Poster Abstract
P10.37 Josh Veitch-Michaelis (Astrophysics Research Institute)
Theme: Data processing pipelinesRASCAL: Towards automated spectral wavelength calibration
_RASCAL: Towards automated spectral wavelength calibration_Wavelength calibration is a routine and critical part of any spectral workflow. This usually involves matching peaks in an arc lamp spectrum to a catalogue of known emission lines. Existing automated solutions like template matching and cross-correlation (e.g. PypeIt) make assumptions about the repeatability of the calibration setup; simple effects like variation in relative intensity of the emission lines can cause the fit to fail. As a result calibration still involves manual input and may take minutes, even for an experienced user, if the spectrum is particularly crowded. There is growing need for an automated solution that is easily transferable to multiple instruments and robust to system re-configuration e.g. grating position, lamp type.
In order to address this, we have developed RASCAL (RANSAC-Assisted Spectral CALibration): a Python package for semi-automated wavelength calibration for low- and mid-resolution spectographs. RASCAL works fully automatically for low-resolution spectra. For mid-resolution only a coarse estimate of the dispersion and the spectral range is needed; we make no prior assumptions about what lamp was used. RASCAL first identifies plausible sets of matching peak/emission line pairs before finding the best solution using a variant of random sample consensus (RANSAC); calibration typically completes in seconds on a mid-range laptop. We have tested using spectra from existing systems including SPRAT (Liverpool Telescope), ISIS (William Herschel Telescope), DEIMOS (Keck), simulations and commercial spectrometers. RASCAL has been developed for the ASPIRED program, is being released as open-source and is designed for easy integration into existing pipelines.