One of the biggest challenges in exoplanet exploration is telling the difference between an environment that is habitable and one that is actually inhabited. Even if astronomers find a rocky planet that may have liquid water on its surface and detect certain gases in its atmosphere, that alone cannot confirm the presence of life. Researchers must determine whether a detected gas comes from biological metabolism or can be explained entirely by processes that do not involve life, such as volcanic activity and atmospheric chemistry.
To examine this problem, a research team led by Samantha Gilbert-Janizek, who conducted the work as a doctoral student at the University of Washington and is now a postdoctoral researcher at the University of California, Riverside, developed a whole-planet coupled model of an Earth without life. The model links heat and material transfer among the core, mantle, crust, ocean, and atmosphere. It starts about 50 million years after Earth's formation and reproduces the planet's evolution up to 5 billion years after formation, roughly 500 million years beyond the present.
The results were published on September 18, 2026, in The Planetary Science Journal (DOI: 10.3847/PSJ/ae9d79). The study, which appeared first as a preprint (arXiv:2602.02267), does not report new observations of an exoplanet. Instead, it uses numerical simulation to test whether an Earth-like planet could maintain a warm, wet climate over a long period without life. The findings could serve as a non-biological point of comparison when future exoplanet surveys look for signs of biological activity.
A coupled model that reproduces 5 billion years of planetary evolution
A planet's atmospheric composition is not determined solely by atmospheric chemistry and weather. Volcanic gases released from the mantle, weathering reactions between the ocean and crust, and changes in the energy received from the host star all influence one another. However, many conventional exoplanet models either calculate the atmosphere alone in a static state or treat interior physics, such as mantle convection, independently. Consistently calculating a planet's long-term interior changes together with its surface climate has been difficult.
The distinguishing feature of this study is that it couples the planetary evolution code VPLanet with Clima, a climate model that calculates the radiative-convective equilibrium of the atmosphere. The geophysical processes from core to crust handled by VPLanet are linked with Clima's atmospheric radiation and convection calculations, and the evolution of the whole planet is tracked in a one-dimensional model. The calculation begins at about 50 million years after formation, when the magma ocean right after Earth's formation is assumed to have solidified, and runs to 5 billion years after formation.
For the basic atmospheric conditions, the team set 1 bar of nitrogen (\mathrm) as the background gas and focused on changes in the greenhouse gases carbon dioxide (\mathrm) and water vapor (\mathrm). The model also incorporates changes in the Sun's luminosity and effective temperature as it evolves, and links outgassing at mid-ocean ridges, carbon cycling through subduction zones, and the fixation of atmospheric \mathrm through silicate weathering of continental crust. This makes it possible to trace how thermal changes in the planet's interior affect volcanic activity and the carbon cycle, and ultimately change surface temperature.
The benchmark used to validate the model was not present-day Earth but pre-industrial Earth. The state at 4.5 billion years of model time was matched to pre-industrial Earth, and the team checked its consistency with the climate and interior state of an era when atmospheric \mathrm was about 280 ppm. Because today's atmosphere is changing rapidly due to human activity such as fossil fuel consumption, the team used the pre-industrial state, which has little human influence, as the reference for evaluating the natural carbon cycle.
However, reproducing pre-industrial Earth does not mean the model can predict unknown exoplanets with the same accuracy. The model includes parameters tuned to the pre-industrial Earth, such as the optical thickness of clouds and the dissolution properties of carbonates. There is also great uncertainty about the atmospheric composition and geological environment of the early Earth, and the model has not been validated as accurately reproducing early conditions.
17 of 19 benchmarks matched within the margin of error
The team compared the results at 4.5 billion years of model time with 19 observational target values set for pre-industrial Earth. Seventeen items fell within the uncertainty range of the targets, and the remaining two differed by only about two standard deviations (), assuming a normal distribution.
The calculated mean surface temperature was 286.9 K, matching the pre-industrial estimate of 286.9 ± 0.1 K. The atmospheric carbon dioxide partial pressure was 28.2 Pa, close to the target of 28.4 ± 0.4 Pa. Heat flow from the upper mantle was calculated at 38 TW, matching the target of 38 ± 3 TW. Thus the model met the set benchmarks not only for surface climate but also for heat transport in the planet's interior.
