Roman’s High-Latitude Time-Domain Survey (HLTDS) provides time-series photometry and slitless spectroscopy at HST-like resolution and sensitivity across tens of deg², obtaining multi-band color information and spectroscopy on a 5-day cadence. Most observations will occur over a two-year window in the middle of Roman’s five-year primary mission.

The HLTDS will monitor fields in both hemispheres, with observations optimized to detect and characterize Type Ia supernovae (SNe Ia) over redshifts from ~0.5 to > 2.5. As one of Roman’s Core Community Surveys, HLTDS is central to meeting the mission’s science requirements: by using SNe Ia to trace the expansion history of the Universe, it will probe the origin of cosmic acceleration and the nature of dark energy. Beyond SNe Ia, HLTDS will chart the transient near-infrared sky, enabling the detection and characterization of rare, high-redshift events, core-collapse SNe, tidal disruption events, super-luminous SNe, kilonovae, pair-instability SNe, active galactic nuclei, dust-extinguished sources, and cool, red objects. The survey will also deliver deep co-added imaging and slitless spectroscopic data, enabling studies of faint and distant galaxy populations.  

The survey was designed through a community process. The Roman Observations Time Allocation Committee's (ROTAC) Final Report (hereafter “ROTAC, 2025”) discusses the rationale for the chosen implementation, expected source yields (see also Rose et al., 2025), and an overview of anticipated science investigations. 

This article summarizes the major scientific drivers and describes the survey implementation in the Astronomer’s Proposal Tool (APT) for Roman.

HLTDS Overview

The HLTDS will monitor one northern and one southern field over the course of Roman's primary five-year mission, with the primary survey observations obtained over a two-year span. In the southern field, spectroscopy will be obtained in wide and deep tiers optimized to characterize SN Ia in two distinct redshift ranges. Similarly, imaging in both the northern and southern fields will be carried out in wide and deep tiers, with filter sets and exposure times optimized for the SN Ia characterization in two different redshift ranges. The wide and deep imaging tiers are optimized to reach a signal-to-noise ratio of 20 at average maximum light for SN Ia at z ~ 0.9 and z ~ 1.7, respectively, and the two tiers will have substantial overlap in the resulting SN Ia redshift distributions. The spectroscopic tiers will provide redshifts and classifications for a subset of the transient sources. The many imaging and spectroscopy epochs will yield deep, coadded datasets with areas approximately 100x and 30x those of the large HST Treasury programs CANDELS (Grogin et al., 2011) and 3D-HST (Brammer et al., 2012), respectively.

The survey is organized into three Components, Core, Pilot, and Extended, executed across the five-year prime mission (see Schematic of HLTDS Component Timing). Each Component sets the overall cadence and sequencing of observations. Within each Component, observation sequences are carried out in four Tiers: Wide Imaging, Deep Imaging, Wide Spectroscopy, and Deep Spectroscopy. The Summary Table of HLTDS Tiers lists the observation specifications for each Tier. 

The HLTDS is allocated 180 days, including exposure times and overheads relating to consecutive slews, filter changes and settle times. The Core Component (88% of the allocation) constitutes the primary survey observations and is planned over two years in the middle of Roman’s five-year mission. The Pilot Component (8%) consists of early observations executed as soon as possible in the mission. The Extended Component (4%) comprises a small number of additional Deep Imaging observations at much longer cadence, scheduled both before and after the primary observations.

Field Selection

The HLTDS fields must lie within Roman's Continuous Viewing Zone to enable a 5-day cadence throughout the two years of the Core Component observations. Field selection also considered: testing cosmic variance; minimizing foreground extinction to maximize transient discovery; leveraging existing and planned contemporaneous multi-wavelength datasets; and ensuring year-round visibility with ground-based facilities. These requirements led to two HLTDS fields, one in the north and one in the south, well separated in right ascension. 

The northern field is ELAIS-N1 (Oliver et al, 2000). 

The southern field is the Euclid Deep Field South (EDFS), which will be observed with both imaging and slitless spectroscopy. 

