December 12, 2023

The last major release, ULLYSES Data Release 7 (DR7) consists of new high-level science products (HLSPs) for 97 ULLYSES stars observed since DR6 (March 2023) and updated HLSPs for 398 stars included in all previous releases, for a total of 495 ULLYSES stars in the final sample. Updated HLSPs from previous DRs incorporate ULLYSES algorithm updates as well as COS and STIS pipeline and calibration improvements. HLSPs from previous DRs should be replaced with newly downloaded products that incorporate these upgrades.

• DR7 Target Breakdown
• T Tauri Stars
• Changes to the ULLYSES Sample
• Code Updates
• Time Series Spectra
• Targets Requiring Special Calibration
• Web and Catalog Updates
• Future Updates
• DR7 Caveats and Known Issues
• Download Data
• Publications
• stisblazefix Appendix

DR7 Target Breakdown

495 ULLYSES stars:

• 355 massive stars have HST/UV spectra (from the LMC, SMC, and additional low-metallicity galaxies NGC 3109, Sextans A, IC 1613, WLM, Leo-P, Leo-A, and the Magellanic Bridge)
• 155 massive stars also have FUSE spectra
• 86 massive stars have STIS/CCD spectra
• 136 survey T Tauri stars have HST/UV spectra
• 57 stars have LCOGT (Las Cumbres Observatory Global Telescope) imaging observations in the V and i’ bands
• 1 star (SZ-82) has LCOGT imaging observations in the u’, V and i’ bands
• 103 stars have STIS/Optical spectra
• 4 Monitoring T Tauri stars
• All monitoring stars have LCOGT imaging observations in the u’, V, and i’ bands
• All monitoring stars have additional archival COS and/or STIS spectra

9 non-ULLYSES stars:

• These targets are present in STIS long-slit observations of T Tauri stars
• Products for closer companions may be used to disentangle their contributions to the spectra of primary ULLYSES T Tauri stars

Finally, HST/WFC3 drizzled images, obtained by the ULLYSES team, are provided for the low-metallicity galaxies NGC 3109 and Sextans A.

T Tauri Stars

In DR7, new archival data were added for 74 T Tauri stars. 10 of these targets have companion stars that are visible in the STIS target acquisition images. However, because these stars were not part of the core ULLYSES sample, only spectra of the primary targets were extracted. Information regarding the companions and other relevant calibration comments are included below for completeness. Unless otherwise noted, the ULLYSES team did not attempt to identify companions by any catalog designation.

1. TWA-3A: One companion, not located in the slit, approximately 1” from the ULLYSES target. This companion is unresolved in the COS aperture. Only the ULLYSES target is extracted in STIS observations.
2. V-CO-ORI: One companion ( V* CO Ori B) not located in the slit, 2.11” from the ULLYSES target. Only the ULLYSES target is extracted in STIS observations.
3. V-V1070-TAU: One companion, located in the slit, approximately 1” from the ULLYSES target. This companion is unresolved in the COS aperture. Only the ULLYSES target is extracted in STIS observations.
4. V-V1098-TAU: One companion, not located in the slit, approximately 1” from the ULLYSES target. This companion is unresolved in the COS aperture. Only the ULLYSES target is extracted in STIS observations.
5. SZ-102: No known companion, but this source has a bipolar jet, creating an elongated PSF.
6. RECX-1: This source is part of a triple system. The A and B components are separated by 0.2”, while the C component is 9” away. The A+B components are unresolved in both the COS aperture and the STIS slit. The A+B components are extracted together, and are collectively referred to as RECX-1 in both the COS and STIS HLSPs.
7. V-V1154-SCO: One companion, not located in the slit, 0.32” from the ULLYSES target. This companion is unresolved in the COS aperture. Only the ULLYSES target is extracted in STIS observations.
8. V-DF-TAU: This source is part of a binary system, with separation of 0.09”. Both components are unresolved in both the COS aperture and the STIS slit. Both components are extracted together, and are collectively referred to as V-DF-TAU in both the COS and STIS HLSPs.
9. V-V410-TAU: This source is part of a triple system. The A and B components are separated by 0.07”, while the C component is 0.26” away. The A+B+C components are unresolved in both the COS aperture and the STIS slit. The A+B+C components are extracted together, and are collectively referred to as V-V410-TAU in both the COS and STIS HLSPs. An additional source was found 1.25” from the triple system. However, it is not clear whether it is a physical companion or a STIS artifact (possibly the "railroad tracks").
10. V-GX-ORI: One companion, located in the slit, approximately 1.38” from the ULLYSES target. Only the ULLYSES target is extracted in STIS observations.
11. V-VW-CHA: This source is part of a triple system. The A and B components are separated by 0.7”, while the C component is 16.8” away (and not present in ULLYSES observations). The A+B components are unresolved in the COS aperture. The B component is a binary with 0.1” separation, and the A component is suspected to be a spectroscopic binary. All components are T Tauri stars. In some observational orientations, the A and B components are both present and unresolved in the dispersion direction of the STIS slit; in such cases, the A+B components are extracted together and referred to as V-VW-CHA in both the COS and STIS HLSPs. In other observations, the A and B components are separated and resolved in the cross-dispersion direction of the slit; in such cases, the A and B components are extracted separately and referred to as V-VW-CHAA and V-VW-CHAB, respectively.