The two items outside the targets' uncertainty ranges were dissolved carbon dioxide concentration in the ocean and the pH (hydrogen ion exponent). Dissolved \mathrm was 10.8 \mathrm} against a target of 8 ± 2 \mathrm}, a somewhat high value. Ocean surface pH was 8.28 against a target of 8.18 ± 0.05, more alkaline than expected. The paper cites the simplified treatment of calcite precipitation and dissolution in the one-dimensional model as one possible cause of these differences.
| 評価指標（モデル時間45億年） | 産業革命前の地球の目標値 | 生命不在モデルの計算値 | 評価と留意点 |
|---|---|---|---|
| 地表平均温度 | 286.9 \pm 0.1,\mathrm | 286.9,\mathrm | 誤差範囲内で一致。ただし、雲のパラメータは目標温度に合わせて調整されている |
| 大気中の\mathrm分圧 | 28.4 \pm 0.4,\mathrm | 28.2,\mathrm | 誤差範囲内で一致。設定した炭素循環の収支と整合している |
| 上部マントル熱流量 | 38 \pm 3,\mathrm | 38,\mathrm | 誤差範囲内で一致。マントルの熱輸送についても目標値を再現 |
| 海洋溶存\mathrm濃度 | 8 \pm 2,\mathrm} | 10.8,\mathrm} | 目標値より約高い。炭酸カルシウムの沈殿・溶解過程の簡略化が影響した可能性 |
| 海洋表面pH | $8.28$ | 目標値より約高い。溶存無機炭素やカルシウム濃度の扱いに課題が残る |
The team then calculated how a lifeless Earth would look from space, anticipating exoplanet observations by future space telescopes. They used PICASO, a radiative transfer code that calculates the absorption and scattering of light in an atmosphere. Reflected-light spectra from 0.2 to 2.0 \mathrm were generated under conditions of a 90-degree phase angle, 50% cloud-cover weighting, and a spectral resolution of .
| 大気成分 | 生命が存在する地球 | 生命不在モデル | スペクトル上の特徴と解釈 |
|---|---|---|---|
| 水蒸気（\mathrm） | 明確な吸収帯がある | 明確な吸収帯がある | 両者に共通して現れる。大気中の水の存在を示すが、液体の水の直接的な証拠ではない |
| 二酸化炭素（\mathrm） | 明確な吸収帯がある | 明確な吸収帯がある | 両者に共通して現れる。生命がなくても大気中に存在する |
| 酸素（\mathrm） | 顕著な吸収帯がある | 吸収特徴が現れない | 光合成による酸素供給がないため、モデルでは有意な蓄積が起こらない |
| オゾン（\mathrm） | 紫外・可視域に強い吸収がある | 吸収特徴が現れない | 酸素がほとんど存在しないため、オゾンによる吸収も現れない |
| メタン（\mathrm） | 特徴的な吸収帯がある | 吸収特徴が現れない | 生物による大量の放出がなく、モデルでは顕著な吸収を生じる濃度に達しない |
Comparing the synthetic spectra, even a lifeless Earth shows clear absorption bands from water vapor and carbon dioxide. The absorption features of oxygen and ozone seen in today's atmosphere did not appear in the lifeless model. Methane, which is produced in large quantities by microbes and other organisms on present-day Earth, also showed no notable absorption in the lifeless model.
The result reaffirms that detecting water vapor or carbon dioxide alone cannot establish the presence of life. At the same time, it does not mean that a planet where oxygen or methane is not detected can be declared lifeless, because an observable atmospheric composition depends not only on a planet's biological activity but also on geological activity, atmospheric chemistry, and the stellar radiation environment.
The spectral calculation here is a forward calculation, which derives light absorption and scattering from a predetermined atmospheric composition. It is not an inversion analysis that accounts for noise in real observational data and estimates atmospheric composition from it. The results are theoretically useful for considering observation methods for future space telescopes such as NASA's planned Habitable Worlds Observatory (HWO), but they have not been confirmed by actual observations.
Four assumptions behind the lifeless Earth
The finding that liquid water can be maintained on the surface without life depends on several assumptions built into the model. If real planets do not satisfy these conditions, they may evolve differently from the calculated results.
The first assumption is that active plate tectonics continues for more than 4.5 billion years. In Earth's long-term carbon cycle, subduction of oceanic crust carries carbonates into the interior, and volcanic activity returns carbon dioxide to the atmosphere. Combined with silicate weathering at the surface, this cycle regulates greenhouse gas concentrations over the long term.
However, whether subduction zones like today's existed during the Hadean and Archean eons of early Earth remains debated. It has also been suggested that Earth may have been in a "stagnant lid" state, in which plate motion is not active. The model assumes that plate motion continues, and does not handle cases in which plate tectonics stops partway through a planet's history or shifts to another mode of geological activity.
