An Improved Atlas of Full-Scan Spectra from ISO/SWS

TL;DR

This work revisits the ISO/SWS full-scan spectral atlas to remove unphysical artifacts introduced by the original band-joining method. It adopts a damped, multi-anchor least-squares framework that leverages multiple anchor bands, photometric data, and Spitzer IRS spectra to produce a single, continuous spectrum across 2.4–45 μm for 1,036 spectra. The approach improves flux consistency, reduces negative flux issues, and provides new deliverables (two IPAC tables) to the community, with the atlas hosted at IRSA and BC Dataverse. The result is a more reliable resource for stellar and nebular spectroscopy, enabling better cross-calibration with contemporary infrared datasets and facilitating studies of variability and spectral trends. The work also documents QA procedures and the practical challenges of post-mmission calibration, highlighting the balance between data-driven fitting and expert judgment in assembling a large, heterogeneous spectral atlas.

Abstract

We present an atlas of full-scan spectra from the Short-Wavelength Spectrometer (SWS) aboard the Infrared Space Observatory (ISO) after reprocessing and improving an earlier version published 22 years ago. The SWS spectra cover the wavelength range from 2.35 to 45.3 μm. They include scans in 12 separate bands, and we have updated the methods used to combine those bands into a single continuous spectrum. The main improvement comes from applying multiple constraints, including new photometry and spectra from the Infrared Spectrograph (IRS) on the Spitzer Space Telescope that have become available since the release of the original products, and individualized attention to each spectrum, to renormalize the separate bands into a more consistent single spectrum. In particular this removed unphysical negative fluxes that were common in the original data products. The new database, with 1035 reprocessed spectra, will be available to the community at IRSA, which also hosts the original processing.

Figures (8)

• Figure 1: Color-color diagram of the SWS sample using IRAS photometry. Symbol colors indicate the 6 groups.
• Figure 2: Example of the normalization algorithm in the previous processing driving the flux density negative. The original pws data are shown in black, and the previous processing results are in gray. Band 3D was used as the anchor band, shown in blue.
• Figure 3: Example of the normalization algorithm in the previous processing apparently producing excess flux density at low flux levels. The color scheme is the same as for Figure \ref{['fig.negflux']}.
• Figure 4: Example of the low-flux portion of a stellar source. The pws data are shown in black, Spitzer IRS data in red, and the previous results in gray. Wavelength ranges for selected bands are schematically marked with blue bars, with the bands labeled.
• Figure 5: Example of a common case in which the previous algorithm yielded good results, and the least-squares method produces essentially equivalent results. Top: the original pws data; middle: the previous results; bottom: the result of the least-squares fitting. The upper two plots have been vertically offset for clarity. The anchors applied are shown in red. The vertical bars show the band segment boundaries, and the determined corrections are shown for selected bands, scaling factors above the plot (in blue), and offsets for the least-squares fit below the plot (in magenta).