Changes to the ULLYSES Sample

Target Names

For each target in the expanded ULLYSES sample, the ULLYSES team has tabulated all aliases that are used in each contributing dataset. These aliases are documented in a CSV file found in the ullyses_utils github repository Of particular note are the following columns:

1. target_name_ullyses: The target name used by the ULLYSES team in the TARGNAME header keyword; this is a well-known name for each target.
2. target_name_hlsp: The target name used by the ULLYSES team in the HLSP filenames. This name cannot include spaces or special characters (except for dashes).
3. target_name_simbad: A name for the target that is resolvable by SIMBAD.
4. target_name_ullyses_apt: If the target was observed by the ULLYSES team, this is the target name used in the phase 2 proposal.

Additional columns list all remaining aliases for each target.

In DR7, the target_name_hlsp values were modified significantly, to homogenize the naming convention and to use consistent catalogs for similar targets. Users should be aware that, as a result, HLSP filenames from previous DRs will not match to DR7 filenames exactly. In addition, a mistake in the naming of a previously known companion to the T Tauri Star CVSO-104 was fixed. The previous name was GAIA-DR3-3217634157789741952, which was in fact the Gaia ID for CVSO-104 itself. This was updated to the true name of the companion, HARO-5-64B.

Additions to the ULLYSES Sample

Archival STIS/CCD data for 86 massive stars in the LMC and SMC were added from SNAP program 16230 (PI: D. Massa). All data were taken using the G230LB and G430L gratings. For a small subset of datasets with multiple G430L observations, only one dataset was used in the HLSPs, due to small velocity offsets between observations. All data were inspected by the ULLYSES team, and problematic exposures were excluded from the sample, as documented in the list of rejected datasets in the ullyses_utils github repository. All CCD data were run through the ULLYSES custom CalSTIS pipeline.

COS and STIS data for 74 T Tauri stars were added in DR7, from various archival programs. All STIS/CCD data, and a subset of some STIS/NUV-MAMA data, were run through the ULLYSES custom CalSTIS pipeline.

COS and STIS data for massive stars in the LMC, SMC, and Leo A were added in DR7, from various archival programs.

Code Updates

PyPI Release

The ULLYSES HLSP creation code has been publicly available since DR5 (June 2022), but various improvements are now included in DR7. Of particular note is the release of the python package ullyses on PyPI, as well as supporting data and scripts found in the ullyses_utils package. The ullyses package can be installed following instructions available on the ullyses github repository. Users should take care to search for the exact spelling of “ullyses”, as there are existing packages with similar names.