The second assumption is that atmospheric escape into space is not considered. In particular, processes by which an atmosphere is lost through interaction between a planet's magnetic field and the stellar wind are excluded from the calculation. For planets with a mass and magnetic field similar to Earth's, there are conditions under which the effect of atmospheric escape can be approximated as small, but this does not apply to every planet.
For example, planets orbiting close to red dwarfs (M-type stars) may be exposed to intense radiation and flares from their stars. In such environments, water and nitrogen in the atmosphere could be lost, greatly changing the surface environment. This model, which does not include atmospheric escape, cannot be applied as is to planets around red dwarfs.
The third assumption is that abiotic methane production is not treated as a major climate driver. On Earth, non-biological methane production can occur in connection with serpentinization, in which rocks such as peridotite react with water. However, the methane supplied by such processes is thought to be very small compared with the biogenic emissions on present-day Earth. The model therefore limits the main greenhouse gases to \mathrm and \mathrm.
Still, how much of a role abiotic methane played in early Earth's climate remains open to debate. In particular, how much methane and other greenhouse gases contributed to surface warming when the Sun was dimmer than today is an important question for understanding early Earth's climate.
The fourth limitation lies in how the one-dimensional model handles clouds and weathering. One-dimensional climate models cannot directly reproduce atmospheric circulation or the regional distribution of clouds. The team therefore adjusted a parameter representing the reflective properties of clouds so that the model reproduces the temperature of pre-industrial Earth. While this brings the results closer to the real Earth, it also means uncertainty remains in predictions for unknown planets.
In addition, the model that calculates continental weathering rates draws on data from an Earth with plants and soil microbes. There are not enough observational data to compare directly how quickly chemical weathering proceeds on a bare rock surface without life. Even a model that excludes biological activity thus reflects, in some of its parameters, knowledge based on the environment of an Earth with life.
Therefore, what this study shows is a conditional result: a planet with the same mass and initial composition as Earth, on which plate tectonics continues, could maintain liquid water over a long period without life. It does not prove that an Earth without life would necessarily have followed this history.
If observations differ from the model, does that mean life exists?
How could this lifeless-Earth model help future exoplanet exploration? If a space telescope obtains the atmospheric spectrum of an unknown rocky planet, researchers can use the planet's mass, orbit, and the luminosity of its host star to consider what kind of atmosphere would form through non-biological processes alone. This study provides one of the theoretical frameworks needed for such comparisons.
Even if an observed atmospheric composition matched the lifeless model's prediction, it could not be concluded that the planet has no life. The model calculates one example of non-biological planetary evolution and does not cover every possible geological environment or atmospheric composition. Moreover, even if life does exist, it may be indistinguishable from a lifeless planet in distant observations unless its activity is large enough to change the overall atmospheric composition.
Conversely, if observations deviate greatly from the lifeless model's prediction, that alone does not prove the existence of life. Mineral chemical reactions not included in the model, atmospheric escape, or inappropriate initial conditions could also create differences between prediction and observation. Even if a combination of gases that is hard to explain is found in an atmosphere, it must first be carefully tested whether known non-biological processes can account for it.
The important role of such models is not to directly determine the presence of life but to narrow down which planets should be prioritized for follow-up observations. If results that are difficult to explain by geological activity and atmospheric chemistry alone are obtained, the model could provide a clue for investigating in detail whether the cause is an unknown non-biological phenomenon or biological activity. It may also eventually help decide which planets should be allotted the limited observing time of space telescopes.
In the University of Washington press release, co-author Associate Professor Rory Barnes explained that modeling a planet from its core to its atmosphere as a single system raises confidence in future planetary simulations. This, however, is the research team's own assessment and does not mean that the model's predictive accuracy for unknown exoplanets has been demonstrated by observation. No equation that can definitively determine the possibility of life has been established, and it remains difficult to judge the presence of life simply because a particular gas is detected.
The significance of this study is that, rather than directly searching for the features of a planet with life, it examined in detail a planetary environment that can exist without life. Knowing how far geological activity and atmospheric chemistry alone can maintain an Earth-like environment would also clarify the baseline for identifying changes caused by biological activity. The challenge now is whether the model's range of application can be widened to include the cessation of plate tectonics, atmospheric escape, and different stellar environments. Clarifying, one by one, what can happen on a lifeless planet is an important step toward finding traces of life on distant worlds.