How Astronomers Detect Water On Exoplanets

Photo: Felix Mittermeier / Pexels
No telescope has ever photographed a puddle on another planet. Instead, astronomers detect water on exoplanets by reading how starlight changes as it filters through a planet's atmosphere during a transit, a technique called transmission spectroscopy. Water vapor absorbs light at specific infrared wavelengths, and when a planet crosses in front of its star, some of that starlight grazes the planet's upper atmosphere and picks up water's chemical signature before reaching a telescope's detector. The James Webb Space Telescope (JWST) has used this method to confirm water vapor in planets ranging from scorching gas giants to smaller, cooler sub-Neptunes, though a vapor detection is a very different claim from finding liquid water on a surface.
The core problem: exoplanets are too small and dim to photograph directly
Most confirmed exoplanets orbit stars that outshine them by a factor of a billion or more at visible wavelengths. Photographing surface or atmospheric detail on a planet that faint, sitting immediately next to an overwhelmingly bright star, is beyond nearly every instrument in operation today, with the exception of a small number of directly imaged giant planets in unusually wide orbits far from their host stars. For the roughly 5,800 exoplanets confirmed as of 2024 according to NASA's Exoplanet Archive, direct photography of atmospheric chemistry simply is not an option.
That constraint forced astronomers toward an indirect method, and it turns out starlight itself does the work. The technique that solved the problem is the transit method combined with spectroscopy, sometimes just called transmission spectroscopy. When a planet's orbit carries it across the face of its star as seen from Earth, a small fraction of the star's light does not pass by the planet cleanly. Instead, it grazes through the thin outer layers of the planet's atmosphere on its way toward our telescopes, and that grazing light comes away altered.
The alteration is subtle. During a transit, the overall brightness of the star dips by a fraction of a percent, the basic signal used to discover most exoplanets in the first place through missions like NASA's Kepler and TESS. But layered on top of that overall dimming is a second, much fainter signal: the wavelengths of light that got absorbed by whatever gases exist in the planet's atmosphere. Separating that second signal from the first is what lets astronomers identify individual molecules.
How starlight becomes a chemical fingerprint
Think of a prism splitting sunlight into a rainbow. Every element and molecule absorbs light at its own characteristic wavelengths, carving out a pattern of missing colors in that rainbow that functions like a barcode unique to that substance. Water vapor, carbon dioxide, and methane each leave behind their own recognizable pattern of missing wavelengths, one that laboratory spectroscopy on Earth has mapped in exhaustive detail for decades.
To exploit that barcode, astronomers first capture a spectrum of the star by itself, when the planet is safely out of the way. Then they capture a second spectrum during the transit itself, when starlight is passing through the planet's atmosphere on its way to the telescope. Subtracting the first spectrum from the second removes the star's own light and leaves behind the atmospheric fingerprint: a series of small dips at exactly the wavelengths that water vapor, or whatever other molecule is present, is known to absorb.
This is not a single quick measurement. Because the atmospheric signal is a tiny fraction of the overall transit dip, which is itself already a small fraction of the star's total brightness, astronomers typically need to combine data from multiple transits of the same planet to build a spectrum with enough statistical confidence to identify a molecule reliably. A single noisy transit can produce a tentative, low-confidence hint; a robust detection usually requires repeat observations.
Why does this only work during a transit?
During transit, starlight passes directly through the thin outer atmosphere of the planet on its way toward Earth, picking up absorption features from any gases present in that atmosphere. Outside of a transit, astronomers see only the star's own light with no atmospheric filtering involved, so the chemical signal that reveals water vapor disappears from the data entirely.
What does a "detection" actually look like in the data?
A detection is a statistically significant dip in brightness at the exact wavelength water vapor is known to absorb, measured against the noise level of the rest of the spectrum. Astronomers report detection confidence in units of "sigma," where higher sigma values mean lower odds the pattern is a random fluctuation rather than a real signal, and a claim below roughly 3-sigma is generally treated as tentative rather than confirmed.
Why JWST changed what's detectable
The Hubble Space Telescope pioneered exoplanet transmission spectroscopy starting in the 2000s and successfully detected water vapor on large, hot planets, including so-called hot Jupiters orbiting close to their stars. But Hubble's near-infrared instruments were not sensitive enough to reliably pick up the much weaker atmospheric signal produced by smaller, cooler planets, where the transit dip itself is shallower and the atmosphere makes up a proportionally tinier fraction of the planet's size.