Algorithm Enhancements

The code to coadd spectra has now been further generalized. Previously, the names of a directory and a grating were supplied at runtime, and all files matching the specified grating were coadded. In addition to this method of invoking the coaddition algorithm, users may now also specify an explicit list of files to be coadded (from multiple directories, if desired) in place of a directory. In addition, a new script to coadd and write products for any COS, STIS, or FUSE data, is now available in the ullyses package.

A new abutment strategy has been adopted for all level 4 HLSPs (with suffix preview_spec.fits). This new strategy was derived from that implemented for the HASP (Hubble Advanced Spectral Products) project. In this method, all gratings have a wavelength range within which the data are considered "good", and a priority. When making a level 4 product, the data from the highest priority grating that is within the specified wavelength range is selected over lower priority gratings. In practice, this is implemented by assigning 2 transition wavelengths to each grating product, one for the shortest wavelength that fits in the fiducial wavelength range, and one for the longest wavelength. The full list of transition wavelengths from each input grating product is then traversed, and at each transition a decision is made as to which grating has the highest priority and will be used until the next transition wavelength. The transitions are abrupt, and the resolution and wavelength spacing will change from grating to grating. While this method works well for most targets, there are a select number of targets where the level 4 products might not be ideal. In these rare cases, we encourage use of the level 3 aspec or level 2 cspec files instead. We emphasize that the level 4 products are intended primarily for visual inspection, and level 2 or 3 products should be used for scientific analysis.

The columns of the provenance FITS extension of each HLSP (typically the 2nd extension) were updated to be ordered by the starting exposure datetime of each contributing spectrum. The provenance extension lists select metadata for each contributing spectrum.

FITS header comments were added for datasets that either required special calibration, or have quality issues that users should be aware of. These can be found in the 1st extension headers of each HLSP, in COMMENT keywords. If nothing special needed to be applied to the data, the header keyword is empty.

Bug Fixes

A previously known bug did not include all SCI extensions of STIS multi-extension FITS files in HLSPs. This bug has been fixed, and only affected data that had STIS REPEATOBS exposures.

The suffix of level 3 HLSPs, which abut coadded spectra of the same instrument and resolution, were renamed from cspec.fits to aspec.fits. cspec.fits is already used for level 2 HLSPs (coadded spectra of a single grating).

Documentation

Final documentation of all datasets used to create HLSPs is now available in a CSV file in the ullyses-utils github repository. A list of all datasets originally considered but ultimately rejected for inclusion in HLSPs, and the rationale, is also available in the same repository. Jupyter notebooks that illustrate the use of the ULLYSES software packages are now available in the ullyses repository. For a description of the coaddition algorithm, see the Data Description page. Contributions and feedback on this project are welcome! Contribution guidelines are included in the repository. If you have questions or concerns regarding the software, please open an issue or contact the HST Help Desk.

Time-series Spectra

In DR6, the ULLYSES team began delivering spectroscopic time-series products for archival targets hat would advance the scientific goals of the project. These “serendipitous” time-series are for T Tauri stars that have been observed over multiple epochs (either by the ULLYSES team, archival programs, or a combination of both) and whose flux varies as a function of time and therefore should not be coadded. For these targets, only exposure-level time-series products are delivered (see the Data Description page). Bundling all observations of variable targets in one file allows for convenient time-domain spectroscopic analysis. Note that due to differing resolutions between spectral gratings, these files are also bundled by grating. These files will have the suffix tss.fits. In DR7, new serendipitous time-series spectral products are created for the following 21 T Tauri stars:

1. CVSO-58
2. CVSO-90
3. CVSO-107
4. CVSO-109
5. EM-LKCA-15
6. HD-163296
7. SZ69
8. V510-ORI
9. V-AA-TAU
10. V-CS-CHA
11. V-DF-TAU
12. V-DG-TAU
13. V-DM-TAU
14. V-DQ-TAU
15. V-RY-TAU
16. V-T-CHA
17. V-T-TAU
18. V-UX-TAU-A
19. V-UZ-TAU-E
20. V-V836-TAU
21. V-VW-CHA

Many of theses datasets have also undergone manual calibration to improve the science spectra.

