Pipeline Overview

LBCgo reduces LBC multi-extension FITS files through a sequence of stages, controlled by the top-level function lbcgo(). Each stage can also be called independently via the lower-level functions in LBCgo.lbcproc and LBCgo.lbcregister.

Stage 1: Overscan subtraction and trimming

go_overscan()

A fourth-order polynomial is fit to the overscan strip (defined by the BIASSEC header keyword) and subtracted from each row. The science region is then trimmed to TRIMSEC. Output files carry the _over.fits suffix.

Several LBC header keywords that confuse downstream astrometric tools are removed at this stage: BIASSEC, TRIMSEC, DATASEC, ROTANGLE, PARANGLE, and all duplicate astrometric keywords ending in A (introduced by the LBC readout electronics).

Stage 2: Master bias (optional)

make_bias() / go_bias()

Bias frames (IMAGETYP=zero) are median-combined with 3σ sigma clipping to produce zero.fits. Bias subtraction is disabled by default (bias_proc=False) because the LBC overscan provides a reliable per-image bias level. Enable with bias_proc=True if your data require it.

Stage 3: Master flat field

make_flatfield() / go_flatfield()

Sky flat frames (IMAGETYP=flat, OBJECT=SkyFlat) are overscan-subtracted and median-combined per filter. Frames with a mean pixel value above 55,000 ADU are rejected as saturated before combination. The combined flat is normalised and applied to science frames. Output flat-fielded files carry the _flat.fits suffix.

Stage 4: Target directory organisation

make_targetdirectories()

Flat-fielded science files are moved into subdirectories organised by OBJECT header keyword and filter, e.g. NGC891/g-SLOAN/. Object names with spaces are compressed (NGC 891 → NGC891).

Stage 5: Chip extraction

go_extractchips()

Each multi-extension FITS file is split into per-chip single-extension files named <base>_1.fits through <base>_4.fits. These are the inputs to the astrometric pipeline. Chips can be subsetted via lbc_chips.

Stage 6: Astrometric registration (optional)

go_register()

Controlled by do_astrometry=True (the default). Requires SExtractor, SCAMP, and SWarp to be installed.

SExtractor (go_sextractor()): Detects sources and writes a FITS_LDAC catalog (*.cat) for each chip image.

SCAMP (go_scamp()): Matches the catalog against an astrometric reference (default: GAIA-DR3) and writes a .head file containing the WCS solution. SCAMP is run iteratively (default 3 iterations) with progressively tighter tolerance parameters to improve the astrometric fit.

SWarp (go_swarp()): Resamples each chip image onto a common grid and co-adds them into a final mosaic. Output files are named <object>.<filter>.mos.fits with a companion weight map.

Known limitations

  • V-BESSEL shared between cameras: The V-BESSEL filter is available in both LBC-Blue and LBC-Red. If both cameras observed in V-BESSEL on the same night, run the pipeline separately with lbcr=False and lbcb=False to prevent inappropriate co-addition.

  • Unmatched flat fields: If a flat field is present for a filter but no science frame was taken in that filter (or vice versa), the behaviour can be unexpected. Inspect your calibration file inventory before running.

  • Partial-readout test images: Flat or bias frames taken as partial-readout test exposures will cause the pipeline to fail without a helpful error message. Remove these from raw/ before running.

  • Astrometric fit quality: No automated quality assessment of the astrometric solution is currently implemented. Inspect the SCAMP .xml diagnostic files or the astro*.pdf plots to identify chips with poor solutions before running SWarp.

  • Extended objects: Background estimation and astrometric fitting can be unreliable for fields dominated by large extended sources (e.g., nearby galaxies).

Credits

LBCgo is built on code originally developed by David Sands and subsequently extended in scripts made available by Ben Weiner (bjweiner/LBC-reduction). It makes extensive use of the astropy-affiliated package ccdproc and the Astromatic tools by Emmanuel Bertin.