This article describes how Roman Space Telescope data releases are planned and scheduled during the lifetime of the mission. Information on the kind of data that will be included in the releases and details on how the data releases will be organized are given.

Introduction

Roman Space Telescope (Roman) Wide Field Instrument (WFI) data products are classified in different levels, according to their processing and calibration stage (for more information, please see the WFI Data Format and Data Levels and Product articles). In this article, we will focus on the Level 2 (L2; calibrated rate images), Level 3 (L3; co-added, re-pixelated images), and Level 4 (L4; extracted, high-level products, e.g., catalogs and 1-D spectra). As soon as the uncalibrated detector ramps are obtained (L1), they are run through the current version of the Roman calibration pipeline, romancal , to create the calibrated L2 images. Therefore, L2 data are available as prompt products from the Barbara A. Mikulski Archive for Space Telescopes (MAST) soon after the observations are completed and downloaded from the telescope. The prompt L3 image products are made available within 5 days of receiving the last L2 input data needed to create them, and the initial L4 products are released 7 days after receiving the input data. Due to the large volume of data that Roman will produce, reprocessed data will be provided on a defined schedule as data release (DR) products and made available to users through the MAST archive. These data products will have uniform processing, i.e., the calibration pipeline version and calibration reference data will use the same version (or context; see CRDS for Reference Files for more information) for all data products in the release. DR products will be made available to users approximately six months after all the necessary observations have been collected. All products associated with a data release will be made available in the archive at the same time.

Cadence of Data Releases

The cadence of DRs will depend on the timing of the observations. The selection of the calibration file and pipeline versions, the configuration of calibration pipeline parameters, and the final triggering of a DR will be managed by a designated working group. Two of the Roman Core Community Surveys (CCS), the Galactic Bulge Time-Domain Survey (GBTDS) and the High-Latitude Wide-Area Survey (HLWAS), have gaps between observing seasons that will be used to schedule their DRs. For example, a DR for GBTDS could occur approximately six months after the completion of the first season’s observations (which span about 70 days). For the High-Latitude Time-Domain Survey (HLTDS) and General Astrophysics Survey (GAS) programs, the cadence will vary and evolve throughout mission operations. DRs for the CCS programs will occur independently; for instance, an HLWAS DR will contain only HLWAS products and will not combine data from multiple CCS programs. GAS programs might be grouped into GAS DRs based on the mission observing timeline and any other relevant considerations.

Products of Data Releases

DRs will provide users with uniformly processed and science-ready data. Reprocessing for a DR may begin at any data level, depending on updates to the calibration pipeline or calibration reference files. While L3 and L4 DR products will be versioned and preserved in the archive, lower-level products, such as L2 images, are currently planned to be overwritten and not versioned with the DR that re-processed them. A strategy to allow users to retrieve the L2 images used to create the L3 and L4 DR products is under development. 

For L3 images, two types of products will be produced: DR-SUBSET and DR-FULL (see L3 Prompt and Data Release (DR) products to be made available in the MAST archive table). DR-SUBSET products correspond to specific elements of the survey activities (e.g., pass, segment; see the APT Program Elements article for more information). For GAS programs, DR-SUBSET will align with passes, while for CCS programs, this mapping is customizable and will be detailed in the DR documentation.

The DR-FULL product, instead, represents the deepest possible stack within the survey footprint, using only data from that survey to ensure uniformity in observation specifications. By design, DR-FULL products combine multiple epochs (if available), whereas DR-SUBSET products preserve time-domain information. In contrast to prompt L3 products, which use native sampling, DR L3 products will be oversampled if the observations provide sufficient sampling of the point spread function (PSF), such as through sub-pixel dithering. All L3 data products will be stored in files according to the Skymap Tessellation.

Note that L3 products will be created using  romancal for the HLWAS, HLTDS, and GAS programs, while GBTDS L3 products will be created using the microlensing science pipeline.

DRs will also include several L4 product types for the HLWAS, HLTDS, and GAS observations in imaging mode:

• PSF-matched multi-band photometric catalogs
• Results from a small set of artificial source tests to quantify completeness and systematics
• A difference-imaging catalog to detect variable sources
• A forced-photometry catalog for time-series analysis
• Segmentation maps for detected sources

Similar to DR L3 products, the GBTDS L4 products will be created by the microlensing science pipeline, and the list of products will differ from the list above. In addition, L4 spectroscopic mode data products will be available. 

Example science cases for data release products

Data Release products will support a wide range of scientific investigations. Some examples of how these products may be used include:

• L4 PSF-matched multi-band photometric catalogs with photometric redshifts will be essential for studies of stellar populations and galaxy evolution;

• DR-FULL images will enable accurate measurements of the shapes and sizes of detected sources, which are critical for high-redshift galaxy studies;

• Segmentation maps and object detections from DR-FULL stacks may be used to create template images and perform time-series analyses on individual L2 exposures;

• Light curves derived from L2 images can support various time-domain and variability studies, such as the detection and analysis of variable stars, active galactic nuclei, and tidal disruption events;

• Single L2 exposures will be particularly valuable for precise astrometric and photometric measurements, such as those needed for proper motion studies.

For additional questions not answered in this article, please contact the Roman Help Desk.