Properties of both fields are given in Summary Table of Survey Fields, and their locations are shown in the Figure of HLTDS Fields.  A discussion of the full considerations for the field selection is provided in the HLTDS Definition Committee's report (Appendix C.2; ROTAC, 2025).

Roll Angles and Temporal Coverage

For each target within the field of regard, Roman can perform ±15° roll changes around the nominal attitude, imposing strict scheduling limits on allowable position angles, even in continuous viewing zone fields like those targeted by the HLTDS. As the observatory orbits the Sun, the roll angle for HLTDS observations changes by roughly 1° per day, so each epoch has a slightly different position angle. To mitigate position-angle variations, the survey adopts near-circular mosaics that maximize inter-epoch overlap. However, due to the WFI focal-plane geometry, no mosaic can be perfectly circular; therefore, overlap is highest in the mosaic core, which attains full coverage each epoch ("Area with 100% epoch overlap" in the Summary Table of HLTDS Tiers), and tapers toward the exterior, which is revisited only a handful of times per year. As a useful collateral benefit, varying the roll angle at each epoch distributes the field of interest across different regions of the focal plane, reducing the impact of detector artifacts.

To make this concrete, consider the wide imaging tier in the Southern field. Each epoch maps 8.9 deg², but with a 5-day cadence only 6.8 deg² are revisited every epoch; by contrast, about 13 deg² receive coverage at least once over the year. See Figure of Cumulative Area Covered by Different Numbers of Epochs, which shows the pseudo-dithered mosaic (left) and the cumulative area as a function of the number of epochs when revisiting every 5 days for a year (72 total epochs; right). A practical “effective area” for the survey is the region covered by ≥90% of the 5-day-cadence epochs; this value ("Area with 90% epoch overlap") is reported in the Summary Table of HLTDS Tiers.

Northern Field: ELIAS-N1

The northern HLTDS field, ELAIS-N1 (Oliver et al, 2000) will be observed with imaging only. It has extensive coverage across X-ray, the near- and far-ultraviolet, deep optical and infrared imaging, and radio, and is well positioned for imaging with northern ground-based facilities. The wide imaging tier in this field will cover ~10 deg², while the deep imaging tier covers the innermost ~2 deg² of the wide tier. The Figure of HLTDS Fields shows the approximate HLTDS footprint for the Wide Imaging and Deep Imaging Tiers, compared to the multi-wavelength coverage from other observatories. 

Implementation

The Wide Imaging Tier uses 38 pointings arranged in a torus. The Deep Imaging Tier fills the center region with 7 pointings, providing 2.4 deg² per epoch. In this design, the wide-tier mosaic deliberately omits the center, already imaged by the deep tier, so that its pointings can be pushed outward to expand coverage in the exterior regions. Taken together, the wide and deep tiers cover 14 deg² per epoch, with the central area imaged exclusively by the deep tier. The mosaic pattern is shown in the Figure of HLTDS Mosaic Patterns.

Southern Field: EDFS

The southern HLTDS field, the Euclid Deep Field South (EDFS), will be observed with both imaging and slitless spectroscopy. Roman observations in this region will reach approximately half the cumulative depth of the Large Survey Space and Time (LSST) Deep Drilling Fields, providing a powerful combination of cadence and wavelength coverage for a broad range of science investigations.

The EDFS is divided into two components, EDFS_a and EDFS_b, both covered by the southern HLTDS. The wide imaging tier is centered on EDFS_b, covering approximately 7 deg², while the deep imaging and spectroscopic tiers are centered on EDFS_a. Unlike in the northern field, the deep imaging tier (~4 deg²) is not contained within the wide imaging tier but is spatially offset from it. All spectroscopic observations are centered on the EDFS_a field. The wide spectroscopy tier (~4 deg²) largely overlaps the deep imaging area, and the deep spectroscopy tier covers the innermost ~0.5 deg² of that same region. The Figure of HLTDS Fields shows the approximate area covered by the Wide and Deep Imaging and Spectroscopy tiers in the HLTDS, compared to the Euclid and LSST coverage.

