Citizen science volunteers recently carried out an extensive search using outdated images taken by space telescopes.
Sorting through 37,000 images from the archives of the Hubble Space Telescope, 11,000 citizen scientists joined professional astronomers in the hunt for faint solar system objects that were missed in previous surveys.
The project has helped identify 1,031 previously unknown asteroid trails.
More than 400 of these were the trails of small asteroids less than 0.6 miles wide, which are essential in the study of space impacts of Mars and Jupiter.
Sifting through decades of orbital observations
The Hubble Space Telescope was never designed to spot asteroids.
With its primary purpose being far-off galaxies and nebulae, the solar system objects fell into the telescope’s view unintentionally while observing these distant targets.
There were thousands of such “photo bombers” in the archive.
The team accumulated some 37,000 photos taken by the Hubble from 2002 to 2021, forming a dataset that was just too large to be analyzed by one research team.
This database had to be unlocked through the efforts of the broader society.
Researchers shared thousands of images with the public in the online database, allowing space fans to help with the process of analyzing decades’ worth of astronomical observations.
Spotting curved streaks created by parallax
Asteroids in transit form a clear light track.
Since the orbiting Hubble completes an orbit around Earth in just 95 minutes, parallax turns the linear asteroid trajectory into a distinct curve.
This form is easily recognizable to human eyes.
Both volunteers and the machine-learning algorithm were trained to recognize cosmic rays and distinguish them from real asteroid trails. Candidate detections were also visually checked by the authors.
Thousands of image grids were scanned.
Over 11,000 users evaluated possible images of asteroids on the Zooniverse website and identified and recorded their positions.
Combining crowdsourced tagging with artificial intelligence
Classifications performed by citizen scientists were used to train a machine learning algorithm.
Scientists leveraged the massive dataset of human-tagged asteroids to train the automated deep learning neural network to efficiently scan through Hubble’s remaining archives.
The combined process validated over 1,000 missing trails.
With the help of crowdsourced pattern recognition and automated analysis, the researchers confirmed the existence of a total of 1,701 asteroid trails corresponding to 1,031 previously undiscovered asteroids in the main belt.
Machine algorithms validated the crowdsourced tags.
AI helped find the trails; a separate parallax analysis estimated distances, sizes, and limits on orbital parameters for a filtered subset.
Uncovering tiny fragments of ancient collisions
The recently discovered asteroids were extremely faint.
Thanks to its exceptional optical sensitivity, the Hubble was able to see extremely faint objects that were missed by ground-based surveys.
Approximately 400 of the discovered objects were less than 0.6 miles in diameter.
These small fragments come down to a fraction of a mile across, representing an elusive population of smaller objects created during ancient interplanetary impacts.
Small asteroids trace the history of space collisions, according to NASA Hubble.
Mapping the unseen population of the asteroid belt
Small asteroid populations serve as a probe for planetary evolution theory.
The use of actual data of small asteroids in relation to mathematical calculations increases the validity of mathematical theories that explain collisions and matter fragmentation during the evolution of our solar system.
Archival data is still capable of yielding scientific insights.
By utilizing decades of archive observations by space telescopes alongside citizen science efforts, it becomes evident that there is much more to astronomy to discover even in existing data.
Future surveys will take advantage of these crowdsourced findings.
As next-generation observatories begin mapping the night sky, applying hybrid artificial intelligence and citizen science workflows will allow researchers to track millions of faint asteroids across our cosmic neighborhood efficiently.
The complete study can be read here: García-Martín, P., Kruk, S., Popescu, M., Merín, B., Stapelfeldt, K. R., Evans, R. W., … & Thomson, R. (2024). Hubble Asteroid Hunter: III. Physical properties of newly found asteroids. Astronomy & Astrophysics, 683, A122.
Read the whole thing?
Get the week's signal, not the noise
Our sharpest reporting on energy, climate and nature — free, once a week.
