Automated Spectroscopic Wavelength Calibration using Dynamic Time Warping

TL;DR

This paper introduces an automated wavelength calibration method for one-dimensional spectra using Dynamic Time Warping, which is robust, flexible, and effective even with differing spectral resolutions and missing features.

Contribution

It presents a novel application of Dynamic Time Warping for wavelength calibration in astronomy, integrated into a Python data reduction toolkit.

Findings

DTW effectively recovers non-linear dispersion solutions.

The method is robust against spectral resolution differences.

It handles spurious or missing spectral features.

Abstract

Here we present an automated method for obtaining wavelength calibrations for one-dimensional spectra, using Dynamic Time Warping (DTW). DTW is a flexible and well-understood algorithm for pattern matching, which has not been widely used in astronomy data analysis. Employing a calibrated template spectrum as a reference, DTW can recover non-linear and even discontinuous dispersion solutions without an initial guess. The algorithm is robust against differing spectral resolution between the template and sample data, and can accommodate some spurious or missing features. We demonstrate the effectiveness of DTW in an automated data reduction workflow, using both simulated and real arc lamp spectra in a Python data reduction framework. Finally, we provide a discussion on the utility and best practices with the DTW algorithm for wavelength calibration. We also introduce the PyKOSMOS data reduction toolkit, which includes our DTW calibration methods.