolution studies. Matt leads the MIGHTEE cross-matching effort, combining MeerKAT data with optical datasets, while Fangxia works on combining MeerKAT radio data with data from other SKA pathfinders such as GMRT in India and LOFAR in Europe, to gain insights into the nature of distant radio sources.

Chaka Mofokeng and Sibusiso Mdhluli work with Profs Vaccari and Taylor on their MSc projects, in which they aim to classify radio galaxies using machine learning techniques such as convolutional neural networks (Chaka) and self-organised maps (Sibusiso). This task has mostly been done by means of visual inspection of individual galaxies by astronomers, but this is rapidly becoming impractical and needs to be automated by the time the SKA comes online. Their main aim is therefore to evaluate the effectiveness of the different techniques and to further develop them so that we are ready to use the SKA to the best of its capabilities from day one.

Combining the observing capabilities of new astronomical instrumentation and the data processing capabilities offered by cloud computing technologies, we can now also study if and how the sky changes over time, in unprecedented detail. Boikuthso Mabala and Mfundo Mdwadube work with Profs Vaccari and Taylor on the ADFS-MeerKAT project, which is using MeerKAT and MeerLICHT (an optical telescope at Sutherland) to simultaneously observe the same patch of the sky in the radio and in the optical over a period of two years and determine if and how radio sources change in brightness.

IDIA aims to ensure that communities, learners, the public and other students benefit from its work. Prof Carolina Ödman (pictured) has designed and led this arm of IDIA. IDIA regularly runs hackathons, in partnership with its partners in the Office of Astronomy for Development (OAD), the UK-South Africa DARA Big Data programme and SARAO. In a hackathon, students – from undergraduates to PhD level – come together over an intensive two-day event during which they learn to apply machine learning to different types of data sets. Inspired by the tools of radio astronomy, students acquire the ability to find pulsars in radio astronomy data, create a movie recommender engine or analyse the public’s sentiment about the coronavirus pandemic from Twitter data. 

These events are well attended and were run in-person until March 2020, when an event was held at Sol Plaatje University, and then in various hybrid and online formats since. Participants have come from South African universities as well as all the eight African partner countries in the SKA, plus a few more.

Each event happens in the cloud, with IDIA’s research cloud being made available to the students for their projects. IDIA’s cloud computing platform is also used to support the DARA Big Data Research schools organised by SARAO since 2018. In this context, students spend two weeks carrying out big data research projects. Because these are intense skills-development events, IDIA also brings in people to help develop industry skills. From a CV lab to intervention by entrepreneurs to a mock start-up competition, role-play is used to help the students think about their big data projects and skills from a different perspective, and familiarise themselves with industry thinking and practices.

While scientific skills are being developed, the Development and Outreach arm of IDIA also seeks to ensure that this work benefits, and inspires communities. IDIA and the CRC have been supporting Sibusiso Mdhluli’s Summer School in Acornhoek, the small town he comes from in Mpumalanga for the last couple of years. This school helps Grade 11 learners get ahead on their Grade 12 maths and science programme with the help of previous learners, now science students across the country.

To explain the science that happens, there needs to be equitable language. In collaboration with UWC’s Xhosa department members Dr Sebolelo Makopela and her student Sinethemba Nobom, Chaka Mofokeng and Prof Ödman are developing resources about astronomy in isiXhosa and other African languages. This takes a lot of thinking as the team tries to go beyond just co-opting English scientific terms.

IDIA’s Development and Outreach office is committed to the development of young people using astronomy, and this comes at a time where the continent is getting ready to use the SKA radio telescope. Cape Town was chosen as the destination for the International Astronomical Unions’ General Assembly in 2024, when the global astronomy community is coming together, presenting an opportunity to showcase African talent in science and technology and the warmth of the African spirit.

Prof Carolina Ödman is the Associate Director Development and Outreach, Inter-University Institute for Data Intensive Astronomy based at UWC

Prof Mattia Vaccari is eResearch Director at UWC

Prof Russ Taylor holds a Research Chair supported by UWC, UCT and SARAO

Read the latest edition of Signals Research Magazine HERE
Listen to Prof Ödman’s interview with Lester Kiewit, of CapeTalk, HERE