Bad Astronomy | THOR program accelerates search for near-Earth asteroids

Note: This article was written in part to promote Asteroid day on June 30th, a global effort to raise awareness of the dangers and scientific importance of asteroids. It is June 30th every year, the anniversary of the great Tunguska impact of 1908, and the B612 Foundation mentioned below is one of the founding partners. I was supposed to be in Luxembourg – Asteroid Day HQ – to moderate some panels and talk about asteroids, but a health issue (now solved!) kept me from traveling. Still, I hope you check out all the cool events planned, including live streams with scientists, astronauts, and other experts. Learn things and have fun!

The search for near-Earth asteroids has just made a very big leap thanks to THOR.

Yes, different THOR. This stands for Tracket-less Heliocentric Orbit Recovery and is a method that not only accelerates the finding of asteroids enormously, but also enables the search using old, archived images regardless of the time of recording. It’s faster and can use the huge database of observations that just lie around online. So yeah, that’s a big deal.

Finding asteroids in general isn’t hard, it’s just time consuming. As they orbit the sun, they appear to move slowly across the sky. So you use a telescope to take a picture of a single point, wait a short while—usually go to other points in the sky to observe them—and then observe the same point again. Do it again and now you have three images of the same part of the sky.

Stars don’t move. So when you line up the three images, all the stars appear in the same place, but the asteroid has moved, forming a line of three dots. This is the trace of its movement over that time, hence this short line is called a tracklet. It may be enough to use the centuries-old equations of motion to create a predicted orbit for the object, and the equation describing that orbit can then be projected into the future or past to see where in the sky it will be or not was; future observations or previously archived observations can be searched to see if it is present and the orbit can be refined.

In practice, of course, it’s much more complicated, but that’s more or less how it was done. One problem is that this method takes a lot of computing time and is not very efficient. Another reason is that asteroids don’t always seem to move in straight lines; The movement of the Earth around the Sun — or the movement of an orbiting observatory around the Earth — can wobble these lines, making it harder to spot asteroids. When huge surveys come online in the next few years they will find millions of asteroids (!!) and this method will get bogged down trying to find them all.

Enter THOR [link to paper], a project developed by the Asteroid Institute, a project of the B612 Foundation. The idea here is not to track the asteroids themselves, but to create theoretical test orbits for an asteroid, which is a little backwards from the usual approach. A test orbit is really just the equation for an imaginary orbit, say circular, 300 million km from the Sun at a given inclination and orientation. This produces a series of numbers denoted as orbital parametersand they in turn define an equation that can be solved according to where an asteroid is at any given time.

This test orbit is then projected forwards or backwards onto the times of the other observations, which are then searched for objects that are close to this path. Algorithms for this type of search are widely used and tend to be quite fast.

This method has several advantages – the Asteroid Institute has a good FAQ to explain all of this – but one that really stands out is that it doesn’t necessarily need observations that are close in time and at a specific cadence to work . The position of a potential asteroid in a test orbit can be calculated at the time of any observation from any observatory. Because we know when an observation was made and where in the sky it was taken, it’s possible to see if the potential asteroid was in that observation at the time, even if it was taken weeks or more ago.

This is extremely powerful. There are many – a a lot of – from astronomical observations stored in databases, and in fact the team that developed the algorithm tested it on real data. They used two-week observations from the Zwicky Transient Facility, a massive all-sky survey, to search for potential asteroids and were able to recover over 97% of the previously known asteroids that showed up in the data! Impressive.

They also used data from the NOIRLab Source catalogue, a vast database of astronomical observations, and examined a month’s worth of observations. They found 104 new asteroids in data confirmed by the Minor Planet Center. So it can find both known and new asteroids. This is important because new observations can trigger thousands of alerts about potential asteroids; If these can be quickly sorted out for known asteroids, that saves a lot of time right there.

THOR can carve out asteroids quickly and across different observations, using old imagery to really nail the orbits as well. When these new big surveys come online, it looks like THOR will be incredibly useful for finding many of the asteroids that are expected to be discovered – something like 6 million in the next decade.

That’s a lot of stones. Knowing where they are and, more importantly, where they’re going to be is obviously pretty important, so I’m all for it.

Note: If you are a code nerd, you can find THOR on GitHub.