Abstract
The Hubble Catalog of Variables (HCV) project aims to identify the variable sources in the Hubble Source Catalog (HSC), which includes about 92 million objects with over 300 million measurements detected by the WFPC2, ACS and WFC3 cameras on board of the Hubble Space Telescope (HST). By using an automated pipeline containing a set of detection and validation algorithms, the project pre-processes HSC sources to correct systematic effects and remove bad measurements. The corrected data are used to compute a number of variability indexes to determine the variability status of each source. The final catalog will contain variable stars, active galactic nuclei (AGNs), supernovae (SNs), and potentially new types of variables, reaching an unprecedented depth ($V \leq 27$ mag). The HCV will be available at the Mikulski Archive for Space Telescopes (MAST) and the ESA Hubble Science Archives, with the pipeline deployed at STScI for future updates.
Introduction
While the Hubble Space Telescope (HST) is famous for its stunning deep-space images, it was not originally designed as a time-domain survey instrument. Consequently, the Hubble Source Catalog (HSC)—a collection of 92 million objects—is incredibly "inhomogeneous," containing data from different cameras (WFPC2, ACS, WFC3), varying filters, and inconsistent observation strategies. The primary challenge addressed by the HCV project is distinguishing true astrophysical variability from "noise" introduced by the telescope itself. Systematic effects like "telescope breathing" (small focus changes due to temperature fluctuations), PSF variations, and slight image misalignments can make a perfectly steady star appear to flicker. The HCV project aims to clean this data and provide a reliable gold-standard catalog of variable sources.
Methodology and Pipeline Development
The authors developed an automated pipeline to process lightcurves for millions of sources. Key innovations include:
- Synthetic Error ($Errsyn$): Instead of relying solely on standard photometric error, the team created a "synthetic error" metric that accounts for image artifacts, the "concentration index" (object fuzziness), and positional shifts between exposures.
- Two-Stage Outlier Detection: The pipeline uses robust regression to fit a baseline to each lightcurve, flagging "strong outliers" (likely cosmic ray hits) and "potential outliers" (statistically suspicious data).
- Local Zero-Point Corrections: To fix systematic drifts, the pipeline uses the median brightness of neighbors within a 20-arcsecond radius to re-calibrate target measurements.
- Variability Indexes (VIs): The team computed 20 different statistical indexes for every object, ranging from simple scatter-based metrics to complex correlation-based tests.
Datasets
The project utilizes the Hubble Source Catalog (HSC) version 2, released in September 2016. This dataset includes:
- 92 million objects with over 300 million individual measurements.
- Data spanning 26 years of HST operations.
- Observations across 112 different filter/instrument combinations.
Key Results and Insights
The pipeline identified approximately 8,000 multi-filter candidate variables (MFCV) and 47,000 single-filter candidate variables (SFCV) from an initial pool of ~2 million pre-processed targets. The study found that a combination of Median Absolute Deviation ($MAD > 5\sigma$) and reduced chi-squared ($\chi^2_{red} > 3$) was the most robust way to filter out noise while keeping true variables.
A major insight from this work is that for bright objects, the error budget is dominated by systematic effects rather than photon noise. The $Errsyn$ approach effectively weights the reliability of measurements based on the telescope's state. Furthermore, by reaching a magnitude of $V \leq 27$, the HCV probes much deeper than ground-based surveys, allowing for the study of variability in distant galaxies and faint stellar populations.
Limitations and Future Outlook
The pipeline currently requires at least 5 measurements per lightcurve, leaving millions of objects with fewer observations as "unknowns." Additionally, multi-filter candidates still require visual inspection to confirm they are not artifacts. The ultimate goal is to deploy this pipeline at the Space Telescope Science Institute (STScI) for automatic updates with future HSC releases, making the HCV a living resource for the astronomical community.
Summary updated