TESS: keeping a sky survey running and checking its signals

Restoring the observing sequence

TESS resumed scientific observations on May 3, 2024, after a second interruption in April. The Transiting Exoplanet Survey Satellite had launched on April 18, 2018. Its observing programme tracks changes in starlight, including the small dips produced when planets cross their host stars.

The spacecraft entered safe mode on April 8 and returned to observations by NASA’s April 17 update. It entered safe mode again on April 23. The operations team traced that second interruption to unsuccessful momentum unloading from the reaction wheels: the propulsion system had not been successfully repressurized after the earlier event. NASA’s May 7 update confirmed that the team had corrected this problem. The cause of the April 8 event remained under investigation at that point. NASA’s dated operational updates distinguish the two interruptions.

Cameras, cutouts and light curves

TESS uses four cameras to monitor sectors of the sky for about a month at a time. Each sector spans 24 by 96 degrees, and a full-frame image contains 192 million pixels. During the primary mission, those images were recorded every 30 minutes. By the fifth anniversary in April 2023, the full-frame interval had shortened to 200 seconds. These repeated exposures create records of brightness rather than isolated pictures of a planetary surface. The observations also capture stellar flares, dark spots carried across rotating stars, and oscillations driven by internal sound waves.

The accumulated archive already exceeded 251 terabytes of full-frame images in April 2023, according to project scientist Knicole Colón. Christina Hedges, then leading the TESS General Investigator Office, described cutouts around more than 467,000 objects. Such cutouts provide the measurements used to make light curves: brightness plotted against time. Those archive quantities describe the mission at its fifth anniversary; they are separate from later counts of confirmed planets and candidates.

What an interruption leaves in the data

Sector 78 illustrates how operational history becomes part of a scientific dataset. The April 2025 data-release memorandum records 10.72 days of observations lost to safe hold during the 2024 sector. Including downloads, the total pause was 11.14 days, while science data covered 17.63 days. These figures measure different things. No data are taken in spacecraft safe hold. Camera temperatures can change slightly, and images may shift by up to half a pixel for about two days after observations resume; the coordinate solutions incorporate those shifts.

Data products also differ between targets. Very bright targets and difficult background regions can require custom aperture analysis. Quality flags distinguish spacecraft conditions, scattered light and calibration problems. Sector 78’s release notes specify different treatments:

• Failed photometric processing: original and calibrated target pixels may remain available without a derived light curve.
• Safe-mode and Earth-point cadences: timestamps and measurements carry null values.
• Predicted straylight: inspect full-frame images before exclusion, because the prediction can be conservative.

Testing a transit-shaped signal

A dimming event does not automatically qualify as a planet. LEO-Vetter, introduced by Michelle Kunimoto and colleagues in 2025, applies automated tests to both brightness measurements and pixels. It separates instrumental or stellar false alarms from astrophysical false positives such as eclipsing binaries. Alternating eclipse depths can reveal a binary detected at half its actual period. Difference images help establish whether the changing light comes from the target star or another object nearby. A separate test identifies signals concentrated around data gaps rather than treating every repeated dip as equally trustworthy.

The team demonstrated the system on approximately 200,000 M-dwarf light curves, reducing about 20,000 transit-like detections to 172 planet candidates. The reported 91% completeness and 97% reliability against noise or systematic false alarms came from performance tests using simulated data. They are not a claim that 97% of all TESS detections are planets. The candidate catalogue still required additional assessment of astrophysical false-positive probabilities, transit modelling and inspection. The software’s thresholds can be adjusted for different research purposes.

Open data and human review

Processed observations from TESS and the retired Kepler mission are freely available to researchers and the public. Planet Hunters TESS lets volunteers classify light curves online. Exoplanet Watch performs a different task: participants collect or process observations of known planets and submit them to a public archive. The project also supplies robotic-telescope data, allowing people without their own equipment to contribute. Participants can receive credit when their observations are used in a scientific paper. These activities cover both the search for new objects and further measurements of objects already identified.

