STIPS

Telescopes of the 1-meter class are often called the workhorses of observational astronomy. They follow up supernovae and other transients, time exoplanet transits, monitor variable stars, and train students, and many hold archives that go back decades. Their role is growing as surveys like the Zwicky Transient Facility and Rubin Observatory’s Legacy Survey of Space and Time (LSST) find far more transients than large telescopes can follow. Yet few of these telescopes have an actively maintained, public image-reduction pipeline. Most groups end up writing their own, which falls hardest on the users these telescopes serve best: community colleges, outreach programs, and small research groups.

STIPS, the Small Telescope Image Processing Suite, is an open-source package that runs the LSST Science Pipelines on these telescopes. I am the lead developer, working with Kyle Westfall at UC Observatories, and a paper describing it is in preparation for the Astronomical Journal.

Borrowing a survey’s software

Vera C. Rubin Observatory built the LSST Science Pipelines to process a ten-year survey of the southern sky that will make some 32 trillion observations of 40 billion objects. Along with state-of-the-art image-processing algorithms, they include a data management layer, the butler, that stores every product together with the configuration and software versions that made it.

The pipelines can be adapted to other cameras, and they already process data from Hyper Suprime-Cam, DECam, and others. But each new camera normally needs its own “obs” package, a piece of Python that encodes the detector geometry, header conventions, and calibrations, and writing one takes detailed knowledge of the pipeline internals. Running the pipelines then means learning the butler’s data model, pipeline definitions, and a set of bespoke command-line tools. That complexity has kept the software out of reach for most small telescopes.

How STIPS works

STIPS replaces the obs package with a short, declarative description of the camera. Each instrument gets a directory whose central file, the instrument profile, lists the filters, the header keywords, the observatory site, and the detector geometry; everything else has a default. When the software runs, STIPS builds the instrument classes the pipelines need directly from that profile, so a new telescope needs only the configuration that differs from the defaults.

A reduction is then controlled by a single YAML file that names the instrument, the target, the nights and bands, and where the difference-imaging templates and reference catalogs come from. One command takes the data from raw frames to a chosen end point:

• Ingestion and calibration: screen and ingest the raw frames, then build and certify each night’s bias and flat field
• Single-visit processing: correct each frame for the instrument signature and calibrate it against Gaia DR3 and Pan-STARRS, with color terms for the instrument’s filters
• Difference imaging: subtract a template, built from the telescope’s own images or taken from a survey, to isolate whatever changed
• Photometry and light curves: forced PSF photometry at the target, or differential aperture photometry for bright stars and transits
• Data products: every product is stored in the butler with its configuration and software versions, and light curves and calibration metrics are exported as tables

Nights are processed independently, so a failure on one night doesn’t stop the rest, and the same pipelines can be sent to a Slurm or HTCondor cluster. Because STIPS calls the LSST software in separate processes, it installs in an ordinary Python environment, and container images with everything installed ship with each release.

Testing it on real telescopes

STIPS currently supports two instruments: the Direct Imaging Camera on the 1-m Nickel Telescope at Lick Observatory and Y4KCam on the CTIO 1.0-m telescope in Chile. Every number below comes from reprocessing the raw data from scratch.

• Calibration: against Landolt standard stars, mean Nickel residuals are within 0.012 mag in BVRI, and astrometric fits to Gaia DR3 have median residuals of 14–48 milliarcseconds. On six Y4KCam nights from 2011, 870 measurements of 71 Landolt standards agree with the catalog to medians between −0.028 and +0.039 mag, with a robust scatter (NMAD) of 0.022–0.044 mag.
• Supernovae: Nickel light curves of SN 2023ixf and SN 2020wnt agree with independent reductions of the same data to within 0.10 mag in the median.
• Exoplanets: a single night of Nickel differential photometry recovers the transit of HD 189733 b with a depth of 2.40 ± 0.06% and a mid-transit time within a minute of the published ephemeris.
• Archival data: Y4KCam images from a cloudy night in 2011 recover a transit of WASP-19 b with a depth of 2.16 ± 0.14%, consistent with the published 2.077 ± 0.014%, and a mid-transit time −1.1 ± 1.0 minutes from the ephemeris.

Links

I first presented STIPS at the 248th meeting of the American Astronomical Society in Pasadena in June 2026.

• Paper: Gause, D. & Westfall, K. B., STIPS: Bringing the LSST Science Pipelines to 1-Meter-Class Telescopes, in preparation for AJ
• Code: github.com/dangause/stips (BSD 3-Clause)
• Documentation: stips-docs.readthedocs.io
• Poster: AAS 248 iPoster