A space telescope's capability is often described in terms of how large and how far it can image individual objects. But that yardstick alone misses the point of the Nancy Grace Roman Space Telescope. Its primary mirror is the same 2.4-meter diameter as Hubble's, yet a single exposure covers more than 100 times the area, and by repeatedly surveying the same patch of sky, Roman captures planets and galaxies as statistical populations. Understanding Roman means looking past mirror size to its optical design—which channels light across a wide focal plane—and to its observing strategy, which compares vast numbers of objects under identical conditions.
NASA plans to launch Roman aboard a Falcon Heavy rocket on August 30, 2026, at 7:26 a.m. Eastern Daylight Time (8:26 p.m. Japan Standard Time that same day). After launch, the telescope will deploy its solar panels and antenna, head toward the second Sun-Earth Lagrange point (L2), and spend 90 days activating and calibrating its instruments. The first new images are expected in early 2027. What begins there isn't a changing of the guard from Hubble and the James Webb Space Telescope (JWST)—it's an observing network that pairs Roman's wide-area candidate-finding with the deep, narrow-field follow-up of existing telescopes.
Same 2.4-Meter Mirror, Yet More Than 100 Times the Sky
Given identical primary mirror diameters, one might expect Roman and Hubble to see similarly. But while mirror diameter strongly affects light-gathering power and diffraction limit, it doesn't by itself determine how much sky fits into a single image. Roman uses a three-mirror anastigmat design—combining an elliptical primary mirror, a hyperbolic secondary mirror, and an elliptical tertiary mirror—that shortens the focal length to roughly a third of Hubble's. This design suppresses multiple types of aberration across a wide field, producing sharp images not at the detector's center but across an annular region.
The 18 detectors making up the Wide Field Instrument (WFI) are arranged in an arc that follows this sharp-imaging zone. In camera terms, Roman is like an infrared wide-angle lens, while Hubble and JWST are closer to telephoto lenses. It helps to think of Roman scanning the sky's landscape for rare targets, then JWST zooming in on those objects to study their spectra and fine structure.
That analogy breaks down, though, when explaining the difference in field of view. Each telescope has a different mirror diameter and observes different wavelengths. Detectors and sensitivity also differ—these aren't simply different zoom settings on the same camera. Even the phrase "100 times Hubble" varies depending on which camera you're comparing against. NASA's current technical documentation states that WFI's 0.281-square-degree field of view is about 200 times larger than that of Hubble's near-infrared camera, WFC3-IR, while general mission descriptions commonly use "at least 100 times."
A 300-Megapixel Camera With Full-Field Spectroscopy
WFI's imaging surface consists of 18 Teledyne H4RG-10 detectors, each with 4096×4096 pixels—roughly 300 million pixels total. Each pixel corresponds to 0.11 arcseconds of sky, and the field of view, excluding gaps between detectors, spans 0.281 square degrees. Roman builds large mosaics by slightly shifting pointing between exposures, filling in the gaps between detectors.
Each pixel's mercury-cadmium-telluride photodiode converts light from part of the visible spectrum through near-infrared into electrical signals. Eight imaging filters cover wavelengths from 0.48 to 2.3 micrometers, letting researchers study objects' colors, redshifts, and light obscured by interstellar dust. The detectors are cooled to 89.5 Kelvin—because if the instrument measuring infrared light is itself warm, its own thermal radiation would overlap with the signal from observed targets. Operating near L2 with the sunshield permanently facing the Sun keeps both power generation and optical/detector temperatures stable.
The same filter wheel that holds the imaging filters also carries a prism and grisms. When inserted into the light path, these disperse light from 0.75 to 1.93 micrometers by wavelength, simultaneously recording spectra for numerous galaxies and stars within the field. Unlike conventional spectrographs that route light through narrow slits for individual targets, this method doesn't require re-targeting each galaxy one by one. This parallelism proves valuable for large-scale measurements of how fast galaxies are receding and which cosmic era they belong to.
Exploiting Roman's survey speed—up to 1,000 times faster than comparable missions—requires stabilizing images across a wide field and continuously tracking sensitivity differences among the 18 detectors. Built on this optical foundation, attitude control and internal calibration must work together with image processing as a single integrated measurement system.