JWST changed that. Operating from the Sun-Earth L2 Lagrange point roughly 1.5 million kilometers (930,000 miles) from Earth, a location explained in more detail in how Lagrange points hold space telescopes in place, JWST carries infrared spectrographs sensitive enough to detect atmospheric signals from smaller planets, including sub-Neptunes and, in some cases, rocky planets orbiting small, dim red dwarf stars. One of JWST's first widely publicized results, released in 2022 as part of its Early Release Observations program, was a detailed transmission spectrum of WASP-96b, a hot Saturn-class planet roughly 1,150 light-years away, showing a clear and unambiguous water vapor signature along with evidence of clouds and haze in its atmosphere, observed with the NIRISS instrument's Single Object Slitless Spectroscopy mode (Radica et al., 2023, Monthly Notices of the Royal Astronomical Society).
In 2023, JWST data was also used to identify carbon dioxide and methane in the atmosphere of K2-18b, a sub-Neptune roughly 124 light-years away, building on an earlier 2019 water vapor detection made with Hubble data. The K2-18b findings came from a team that included researchers led by Nikku Madhusudhan of the University of Cambridge, and they represented one of the first robust detections of carbon-bearing molecules in the atmosphere of a temperate, potentially habitable-zone planet smaller than Neptune.
The K2-18b biosignature claim and why it remains unresolved
Part of the 2023 K2-18b announcement included a tentative signal consistent with dimethyl sulfide (DMS), a gas that on Earth is produced almost exclusively by marine phytoplankton and is sometimes treated as a candidate biosignature, a chemical that could indicate biological activity if detected with high confidence on another world. That tentative DMS signal generated significant public attention, but it is a useful case study in how cautious the field has to be with low-confidence data.
Multiple independent teams have since reanalyzed the same JWST observations using different data-reduction pipelines and statistical methods, and the picture has grown considerably less certain. A 2025 reanalysis published in the Astrophysical Journal found the original DMS signal was only around 1-sigma significant, meaning there was more than a 30 percent chance the pattern was simply statistical noise rather than a real molecular signature (Taylor et al., 2025, IOPscience). Separately, a joint analysis combining JWST's NIRISS, NIRSpec, and MIRI instruments concluded there was insufficient evidence for DMS or the related molecule dimethyl disulfide (DMDS) in K2-18b's atmosphere at all (Schmidt et al., 2025, arXiv preprint). Laboratory work has further complicated the biosignature argument: researchers demonstrated in 2024 that DMS can form through purely abiotic, non-biological chemical processes under conditions simulating exoplanet atmospheres, and DMS was separately detected on a comet, a body with no possibility of hosting marine life, undercutting the assumption that DMS detection alone would imply biology.
None of this means K2-18b is uninteresting. Methane has been confirmed in its atmosphere at high statistical confidence in multiple reanalyses, and its water-rich, carbon-bearing atmosphere remains one of the most detailed pictures astronomers have of a sub-Neptune's chemistry. But the DMS episode illustrates precisely why the field treats single, low-sigma detections as starting points for further observation rather than settled conclusions, especially when the interpretation touches something as consequential as a potential sign of life.
What people get wrong about "finding water" on another planet
A detected water vapor signature in a planet's atmosphere is not the same as liquid water on a surface, and it is definitely not evidence of life by itself. K2-18b's water and methane detections describe gases mixed into its atmosphere; whether the planet even has a solid or liquid surface at all remains genuinely unresolved and debated among planetary scientists, since sub-Neptunes may lack anything resembling a true rocky surface the way Earth has one, instead transitioning gradually from thick atmosphere to a high-pressure fluid or supercritical layer beneath.
Confirming actual standing liquid water on the surface of any exoplanet, as opposed to water vapor mixed into an atmosphere, has not been done for any confirmed exoplanet as of 2026. Surface liquid water would require independent lines of evidence, such as detecting specific reflectance patterns consistent with liquid oceans or measuring surface temperature and pressure conditions precisely enough to rule out ice or vapor, none of which current instruments can resolve at interstellar distances for small, rocky planets.
Another common error is treating any single spectroscopic measurement as final proof, an error the K2-18b DMS episode illustrates well. Because raw transit spectra are noisy and the atmospheric retrieval models used to interpret them can produce different results depending on assumptions built into the analysis, credible detections are typically confirmed only after repeat observations and cross-checks using multiple independent analysis pipelines, exactly the process that has kept the DMS claim unresolved rather than accepted.
