Research
My thesis focuses on solar system objects (SSOs) represented in the NOIRLab Source Catalog (NSC), which was created by Dr. David Nidever (my advisor).
NOIRLab Source Catalog
The NSC covers ~3/4 of the sky, including most of the galactic plane and both Magellanic Clouds. Exposures are from NOIRLab's Astro Data Archive, and were captured with CTIO 4 m+DECam, KPNO 4 m+Mosaic-3, and Bok 2.3 m+90Prime. The first data release of the NSC was made public in 2018 and has 34 billion measurements of 2.9 billion distinct astronomical objects from 255,454 exposures (Nidever et al, 2018). The second data release has 3.9 billion measurements of 68 billion objects from 412,116 exposures (Nidever et al, 2021). I am assisting in the process of creating NSC DR3, which will include data over 500,000 exposures and improved measurements (Fasbender et al, in preparation).
Most objects in the NSC are relatively stationary (distant stars, galaxies), but those relatively nearby will move noticeably over short periods of time (SSOs). To distinguish between "stationary" and "moving" objects, we develop methods to assign each individual NSC measurement to its unique corresponding object. This is possible because of the high proper motion of SSOs, as observed from Earth. Displacement of an SSO's measurements between exposures results in the formation of a fairly linear path that traces SSO motion when the exposures are stacked. These structures are called "tracklets," and consist of 3+ detections of a unique SSO taken within a single evening.
SSO Detection: CANFind
The Computationally Automated NSC tracklet Finder (CANFind, Fasbender & Nidever 2021) uses an iterative implementation of the clustering algorithm Density-based Spacial Clustering of Applications with Noise (DBSCAN). With different parameters for each application, DBSCAN forms groups of measurements that are likely to belong to the same object, based on their spatial distribution. CANFind discards stationary objects from the data and identifies tracklets within the remaining points. Each tracklet undergoes a validation process to evaluate its likelihood of representing an SSO.
The catalog of 500,000 tracklets detected by CANFind in NSC DR1 is available for download (.zip file).
CANFind is running on NSC DR2; results will be available soon.
SSO Detection: The "Tough" Transform
An adaptation of the Hough Transform (Duda & Hart 1972), tentatively referred to as the "Transformed-Hough Transform" will form tracklets from more dispersed detections.
The above animation cycles through four exposures included in the NSC, captured about 20 minutes apart. Distant stars and galaxies move so slowly from our perspective that they appear to be stationary as time progresses. Objects closer to home (asteroids, comets, etc.) show more displacement between exposures; one can easily be spotted as it moves past the background stars.