The Hubble Spectroscopic Legacy Archive (HSLA)

Andrew Fox,, Jason Tumlinson,, and Molly Peeples,


Data archiving is a crucial component of the operation of an astronomical observatory. Archives ensure the legacy of the observatory, and act as multipliers for its science output, by enabling science investigations unrelated to those in the proposals that obtained the data. To maximize their use by the community, archives need to be well populated, easy to use, and designed so that astronomers can quickly determine which targets of a given type have been observed. In this Newsletter article we discuss a new archival resource for spectroscopic data from the Hubble Space Telescope, named the Hubble Spectroscopic Legacy Archive (HSLA), hosted at the Hubble Archive at the Institute.

With no future space ultraviolet instruments currently planned, the data from the UV spectrographs aboard Hubble have an important legacy value beyond their initial science goals. The HSLA is an effort designed to maximize the longevity of the UV spectroscopic data and to accelerate the scientific study of these data. The HSLA is led from the Institute with involvement from several spectroscopists in the community (see Table 1). The HSLA working group was formed following a workshop held at the Institute in November 2012, Enhancing the Legacy of HST Spectroscopy, chaired by Alessandra Aloisi and Stefano Casertano, in which community input was given on the needs for archival products for spectroscopy, and the Institute began planning the tools and products necessary to meet these needs.

Table 1: HSLA Working Group Members

The first release of the HSLA

The first release of the HSLA was issued in February 2016. This release contains uniformly reduced, co-added, and classified spectra for all COS far-ultraviolet (FUV) observations publicly available at that time. It contains over 11,000 individual exposures on 1,414 distinct targets. The data are packaged into “smart archives” according to target type and scientific themes (such as “solar system,” “early type stars,” “white dwarfs,” and “starburst galaxies”) to facilitate the construction of archival samples for new Hubble proposals and for general science usage. This release was timed to provide a resource for astronomers preparing Cycle 24 Hubble proposals with a Phase I deadline of April 8, 2016. The HSLA products are described and available for download at

As well as providing quick access to the raw and reduced data, and tables sorted by target type, the HSLA offers several additional features. A “quick look” capability plots the co-added spectra and allows users to assess data quality (see Figure 1). A demographic section shows the distribution of instrument modes and programs used to observe a particular target. A light curve shows the variation of the source flux with time, allowing for an assessment of time-variability at different wavelengths.

Figure 1: Co-added spectrum of the AGN NGC 5548 generated by the HSLA, showing flux against wavelength for all data obtained using the COS G130M and G160M gratings. The eight colored stripes show the wavelength regions used to measure the signal-to-noise ratio of the data.

How the HSLA co-addition works

One of the key concepts behind the HSLA is combining spectra across exposures, visits, and programs to give a single co-added spectrum per target. This approach is appropriate for many scientific uses, especially for non-time-variable sources that have been observed on multiple occasions. For COS/FUV observations, two medium-resolution gratings are available (G130M and G160M) and one low-resolution grating (G140L). The HSLA presents the medium-resolution and low-resolution products separately; it does not combine across different resolutions. Figure 2 shows the distribution of usage across COS gratings.

Figure 2: Percentage of COS usage broken down by grating, for all data taken since installation in 2009. The medium-resolution FUV gratings G130M and G160M are the most widely used.

The extracted one-dimensional spectra produced by the CalCOS data reduction pipeline (known as x1d.fits files) are taken as inputs to the HSLA. The co-addition code is a python script that passes through several steps, described in full in the documentation on the HSLA website. In brief, for each target the code decides which files to combine based on the file headers, applies wavelength shifts to align interstellar absorption lines, applies a nearest-neighbor algorithm to sample the spectra onto a single wavelength grid without any attempt to interpolate between adjacent pixels, then co-adds the counts from each exposure. This process preserves the statistical independence of the signals in each pixel. The code then subtracts the background level, uses Poisson statistics to calculate the uncertainties on the counts, then converts to flux at the end.

The HSLA going forward

A preliminary version of the HSLA was released at the January 2016 AAS meeting in Kissimmee, Florida. Institute astronomers gave demonstrations on how to use the HSLA (see Figure 3), and distributed USB sticks loaded with all the co-added COS FUV data to astronomers who attended the demonstration. Future releases of the HSLA will contain co-added COS/NUV data and STIS UV data, and will update the COS/FUV data products for any changes to the CalCOS calibration pipeline, such as improvements to the wavelength-scale dispersion solution and changes to the reference files.

Figure 3: Demonstration of the HSLA at the January 2016 AAS meeting in Kissimmee, FL. given by Institute astronomer Molly Peeples.

Questions about the content of the HSLA can be sent to The HSLA makes use of Spectator, an open-source python code written by Molly Peeples, to create its plots of demographics and spectra. Spectator is available on GitHub.