How transmission spectroscopy compares to other detection methods
Transmission spectroscopy is not the only technique astronomers use to study exoplanet atmospheres, though it is currently the workhorse method for smaller, temperate planets. Direct imaging spectroscopy, by contrast, tries to isolate a planet's own light from its star's glare using coronagraphs or starshades that block the star's light, then splits that isolated planetary light into a spectrum. Each approach has different requirements and works best on a different population of planets.
| Method | What it measures | Best suited for | Key limitation |
|---|---|---|---|
| Transit transmission spectroscopy | Starlight filtered through a planet's atmosphere during transit, revealing absorption features | Planets that transit their star as seen from Earth, especially small and temperate worlds around compact stars | Requires a favorable transit geometry; signal is extremely weak for small planets around large stars |
| Direct imaging spectroscopy | A planet's own reflected or thermal light, isolated from its star with coronagraphs or starshades | Large, young, self-luminous planets in wide orbits far from their star | Cannot yet resolve small, close-in, rocky planets; requires extreme starlight suppression |
| Emission spectroscopy (secondary eclipse) | Starlight minus the combined star-plus-planet light when the planet passes behind its star | Hot planets with strong thermal emission, mapping day-side temperature and composition | Only samples the planet's day side; weak signal for cool, small planets |
| High-resolution cross-correlation spectroscopy | Doppler-shifted molecular lines from ground-based high-resolution spectrographs | Confirming specific molecules on bright, well-studied planets from Earth-based telescopes | Needs very bright targets and sophisticated removal of Earth's own atmospheric signal |
Transmission spectroscopy dominates current water-vapor detections because it works on transiting planets, the largest and best-characterized population astronomers have, while direct imaging remains largely limited to giant planets in wide orbits.
What's the difference between the transit method and direct imaging?
The transit method reads chemical signatures in starlight that has passed through a planet's atmosphere as the planet crosses in front of its star, while direct imaging tries to physically separate a planet's own faint light from its much brighter star using instruments like coronagraphs. Direct imaging is currently limited mostly to large, young, self-luminous planets orbiting far from their stars, where the light-suppression problem is more manageable.
What comes next for water detection on smaller, rocky worlds
Detecting water vapor reliably on Earth-sized, rocky exoplanets remains one of the hardest open problems in exoplanet science, and it is a major driver behind JWST's ongoing observing programs. The most promising near-term targets are rocky planets orbiting small, cool red dwarf stars, the kind of stars discussed in why red dwarf stars matter for life searches, because the size ratio between a small star and an Earth-sized planet produces a proportionally larger, more detectable transit signal than the same planet would produce orbiting a Sun-like star.
Even so, red dwarf planets bring their own complications. These stars are prone to intense flare activity and high-energy radiation that can strip away a planet's atmosphere entirely over billions of years, meaning a rocky planet's atmosphere, if it has one at all, might look nothing like the primordial composition it started with. JWST observations of several nearby rocky planets orbiting red dwarfs, including members of the TRAPPIST-1 system, have so far struggled to detect clear evidence of a substantial atmosphere at all in some cases, underscoring how much harder this measurement gets as planets shrink toward Earth's size.
Future instruments are being designed specifically to push past this limit. Proposed mission concepts, including next-generation space telescopes with enhanced starlight-suppression optics, aim to directly image and spectroscopically characterize Earth-sized rocky planets around Sun-like stars within the coming decades, a capability that would let astronomers search for water and other atmospheric gases without relying on the geometric luck of a transit at all. Ground-based efforts also continue in parallel, including high-resolution spectrographs on large telescopes that use different techniques such as Doppler cross-correlation, complementary to the space-based transit method described in this article, sometimes paired with the kind of long-wavelength observing described in how radio telescopes see the invisible universe for entirely different classes of astronomical signal.
The takeaway
Detecting water on an exoplanet means reading a chemical barcode hidden in starlight that has passed through an atmosphere thousands of trillions of kilometers away, not spotting oceans through a lens. The technique, transmission spectroscopy during planetary transits, depends on instruments sensitive enough to catch faint absorption dips, which is exactly the capability JWST added when Hubble's older instruments hit their limits on WASP-96b, K2-18b, and dozens of other worlds. The K2-18b biosignature saga is a useful reminder that a water or gas detection, however well-publicized, is only as solid as its statistical confidence and its survival through independent reanalysis, and that surface liquid water remains a claim no exoplanet has yet earned.
