Tracing the Ghosts of Clusters: StarStream Reveals Hidden Stellar Streams in the Milky Way

Yingtian Chen and collaborators present StarStream, a new algorithm that automatically finds stellar streams in the Milky Way using data from the European Space Agency’s Gaia mission. Stellar streams are long, thin trails of stars that once belonged to globular clusters, ancient star groups that orbit the Milky Way but are slowly torn apart by its gravity. Using StarStream, Chen’s team discovered 87 such streams, doubling the number previously known and revealing that many globular clusters are actively losing stars today.

Mapping the Milky Way’s Stellar Streams

The Gaia spacecraft provides astronomers with six-dimensional maps of stars, positions and motions across the entire sky. Traditional methods to find stellar streams relied on recognizing visually thin, orbit-aligned patterns in these maps, but many streams are irregular, wider, or misaligned due to complex gravitational effects from the Milky Way’s disk, bar, and dark matter halo. To overcome this, Chen developed StarStream, which uses kernel density estimation and physical modeling of stream formation to identify real streams without assuming any particular shape. This allows the algorithm to detect streams even when they appear fuzzy or curved, revealing structures that previous visual searches missed.

How StarStream Works

StarStream compares the distribution of Gaia stars around each globular cluster to models of how tidal streams should look, using a realistic Galactic potential and synthetic “tracer” stars ejected from simulated clusters. The program assigns each star a probability of belonging to a stream based on its position, motion, and color. By testing the method on mock data, Chen confirmed that StarStream can detect streams with roughly 60% completeness and purity in regions where dust and background contamination are low, particularly at high Galactic latitudes (|b| > 30°).

A New Population of Streams

Applying StarStream to Gaia’s third data release, the team identified 87 potential globular cluster streams, including 34 high-quality detections. Many newly found streams are wide, short, or misaligned by more than 10° from their clusters’ orbits, proving that streams can take diverse forms. The researchers even rediscovered known systems like Palomar 5 but with more stars than previous surveys found. Some streams, such as those around NGC 4147, are so diffuse they appear almost circular rather than filamentary, yet still show coherent motion and color signatures that confirm their common origin.

Measuring Cluster Mass Loss

One of StarStream’s strengths is that it can provide unbiased measurements of how quickly clusters are losing stars, known as the mass loss rate. By comparing the number of detected stream stars to realistic simulations, Chen estimated that most clusters shed between 1 and 100 solar masses per million years. Interestingly, the rate increases for clusters that are larger and less dense, suggesting that “fluffy” clusters like Palomar 5 and Whiting 1 are nearing complete tidal dissolution. These results align with theoretical predictions that clusters rich in black holes lose mass faster as they evolve.

Implications for the Milky Way’s History

The discovery of so many new stellar streams reshapes astronomers’ understanding of the Milky Way’s structure and evolution. Streams preserve detailed records of their clusters’ past orbits and interactions, offering a way to map the galaxy’s gravitational field and the distribution of its dark matter. Chen’s work shows that nearly three-quarters of globular clusters at high latitudes display tidal streams, implying that cluster disruption is a common and ongoing process in the Milky Way.

A Step Forward in Galactic Archaeology

By releasing StarStream as an open-source tool, Chen and collaborators have provided astronomers with a powerful new way to uncover faint stellar structures hidden in Gaia’s massive dataset. As future surveys provide even better data, including measurements of stellar chemistry and velocity, methods like StarStream will help piece together how the Milky Way’s oldest clusters have been shaped, stretched, and stripped over billions of years.

Source: Chen