DR7 includes newly released LCOGT u’, V, and i’ photometry associated with the second epoch of observations for two stars from the T Tauri star monitoring sample: V-BP-TAU and V-GM-AUR. The photometry for V-BP-TAU was obtained between 23 July 2022 and 27 February 2023. The photometry for V-GM-AUR was obtained between 27 July 2022 and 2 March 2023.

Targets Requiring Special Calibration

Before making HLSPs, various corrections are required to obtain fully optimized 1D spectra. These corrections are detailed below. All targets requiring custom calibration will have a level 0 HLSP (manually calibrated 1-D spectra) ending in the suffix spec.fits

FUSE

Some FUSE data have been manually re-processed to account for spectral channel drifting and erroneous background subtraction. The Jupyter notebooks used to perform these custom corrections are available in the `ullyses-utils` repository. Three new targets are included as part of DR7:

1. NGC346-MPG-435
2. AV-238
3. SK-68D80

COS

For ULLYSES observations of the four monitoring T Tauri stars (V-BP-TAU, V-GM-AUR, V-RU-LUP, V-TW-HYA), COS/NUV vignetting is corrected by scaling the flux in vignetted regions to the flux in the same region with a different COS configuration (that does not suffer from similar vignetting). For all other targets, however, COS/NUV observations lack such overlapping spectra. For these targets, the vignetted regions-- defined as the first 200 pixels of each NUV stripe-- are now flagged and discarded. For four targets, the star was not properly centered in the COS aperture, and the resulting spectra were shifted in the dispersion direction. The offsets were calculated and corrected with custom CalCOS processing. This affects the following targets:

1. LMC079-1
2. SZ-10
3. V-GM-AUR
4. V-TW-HYA

STIS

All STIS/CCD, and some first-order STIS/NUV-MAMA, data of T Tauri stars required tailored calibrations. Special calibration steps taken for these observations can include: custom hot pixel identification and flagging, de-fringing for G750L observations, and customized spectral extraction parameters for T Tauri stars and any companions.

For a subset of STIS echelle datasets, the stisblazefix tool was used to empirically correct for misalignments in the blaze function. For more information, see the appendix below.

Although CalSTIS flags the lowest orders of commonly used echelle configurations- orders 86, 66, and 73 for E140M/1425, E230M/2707, and E230M/2415, respectively- as having a poorly determined residual background, the fluxes of ULLYSES sources in these orders are accurately determined. To maximize wavelength coverage, the ULLYSES team manually removes the affected Data Quality (DQ) flag, DQ=2048, for these orders only. After removal, the data are included in all subsequent HLSPs.

Web and Catalog Updates

Preview plots of ULLYSES spectra are now available on the ULLYSES target table webpages ( LMC, SMC, Low-Z, T Tauri stars). These plots use plotly and are fully interactive through a web browser, allowing for zooming, scaling, and toggling individual spectra on and off. A full walkthrough of the interactive features of the plots can be found on the ULLYSES quicklook page. New information was added to the database that powers the ULLYSES catalog, and is now available to users for querying or displaying:

1. Gaia G magnitude
2. Gaia BP magnitude
3. Gaia RP magnitude
4. Radial velocity
5. Intrinsic (B-V) color

Future Updates

HLSPs for VLT observations of the ULLYSES core sample will be released in early 2024. These products are provided by the XShootU and PENELLOPE groups.

DR7 Caveats and Known Issues

• For regions where all contributing datasets have data quality issues, the flux has been set to 0.
• Whenever the FUV flux is very low (e.g., in weak T Tauri continua or the cores of strong interstellar lines), the statistical nature of the background subtraction algorithms used by CalCOS and CalSTIS can produce slightly negative values. When this occurs, these values should be interpreted as being identically 0. For such regions, the eff_exptime column will also be zero, providing a binary data quality estimate for each product.
• COS spectra using G140L/800 incorrectly contain non-zero flux in the region <912A which is poorly calibrated. The COS team is working on the appropriate updates to the corresponding reference files.