Beyond Transits: A Gravitational-Lensing Census of Planets
Roman's Galactic Bulge Time-Domain Survey will observe the direction of the galactic center at 12.1-minute intervals across six seasons. Allotted roughly 15 months total, the survey is designed to detect more than 1,000 wide-orbit planets through gravitational microlensing. The same images will also capture periodic stellar dimming from transits, with an expected detection of over 100,000 planets via the transit method. Both figures represent pre-launch projections, not confirmed discoveries.
Microlensing relies on light from a distant background star, the mass of an intervening object crossing in front, and a chance alignment between the two and Earth. Because mass bends spacetime, when a foreground star appears to pass near a background star, the background star's light is magnified. If the foreground star hosts a planet, this smooth brightening curve gets a brief deviation. Even a planet unbound to any star—drifting freely—can leave a brief magnification signature if it has mass.
This method complements the transit technique, which tends to favor planets orbiting close to their host stars. Roman extends sensitivity to cold planets near and beyond the habitable zone, ice giants resembling Uranus and Neptune, and free-floating planets. Rather than selecting individual "second Earths" for direct imaging, Roman functions as an instrument that measures which orbital regions host more planets as a statistical population—revealing what kinds of planetary systems are typical across the galaxy.
Microlensing events are transient phenomena that never repeat with the same configuration. Since the same event cannot be re-observed, brief planetary signals must be captured at fine time resolution from the start. The 12.1-minute cadence answers this irreversibility, with wide field of view and time-domain coverage working as a pair.
Carving Out Darkness to Reveal Faint Planets
While microlensing tallies numerous planets through gravitational traces, the Coronagraph Instrument extracts direct light from a small number of nearby planets. The challenge isn't that planets are faint—it's that a bright star sits right next to them. Even when a physical mask blocks the star's image inside the telescope, light diffracts around the mask's edges, and minute mirror imperfections create false point-like images called speckles. A planet's signal gets buried within this noise.
Roman's coronagraph uses precision masks to shape diffracted light and two deformable mirrors to correct the wavefront. A wavefront sensor measures residual starlight, and numerous actuators make microscopic adjustments to the mirror surface. Repeated correction cycles reduce speckles within a targeted detector region, creating a "dark hole"—a region of suppressed brightness. The input is mixed starlight and planetary light; the processing involves light-blocking and active wavefront control; the output is a visible-light image or low-resolution spectrum with starlight suppressed.
Formal performance requirements remain limited. The only mandatory requirement is imaging point sources at a contrast ratio of 10^-7 relative to the star, at separations of 6 to 9 λ/D, within a 10% bandwidth centered on 575 nanometers. Most other wavelengths and spectroscopic/polarimetric modes are classified as best-effort goals. The target is achieving contrast 100 to 1,000 times better than existing facilities, enabling visible-light detection of mature Jupiter-like planets and dust disks in nearby systems. This represents a technology demonstration allotted a total of three months of observing time within the first year and a half—not an instrument designed to detect biosignatures in Earth-like planetary atmospheres.
Still, the experience of operating deformable mirrors and precision masks together in space carries significant value. Running continuous operations from wavefront measurement through post-observation processing, Roman will create dark holes under thermal variations and pointing shifts that ground-based testing cannot fully replicate. It must also maintain that state and subtract residual speckles using reference stars. This operational data will help determine design margins and calibration procedures for the future Habitable Worlds Observatory (HWO), which aims to directly image Earth-like planets.
Three Rulers for Narrowing Down Cosmic Expansion
Roman's largest observing program, the High-Latitude Wide-Area Survey, will conduct multi-color imaging and spectroscopy across a broad region away from the Milky Way's disk. Its goal isn't to "see dark energy" directly—it's to measure how much the universe expanded at each cosmic era and how matter clustering evolved, using multiple rulers with different error characteristics.
The first ruler is Type Ia supernovae. The High-Latitude Time-Domain Survey will repeatedly observe the same region over roughly two years, recording brightness changes over time and spectra. Because Type Ia supernovae can be calibrated to a standard peak luminosity, their apparent brightness yields a luminosity distance; comparing this against redshift traces historical expansion rates. Though allotted only about six months of observing time, spreading visits across the primary mission timeline provides the needed temporal baseline.
