The first common source of contamination is the atmosphere itself (yes, again). There is more than 60 miles (100 kilometers) of air between us and the vacuum of space, and it is constantly shifting, with pockets of warm and cool air competing for dominance. Every time the atmosphere shifts, it slightly changes our view of celestial targets.
To account for this, the EHT team dedicated a fraction of each observing run to training their instruments on a well-known radio source. They then used observed variations in that source to create a real-time model of atmospheric turbulence and its effect on the Sgr A* data, allowing them to remove any atmospheric distortions.
Beyond the atmosphere, there is also a lot of galactic material between us and Sagittarius A* — about 26,000 light-years' worth. And while interstellar space is almost a vacuum, it isn’t perfectly so, meaning dust grains throughout the galaxy interfere with the radio emissions picked up by the Event Horizon Telescope.
One effect of that dust is to gently scatter the radio waves coming from Sgr A*, making it appear broader than it really is. The second effect is that large, random clumps of interstellar dust introduce small blotches that aren’t a part of the black hole system at all. That meant the team had to work hard to develop models of those effects before they could likewise subtract the from the final image of our home galaxy’s black hole.
Lastly, astronomers had to consider the inherent variability present in the disk surrounding Sagittarius A* itself. Previous, much lower-resolution observations suggested that our supermassive black hole’s disk can double in brightness over the course of only a few years or less. Astronomers have even caught the occasional flare popping up around the black hole and disappearing within a single day.
The EHT team needed to train their telescopes on Sgr A* for several hours. They required all that data to ensure the signal clearly rose above the noise — otherwise, the observation would be so noisy it would be nearly useless. But because the black hole’s disk changed and varied in brightness over all that time, it was like taking a picture of a dog chasing its tail. The team couldn’t simply combine several hours’ worth of data into a single blurry mess.
To tackle this, the team divided the data stream into small chunks, no longer than a few minutes each. They then processed each chunk separately, then combined all the clean chunks together to make a single, average image. As a self-check for consistency, the team used separate software pipelines with different methods for cleaning and processing the data.
The end result of all this work — after years of preparation, days of observation, and years of analysis — is a gorgeous portrait of the gravitational goliath hiding in the center of our galaxy: Sagittarius A*.