The SOXS Data-Reduction Pipeline

TL;DR

The paper presents the design and implementation of a Python-based data reduction pipeline for the SOXS spectrograph, enabling rapid, calibrated, and quality-checked spectral data processing for transient event studies.

Contribution

It introduces a new automated, Phase 3 compliant data reduction pipeline for SOXS, capable of processing and calibrating optical and NIR spectra within 30 minutes.

Findings

Pipeline achieves calibration and data delivery within 30 minutes.

Generates quality control metrics for instrument health monitoring.

Supports tailored data processing for the user community.

Abstract

The SOXS is a dual-arm spectrograph (UV-VIS & NIR) and AC due to mounted on the ESO 3.6m NTT in La Silla. Designed to simultaneously cover the optical and NIR wavelength range from 350-2050 nm, the instrument will be dedicated to the study of transient and variable events with many Target of Opportunity requests expected. The goal of the SOXS Data Reduction pipeline is to use calibration data to remove all instrument signatures from the SOXS scientific data frames for each of the supported instrument modes, convert this data into physical units and deliver them with their associated error bars to the ESO SAF as Phase 3 compliant science data products, all within 30 minutes. The primary reduced product will be a detrended, wavelength and flux calibrated, telluric corrected 1D spectrum with UV-VIS + NIR arms stitched together. The pipeline will also generate QC metrics to monitor telescope, instrument and detector health. The pipeline is written in Python 3 and has been built with an agile development philosophy that includes adaptive planning and evolutionary development. The pipeline is to be used by the SOXS consortium and the general user community that may want to perform tailored processing of SOXS data. Test driven development has been used throughout the build using `extreme' mock data. We aim for the pipeline to be easy to install and extensively and clearly documented.