Transit timing is one practical contribution. Citizen-science observations have helped refine when important targets should be observed, saving time on facilities including the James Webb Space Telescope. Rob Zellem, Exoplanet Watch’s project scientist, describes participation using a smartphone or laptop as accessible to people worldwide. Joshua Pepper, a NASA programme scientist, sees human evaluation as a complement to automated classification of unusual signals. These are distinct roles: software can flag patterns, while volunteers and professional teams examine their meaning.

Why repeated sectors matter

TOI 700 e required an additional year of TESS observations before its weak transit signal could be identified. Announced in January 2023, the planet is about 95% of Earth’s size and orbits every 28 days. Its small, cool host star lies around 100 light-years away in Dorado. Three other planets were already known in the system. TOI 700 d, another Earth-size world, follows a 37-day orbit. The later observations also allowed the team to refine the planets’ estimated sizes, making them about 10% smaller than the original calculations.

Emily Gilbert led the research, and Ben Hord explained that the faint signal needed the additional observing year. Planet e lies in the optimistic habitable zone, which allows for liquid surface water at some point in a planet’s history. Planet d occupies the narrower conservative zone, associated with conditions that could support surface water over much of a planet’s lifetime. These zones describe possible conditions; they do not establish that either planet has water. The team continued follow-up work using space observatories and ground-based facilities. Planet e may also be tidally locked, with one side continually facing its star.

Candidate, confirmed planet, possible atmosphere

TOI-715 b provides another example of the distinction between discovery and physical characterization. The planet, described in January 2024 research, lies about 137 light-years away and is roughly one and a half times Earth’s width. It completes an orbit around its red-dwarf star in about 19 days. Its orbit places it in the conservative habitable zone. A suitable atmosphere and other conditions would still be necessary for liquid water to remain on its surface. The same system also contained a possible second planet, slightly larger than Earth, awaiting confirmation.

Georgina Dransfield of the University of Birmingham led the research team. Confirmation observations involved Gemini-South, Las Cumbres Observatory telescopes, ExTrA, SPECULOOS and TRAPPIST-south. Smaller and cooler red dwarfs allow temperate planets to orbit closer to their stars than planets around hotter stars, giving TESS more frequent opportunities to see transits. The confirmed planet’s suitability for atmospheric investigation with Webb depends on properties including its mass and atmospheric extent. A potentially prominent atmosphere would be easier to examine than one held close to a dense planet’s surface; the discovery alone does not settle those properties.

Following a transit with spectroscopy

Two teams announced Gliese 12 b in May 2024 using TESS observations and other facilities. The planet lies almost 40 light-years away in Pisces and orbits every 12.8 days. Its host star has about 27% of the Sun’s size and 60% of its surface temperature. The planet is approximately Earth-size or slightly smaller, comparable to Venus. It receives about 1.6 times the energy Earth receives from the Sun. The estimated surface temperature of 42 degrees Celsius assumes no atmosphere; an atmosphere was not established by that calculation.

Masayuki Kuzuhara and Akihiko Fukui led one team, while Shishir Dholakia and Larissa Palethorpe led the other. Both teams found no evidence of extreme activity from the host star. They proposed further investigation of whether the planet retains an atmosphere. During a transit, starlight passing through an atmosphere can acquire absorption features from different molecules. Facilities such as Webb can use those features for transmission spectroscopy. That follow-up measurement addresses a different question from TESS’s detection of changing brightness: identifying an orbiting planet does not by itself identify the gases around it.

Stellar eclipses in the same survey

TESS’s light curves also reveal systems whose dimming comes from stars. TIC 290061484, described in October 2024, contains an inner pair orbiting every 1.8 days and a third star circling that pair in 25 days. The previous shortest outer period for this type of system was 33 days, recorded in 1956. The trio lies in Cygnus and is nearly edge-on from Earth’s perspective, allowing the stars to eclipse one another. Its orbital arrangement fits within an area smaller than Mercury’s orbit around the Sun.

Machine learning narrowed the TESS data to interesting dimming patterns, after which citizen scientists helped inspect the candidates. Members of the Visual Survey Group worked with professional researchers on the discovery. Veselin Kostov, who led the study, explained that the geometry allowed measurements of the stars’ orbits, masses, sizes and temperatures. Saul Rappaport described the group’s interest in compact multiple-star systems and unusual pulsating stars. The result belongs to stellar research rather than a new-planet tally, even though it emerged from the same mission and the same broad method of recording brightness over time.