Implementation

In the southern field, the wide and deep imaging tiers are implemented as two independent mosaics. The wide imaging mosaic consists of 27 pointings, covering 8.9 deg² per epoch, while the deep imaging mosaic comprises 16 pointings, covering 5.3 deg² per epoch. Both spectroscopic tiers overlap the deep imaging region: the wide spectroscopy mosaic includes the outer 14 pointings, and the deep spectroscopy mosaic consists of the central two, covering 0.5 deg² per epoch. The full spectroscopic footprint covers 5.3 deg² per epoch at differing depths, with the central region observed only by the deep tier. The mosaic patterns are shown in the Figure of HLTDS Mosaic Patterns.

Observational Specifications

The same exposure times are used for each epoch of the the wide and deep tiers, regardless of survey component (Core, Pilot, or Extended), to ensure uniform depth in every filter within each tier. 

To meet the survey’s supernova requirements, exposures shorter than 500 s are not dithered. Simulations showed that adding dithers at these durations would introduce sufficient overhead to reduce areal coverage and cadence, lowering the expected SN yield below requirements (Appendix C.2; ROTAC, 2025). Accordingly, observations with exposure times shorter than 500s use single pointings. For exposure times longer than 500s, dithering becomes favorable because (1) the 20 s slew and settle overhead is less than 4% of the total execution time, and (2) a larger fraction of pixels benefit from cosmic ray rejection, improving the effective depth. For these longer exposures, standard APT dither patterns are used (see the Summary Table of Exposure Time and Depth), and the combined dithered exposures achieve the target depth.

For short exposures (<500 s), where no dithers are used, small offsets are applied when changing filters so the same source does not fall on the same pixels, mitigating the impact of bad pixels and detector artifacts. Offsets follow the same step size and direction as in the the four-point gap-filling dither BOXGAP4_1, but are implemented as equatorial (RA/Dec) target shifts.

With this scheme, a single filter in an epoch may still show chip gaps, but the combined pass across all filters fills those gaps, delivering complete area coverage for the epoch.

The specific selections for each tier, including MA Tables and dither patterns, are given in the subsections below and summarized in the Summary Table of Exposure Time and Depth. 

Core Component

The majority of the HLTDS observations occur over two years in the middle of Roman's five-year primary mission. These data provide the primary discovery and characterization set for most transients and variable sources. Although spectral elements and exposure times are largely optimized for SN Ia discovery and precise measurements of their light curves, the wide and deep filter sets support photometric classification of a broad range of transients.  

Both fields are visited with a 5-day cadence, with the northern and southern schedules decoupled to maximize scheduling flexibility. In the south, both spectroscopic tiers are observed at every epoch. For the imaging tiers, only the bluest filters in each tier is repeated at every epoch: F062 for the wide and F087 for the deep tier. The remaining four filters in each tier are each observed with a 10-day cadence.  However, they are observed in alternating pairs, with one of the pairs observed every 5 days, resulting in color information being obtained every 5 days.  Filter pairs were chosen to (1) ensure full wavelength coverage at each epoch and (2) minimize optical element wheel motion:

• Wide tier pairs: (F106, F158) and (F087, F129)

• Deep tier pairs: (F106, F158) and (F129, F184)

To further reduce optical wheel movement, interleaved wide and deep epochs using similar filters are combined, yielding two four-filter sets that alternate every 5 days:

• F062 (wide), F087 (wide + deep), F129 (wide + deep), F184 (deep)

• F062 (wide), F087 (deep), F106 (wide + deep), F158 (wide + deep)

Pilot Component

Early in the mission, HLTDS will obtain an initial set of observations in each survey field to prepare for the Core Component data. Specifically, eight observations per field, covering all tiers at a ~16-day cadence, will:

• enable the generation of deep template images for difference imaging
• collect reference slitless spectroscopy to support modeling and subtraction of the host galaxy component
• provide an empirical estimate of the SNe Ia rate at z >~ 1 to inform the survey implementation review (see Future Evaluation)

Although the cadence is the same, southern and northern sequences may be fully decoupled to minimize scheduling constraints.