Read next:
- How Lagrange points hold space telescopes in place
- Why red dwarf stars matter for life searches
- How radio telescopes see the invisible universe
FAQ
Q: How do astronomers actually see water vapor on a planet light-years away?
They analyze starlight that has passed through the planet's atmosphere during a transit, looking for absorption dips at the specific infrared wavelengths, near 1.4 and 1.9 micrometers, that water vapor is known to absorb in laboratory measurements. Multiple transits are usually combined to build enough statistical confidence for a reliable detection.
Q: Does detecting water vapor mean a planet could have life?
No. A water vapor detection describes a gas mixed into the atmosphere, not liquid water on a surface or any sign of biological activity. Claims of potential biosignatures on planets like K2-18b, including a tentative dimethyl sulfide signal, remain scientifically contested after multiple independent reanalyses found the evidence too weak to confirm.
Q: Why is JWST better at this than Hubble?
JWST's infrared spectrographs are sensitive enough to detect atmospheric signals from smaller, cooler planets, including sub-Neptunes and some rocky worlds, where Hubble's older near-infrared instruments could only reliably detect signals from large, hot planets like hot Jupiters. JWST's 2022 observation of WASP-96b demonstrated this leap with a clear, high-confidence water vapor spectrum.
Q: What's the difference between the transit method and direct imaging?
The transit method reads chemical signatures in starlight passing through a planet's atmosphere as it crosses its star, while direct imaging tries to photograph a planet's own light separately from its star using coronagraphs or starshades, a technique currently limited mostly to large, young planets in wide orbits.
Q: Can this method detect water on rocky, Earth-sized planets?
It's becoming possible for some nearby rocky planets orbiting small, dim red dwarf stars, where the star-to-planet size ratio makes the atmospheric signal proportionally stronger, but early JWST attempts on planets like those in the TRAPPIST-1 system have often struggled to confirm a substantial atmosphere at all, showing how far this remains from routine.
Q: Has any exoplanet been confirmed to have liquid surface water?
No. As of 2026, no exoplanet has directly confirmed surface liquid water; only atmospheric water vapor has been detected, and that detection alone cannot establish whether a planet has any surface at all, let alone a liquid one.
Q: What does "sigma" mean when scientists describe a detection?
Sigma is a statistical measure of confidence describing how unlikely it is that an observed pattern is random noise rather than a real signal. Higher sigma values indicate stronger evidence, and claims below roughly 3-sigma, like the original K2-18b dimethyl sulfide signal at around 1-sigma, are generally treated as tentative rather than confirmed.
Follow new articles
Add SciHub101 to your feed reader to see new science and technology stories.
About the Author
SciHub101 is an independent science and technology publication founded and edited by Buğra Sözeri, a software engineer and app developer based in Türkiye. AI tools assist with drafting and research. Source documentation in the existing archive is uneven, and articles are being reviewed and corrected. The Editorial Team byline identifies editorial responsibility; it does not certify specialist or independent human review. Where we compare products, anything built by our own team is labelled as such — see the editorial guidelines and About page for how corrections and disclosures work.
Related Articles



The Exoplanet Atmosphere Revolution: What JWST Has Found So Far
For most of the history of exoplanet science, astronomers could detect planets but not smell them. They could measure orbital periods, estimate masses, and infer temperatures — but the chemical composition of another world's atmosphere was almost entirely out of reach. JWST changed that. In its first two years, it detected carbon dioxide on a gas giant 700 light-years away, found evidence of methane and CO₂ on a possible ocean world 120 light-years out, and ruled out thick atmospheres on the rocky TRAPPIST-1 planets — all using a technique called transmission spectroscopy that is as elegant as it is precise.

JWST's WASP-39b: First Clear CO₂ in an Exoplanet Atmosphere
In 2022, the James Webb Space Telescope returned the first clear, detailed evidence of carbon dioxide in a planet outside the solar system. The same planet later gave up sulfur dioxide made by starlight chemistry, water, carbon monoxide, sodium, and potassium, and showed that its morning and evening edges differ.

How Space Telescopes See the Early Universe
Space telescopes do not look back in time by magic. They collect ancient light stretched by cosmic expansion. Here is how redshift reveals the early universe.

Nearest Exoplanets That Could Host Life
A guide to the nearest potentially habitable exoplanets, from Proxima b to TRAPPIST-1e, what JWST has found, and why 'habitable' rarely means confirmed.