The second ruler is baryon acoustic oscillations (BAO) and galaxy clustering. Pressure waves traveling through the early universe's plasma left behind a characteristic distance scale at which galaxies tend to cluster. Measuring how this reference scale appears at different redshifts reconstructs the history of how cosmic expansion stretched distances. WFI's wide-field slitless spectroscopy builds three-dimensional maps of the vast numbers of galaxies within the same sky region.
The third ruler is weak gravitational lensing. The gravity of foreground ordinary matter and dark matter introduces tiny stretching or compression to distant galaxies' images. A single galaxy's image alone can't be distinguished from its true shape. But by examining vast numbers of galaxies and statistically analyzing correlations in shape distortion between nearby angular positions, researchers can estimate the distribution of matter and structure growth at different cosmic eras.
Not all shape correlations stem from gravitational lensing, though. Galaxy formation environments also correlate with galaxies' intrinsic orientation and shape. This intrinsic alignment signal mimics weak lensing signals, and miscorrecting for it can bias cosmological estimates. Even if the expansion history matches predictions, if structure growth—after subtracting these systematic errors—doesn't match expectations, this discrepancy could reveal evolving dark energy or modifications to gravitational theory.
The reason for using three methods isn't simply to accumulate more data points. Supernova calibration, galaxy redshift measurement, and the point-spread function estimation needed for weak lensing each carry different systematic errors. Cross-checking results within Roman's own data, and further comparing against wide-field visible-light imagery from Euclid and the Vera C. Rubin Observatory, helps isolate errors specific to particular wavelengths or detectors. What Roman ultimately delivers isn't a snapshot of dark energy—it's the allowed range for the cosmological constant, any time-varying component, and possible modifications to general relativity.
20 Petabytes of Public Data and Discovery Infrastructure
Roman's three core surveys will consume no more than 75% of the observing time during the five-year primary mission. Researchers can propose separate wide-area observations for most of the remaining time. The first approved program, the Galactic Plane Survey, plans to map up to 20 billion stars using a total of 29 days distributed across the first two years. Questions that would require too much observing time with a narrow field of view become feasible under Roman's wide-angle design—without straining the overall mission schedule.
Processed data is expected to reach 20 petabytes over five years, with no proprietary period on the scientific data. Rather than copying the entire massive dataset to their own facilities, researchers will bring their computations to calibrated images and catalogs within the Roman Research Nexus cloud environment. The center of gravity shifts from the traditional model—where teams whose proposals get approved claim results first—toward a model where multiple teams simultaneously verify the same public dataset.
Publishing data doesn't automatically create equal participation, though. Handling 20 petabytes requires code, statistical expertise, and calibration knowledge, along with tracking sensitivity differences across detectors and variations in observing conditions. For weak lensing specifically, the question becomes whether researchers can preserve subtle shape distortions without corrupting them, while managing systematic errors including intrinsic alignment. Roman's scientific performance depends not only on mirror and detector specifications, but also on who can verify the archive, cloud computing, and analysis software—and to what extent.
Even so, the flow from wide-area discovery to detailed follow-up observation will change fundamentally. Roman will identify rare high-redshift galaxies, transient objects, and planet candidates; JWST and Hubble will follow up with narrower fields of view and different wavelengths. Euclid and Rubin will supplement information across even wider sky regions or in visible light, cross-checking measurement errors for the same objects. Roman doesn't replace other telescopes—it supplies each with targets worth pursuing.
On the path from the coronagraph toward HWO, distinguishing what Roman can demonstrate from what remains beyond reach matters. Roman targets nearby Jupiter-like planets, while HWO concepts aim to directly image and measure atmospheres of Earth-like planets orbiting Sun-like stars. If Roman succeeds in extracting planetary light, it will demonstrate that active wavefront control can operate in space. Whether this reaches the point of detecting biosignatures depends on whether even higher stability and light-blocking performance can be designed and verified, building on these results.
What matters most immediately after launch isn't a dramatic tally of discoveries—it's whether, during the 90-day commissioning period, image quality and calibration hold up across the wide detector surface, whether the coronagraph achieves its planned dark hole, and whether the massive data-processing pipeline functions end to end. Once these three elements come together, Roman transforms from a telescope that photographs wide areas of sky into a discovery infrastructure that continuously generates both new observation targets and design requirements for next-generation instruments.