Extended Component

The HLTDS will also extend the temporal baseline of the Deep Imaging Tier beyond the two-year Core Component. A total of eight additional observations, covering both the northern and southern deep imaging area, will provide temporal coverage across the full five-year span of Roman's primary mission. Three of these observations will occur between the Pilot Component and the Core Component observations, spaced approximately 65 ± 5 days apart and concluding 35 days before the start of the Core Component. The remaining five epochs will be obtained after the Core Component ends, beginning 35 days after the final Core Component observations and spaced by 85 ± 10 days to maximize temporal coverage through the remainder of the mission. This extended program will enable the discovery and characterization of long-duration, high-redshift transients, including superluminous and pair-instability SNe at redshifts up to z ~ 5.

Scheduling of the Observations

Pilot observations are recommended as early as possible, ideally within the first year of the mission. However, their timing depends on the first Galactic Bulge season, coronagraph observations, scheduling of HLWAS early observations, and the Cycle 1 General Astrophysics Surveys program selections. These, in turn, depend on Roman's actual launch date and commissioning timeline. Similarly, the start of the Core Component observations depends on the interplay with Roman's other surveys, particularly on the timing of the third Galactic Bulge season (see Galactic Bulge Time-Domain Survey for more details). The Core Component will not begin until after the third high-cadence bulge season has completed. 

For the Core Component, the goal is to revisit each field on a 5-day cadence. The current implementation targets a per-field cadence of 5 ±1 day, with the option to relax to ±2 days if needed to ease scheduling pressure. Regardless of the adopted tolerance, epochs will be planned so that the average cadence during the primary component is as close to 5 days as possible. At present, north and south passes are interleaved, yielding HLTDS epochs every 2.5 days overall, though the passes could be decoupled if uneven interleaving better satisfies scheduling. Within each field, the wide and deep imaging tiers are coupled to minimize slews and reduce filter-wheel rotations.

Although the community-defined surveys are fully specified, the exact scheduling of observations has yet to be finalized. The timing depends on several factors, including the start of Roman’s science operations, the selection of General Astrophysics Surveys, and target choices and timing within the coronagraph instrument allocation. Even so, much is already understood about how Roman observations will be scheduled. A summary is available in the observation plan for the first two years article.

Future Evaluation of the Survey Implementation

Implementation of the HLTDS will be reassessed after Roman's on-orbit performance is characterized, initial survey data are in hand, and as warranted by emerging science results from Roman or other surveys. The HLTDS definition committee has identified astrophysical and technical performance metrics to be evaluated after the Pilot Component is executed. These include an estimate of the high redshift SNe Ia rate, performance of cosmic ray rejection for non-dithered exposures, and the utility of the prism data. These metrics are detailed in the "Technical Checkpoints from Pilot Survey" section in the HLTDS Definition Committee Report (Appendix C.2; ROTAC, 2025). The evaluation will incorporate input from, and discussion with, the science community. Any recommended changes will be reviewed by a committee of community members with broad, HLTDS-relevant expertise and must be approved by the NASA Project Science Office.

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

References

• Core Community Surveys: Overview and Definitions, maintained by NASA GSFC, accessed 19 September 2025.
• "Roman Observations Time Allocation Committee: Final Report and Recommendations", Roman Observations Time Allocation Committee and Core Community Survey Definition Committees, arXiv:2505.10574, 2025. doi:10.48550/arXiv.2505.10574.
• Roman's Core Community Surveys Are Now Defined, Gilbert, STScI Newsletter Volume 42 Issue 01, 2025. 
• The Hourglass Simulation: A Catalog for the Roman High-latitude Time-domain Core Community Survey, Rose et al., arXiv:2506.05161, 2025.
• Brammer et al., 2012 https://ui.adsabs.harvard.edu/abs/2012ApJS..200...13B/abstract
• Grogin et al., 2011 https://ui.adsabs.harvard.edu/abs/2011ApJS..197...35G/abstract