A brightening candidate in 2024

In April 2024, Michelle Kunimoto, William DeRocco, Nolan Smyth and Steve Bryson reported a possible free-floating planet from a different kind of signal. Their preliminary search used 1,286,238 light curves from Sector 61, observed between January 18 and February 12, 2023, at 200-second cadence. They looked for brief brightening consistent with gravitational microlensing, rather than the repeated dimming of a transit. The candidate event was associated with source star TIC-107150013. Its fitted Einstein-crossing time was about 0.074 days, and its inferred lens mass depended on the lens’s distance.

The team removed poor-quality measurements and data near orbit boundaries, where scattered light can produce systematic effects. They also detrended slower stellar variations. An alternative extraction using the TESS-Gaia Light Curve procedure checked contamination from nearby stars against Gaia positions and brightnesses. The signal appeared in both extractions, and the team preferred the alternative curve for its lower scatter. Their interpretation remained a candidate: consistency with a lensing model did not provide an independently measured planet mass and distance. The paper proposed extending the search to more sectors and fainter targets.

Checking the expected event rate

Hongjing Yang and colleagues challenged that interpretation in a May 2024 analysis. Their calculation combined source-star distances and radii, TESS photometric accuracy, finite-source effects and a free-floating-planet mass function from Galactic-bulge microlensing surveys. Under their assumptions, the search strategy was expected to detect only about 0.0018 events in Sector 61 across the planetary mass range. They considered the reported event unlikely to be genuine microlensing, consistent with an alternative stellar-flare interpretation from long-term OGLE observations. This was an attributed research assessment, not another confirmed planetary discovery.

The calculation also explored its limitations. Missing stellar information, simplified velocities and blended light can affect sensitivity estimates. A proposed search using lower per-point signal thresholds and more measurements per event increased the estimated yield, but correlated noise could reduce that gain. Extending the search to fainter stars and the first seven years of TESS data still produced an expectation of roughly one event under the proposed improvements. That expectation is a modelled survey yield, not an observed count.

Keeping the census dated

A sky mosaic released in May 2026 assembled observations from 96 TESS sectors spanning 2018 to September 2025. Its markers separated 679 confirmed planets from 5,165 candidates. The distinction is part of the measurement record: candidates await verification and cannot simply be added to the confirmed total. Black areas identify regions not yet imaged for that mosaic. The image also contains the Milky Way’s bright plane and the Large and Small Magellanic Clouds, rather than showing only individual target stars and their potential planets.

The census belongs to the September 2025 observation endpoint, although NASA published the assembled view the following May. Associate project scientist Rebekah Hounsell described the range of planet sizes found, from Mercury-like worlds to planets larger than Jupiter. Project scientist Allison Youngblood also emphasized uses of the archive beyond planets, including young stellar populations, changing galactic activity and nearby asteroids. The mosaic draws these repeated observations into a broad view of the survey’s coverage. Its dated categories preserve the difference between an object flagged by the survey and one whose planetary status has passed further checks.

A separate microlensing result in 2026

Gaia23bra b, announced in July 2026, is distinct from the 2024 free-floating candidate. It is a planet bound to an orange-dwarf star, with about 1.6 times Jupiter’s mass and an orbital distance comparable to Jupiter’s. Its host has approximately 80% of the Sun’s mass, and the system lies nearly 40,000 light-years away. Gaia first flagged the brightening event in 2023. Researchers subsequently found that archived TESS observations covered the same region with denser time sampling, revealing additional features associated with a planetary companion.

Mallory Harris led the study, with colleagues including Diana Dragomir and Michael Fausnaugh. Microlensing magnifies the light of a background star when a foreground mass passes close to the line of sight; a planet can introduce further structure into that brightening. This differs from the light blocked during a transit. The two techniques provide access to different planetary populations and different physical information. A microlensing alignment is a one-off observing opportunity, so detailed repeat observations of the event are difficult. In this case, TESS’s already-recorded time series supplied information absent from Gaia’s more sparsely sampled observations.