Data Description & Download

A full description of the ULLYSES data products and how they are created can be found here. Data may be downloaded from the ULLYSES search form (HLSPs only), the MAST Data Discovery Portal (HLSPs and their contributing data), or directly as a High-Level Science Product collection using the DOI.

Publications

A description of the ULLYSES observations and data products is given in:

Ultraviolet Legacy Library of Young Stars as Essential Standards (ULLYSES): Data Release I, Roman-Duval, J. et al. 2020

For more information on how to cite ULLYSES data, see ULLYSES References.

stisblazefix Appendix

We applied the stisblazefix correction to 461 medium-resolution and 11 high-resolution ULLYSES STIS echelle datasets. This correction empirically adjusts the alignment of the blaze function by minimizing discrepancies in the overlapping flux of neighboring orders. The linear fit to the applied shift sometimes converges on a false solution, so we investigated the signal-to-noise (SNR) regime where stisblazefix gives reliable results. To do so, we parameterized STIS echelle datasets by their SNR and subsampled high-SNR calibration datasets from the MAMA Spectroscopic Sensitivity and Focus Monitor (Proposals 7096, 7657, 7673, 7810, 8067, 8421, 8424, 8810, 8857, 9265, 8915, 8916, 8919, 9628, 10033, 10039, 11860, 12414, 12775, 13145, 13548, 13994, 14429, 14833, 14972, 15381, 15561, 15750, 16352, 16559, 16960, and 17389) to determine the behavior of the algorithm at lower SNR. We selected datasets with nonzero exposure time and a "NORMAL" exposure interruption indicator flag ("sci_expflag" in MAST). Additionally, we excluded OE5CE4010, OE5CE4050, O3ZX050D0, & O3ZX050E0 upon reviewing the corresponding STIS acquisitions, and O4LU01060 due to a detector power reset. This resulted in a test sample of 225 high-resolution and 236 medium-resolution echelle observations of white dwarf flux standards BD+28D4211, BD+75D325, and G191-B2B/G191B2B taken from 1997-09-15 to 2023-09-28 to determine the SNR limits to apply to the ULLYSES sample.

We estimated the average continuum SNR of each echelle X1D dataset by calculating the FLUX/ERROR column ratio, trimming out 100 pixels from both edges of each order, calculating the 75th percentile value for each order, and taking the mean across all orders. This proved to be robust to emission & absorption lines and to the blaze function edges. See Figure 1 for an example SNR estimate.

We randomly subsampled the monitor datasets at levels of [0.5, 1, 2, 5, 10, 25, 50]%, giving a range of continuum flux levels, and hence SNR. The subsampled datasets were processed with CalSTIS and stisblazefix to derive the blaze shift offset at the end orders and the mean across all orders. We differenced these values with those derived from the corresponding full dataset. A more detailed analysis may wish to derive additional data points from more subsamples of the input datasets, especially at lower SNR levels.

When plotted against SNR (see Figures 2 and 3), the scatter in the stisblazefix mean blaze correction increases to ~3 pixel at SNR=10 for H-modes and >2 pixels at SNR=5 for the M-modes. These thresholds also conservatively correspond to a sudden increase in scatter of the calstis-calculated blaze shift value (Δ in the SCI ext "BLZSHIFT" keyword). We applied these SNR thresholds when determining which ULLYSES datasets should be corrected with stisblazefix. Additionally, the stisblazefix diagnostic plots were manually reviewed for obvious bad fits; nothing was rejected. For M-mode data we used an SNR=5 threshold to apply stisblazefix to 461/590 ULLYSES datasets. For H-mode data we used an SNR=10 threshold to apply it to 11/66 ULLYSES datasets. All datasets that have had stisblazefix applied have COMMENT keywords noting so.