Determining the size distribution of asteroids is critical to understanding the collisional history and evolution of the inner Solar System. A study published in Astronomy and Astrophysics journal highlights 1,031 previously uncategorised asteroids found from archival Hubble data. A group of volunteers known as “citizen scientists” trained the AI model to spot faint streaks of light left behind by the small space rocks.

The study aims to improve the knowledge of the size distribution of small asteroids in the main belt by determining the parallaxes of newly detected asteroids in the Hubble Space Telescope (HST) archive and, subsequently, their absolute magnitudes and sizes. The streak appearance of asteroids in Hubble photos is a result of the telescope raising around Earth as it takes long-exposure images. The rock would typically go unnoticed in these images because they are a million times fairer than the faintest stars. However, the streaks are more noticeable, allowing astronomers to gain insights about their relative size and orbital characteristics.

Asteroids appear as curved trails in HST images because of the parallax induced by the spacecraft’s fast orbital motion. Taking into account the trajectory of this latter, the parallax effect can be computed to obtain the distance to the asteroids by fitting simulated trajectories to the observed trails. Using distance, we can obtain the absolute magnitude of an object and an estimation of its size assuming an albedo value, along with some boundaries for its orbital parameters.

Pablo García-Martín, a researcher at the Autonomous University of Madrid in Spain, and the study’s lead author stated that they are surprised to such a large number of candidate objects.

Analysis using AI

In the study, the scientists analysed a set of 632 serendipitously imaged asterasteroids found in the ESA HST archive. Images were captured with the ACS/WFC and WFC3/UVIS instruments. As part of a previous study, a machine learning algorithm (trained with the results of a citizen science project) was used to detect objects in these images. 

The raw data used for the study consisted of 1031 asteroid trails from unknown objects, which did not match any entries in the Minor Planet Center (MPC) database using their coordinates and imaging time. They also found 670 trails from known objects. The study supports the idea that the asteroids are fragments of larger asteroids that have collided and broken apart over billions of years.

According to the conclusion stated in the study, One of the advantages of applying machine learning to find Solar System objects in complete astronomical archives is the large number of potential results obtained. This allows the researchers to apply purposely strict filtering conditions to improve accuracy while still keeping a large enough sample to obtain statistically meaningful results.

The team hopes to use similar AI techniques to search through different archival datasets and discover more hidden space objects.

Research Paper: Hubble Asteroid Hunter

Image: Unsplash