Active Asteroid Identification with SNAPS in Preparation for LSST

Introduction

Large-scale surveys of the sky are increasing the data volume for Solar System science. Deriving properties for a large number of primitive bodies makes it possible to identify rare cases that may offer important clues about the formation and evolution of the Solar System. We present results of our successful search for active asteroids– objects whose physical properties change over time and exhibit comet-like behavior. Asteroid activity is hypothesized to occur with physical instabilities, causing the sublimation of materials off the surface (Jewitt et al. 2012). Using the Solar System Notification Alert Processing System (SNAPS; Trilling et al. 2023) database, we have identified both known and unknown active asteroids, and demonstrated our ability to do so with future surveys, like LSST. 

The Solar System Notification Alert Processing System (SNAPS)

SNAPS is an alert broker that receives and processes data from all-sky surveys, deriving properties for asteroids such as rotation period, absolute magnitude, color, and produces lightcurves. SNAPShot1 (Trilling et al. 2023) ingested Zwicky Transient Facility (ZTF) alert data from 2018 July to 2020 May. SNAPShot1 alone accounts for 31,693 asteroids with a total of 5,458,459 observations. This large scale of data allows for the application of astroinformatics, a cross-disciplinary field that encompasses astronomy, data science, statistics, computer science, among others. The combination of these practices can be used to constrain behaviors for populations of small bodies, as well as statistical outliers. SNAPS plans to ingest the Legacy Survey of Space and Time (LSST) data in a similar fashion, and provide real-time alerts for objects of interest to the science community. LSST will report hundreds of measurements for each of some 5 million asteroids, and the first few months will produce the first LSST measurements of more than a million asteroids. It is important that the community is prepared with tools such as SNAPS to maximize the ability to draw conclusions about Solar System origins and evolution.

Methods

We investigate asteroids that vary in properties such as absolute magnitude and color over time, as well as evaluate postage stamps from the ZTF alert stream. We tested numerous methods to automate the candidate flagging process such as null-hypothesis significance testing and machine learning. Figure 1 shows the null-hypothesis method results, where the null-hypothesis assumes that an asteroid does not change in absolute magnitude over time. We hypothesize that outlier candidates reside in the 10-20-10-5 p-value range, a high probability of not fitting this criterion. Additionally, a Convolutional Neural Network (CNN) model has been trained to linearize postage stamps and classify objects as inactive or active. As shown in Figure 2, one object has a visible tail in the top right quadrant (North-West direction) of the postage stamp. The anti-solar vector aligns closely with the tail’s signature, agreeing with the expectation that comet tails either manifest in the anti-motion or anti-solar direction. This main belt asteroid also shows signs of recurrent activity, appearing elongated in observations taken a year apart.

Current and Future Work

We will continue to search for activity in archival and real-time ZTF observations. Applying similar evaluations to forthcoming LSST data on a nightly basis will allow us to identify other interesting objects, and issue alerts to the science community for follow-up. As LSST begins to release data, we plan on integrating LSST and ZTF results, in search of common targets. In turn, we may constrain outlier behavior for asteroids over the course of multiple epochs, but also be able to identify activity with a singular observation.

Figure 1. Probability-value histogram for all asteroids in SNAPShot1. Lower p-values have a lower probability of satisfying the null hypothesis.

Figure 2. Postage stamp for active asteroid in both minmax scale (left) and z-scale (right). The yellow vector represents the anti-solar vector while the red denotes the anti-motion vector (not to scale). The images are oriented with North up and East to the left.

This work is funded in part by the Arizona Board of Regents Technology and Research Initiative Fund, and by NASA and NSF grants to DET and MG.

Jewitt et al. 2012, AJ, 143, 66

Trilling et al. 2023, AJ, 165, 111