The paper in brief

  • Ground-based spectropolarimetry can provide better knowledge of the atmospheres and surfaces of planets than can conventional spectroscopy.

  • Sterzik et al.1 (page 64) apply the method to Earthshine — sunlight reflected by Earth as seen by the Moon, and then reflected back to Earth.

  • The authors' analysis reveals the contribution to Earthshine of clouds, oceans and even vegetation on Earth's surface.

  • The technique might therefore now be used to detect vegetation and other signatures of life on extrasolar planets.

A long way off

Christoph U. Keller

Sterzik and colleagues1 used the highly sensitive 8.2-metre-diameter Very Large Telescope in Chile to detect life on Earth. This may sound like killing a mosquito with a cannon, but it is really a breakthrough: it is the first time that terrestrial life has been detected using a method that can, in principle, also be used to find life on planets outside our Solar System. But we are still far from detecting life on exoplanets remotely.

By applying spectropolarimetry to Earthshine, Sterzik et al. were able to determine the fraction of clouds, oceans and even vegetation on the sunlit part of Earth as seen from the Moon. Spectropolarimetry is a technique that provides not only the intensity spectrum of a light source (spectroscopy), but also the polarization state of the light (polarimetry). In this case, the authors measured the linear polarization content of Earthshine — that is, the degree to which its electric field vibrates in a fixed plane. By comparing the observed signal with models of the polarization signal that had different fractions of clouds, oceans and vegetation, the authors could select the model that best matched the observations.

Linear-polarization measurements of starlight reflected by exoplanets will soon become a common tool for characterizing other worlds. Both the European SPHERE instrument at the Very Large Telescope2 and the US Gemini Planet Imager3 will provide linear-polarization observations that can extract polarized exoplanet light from the much brighter, but unpolarized, starlight. Although these instruments will detect only large gas planets similar to Jupiter and Saturn (Fig. 1), the polarization signal will contain vital information about their atmospheres.

Figure 1: Saturn's polarization.
figure 1

a, Conventional image of Saturn. The planets' rings are visible as a dark band because the rings were almost edge-on at the time of the observations. b, c, Linear-polarization images of the planet for two polarization orientations. Black and white indicate negative and positive polarization signals; grey corresponds to no polarization signal. Two features stand out in the polarization measurements: the polar regions and a band under the rings. The polar regions are known to have different atmospheric properties from the rest of the planet, but the band under the rings has so far eluded explanation. Sterzik and colleagues' similar analysis1 of linearly polarized light from Earth revealed signatures of life on our planet. (Images were obtained with the Extreme Polarimeter at the 4.2-metre-diameter William Herschel Telescope on La Palma, Canary Islands.)

Full size image

The European Extremely Large Telescope — which, in about 10 years' time, will collect 23 times as much light as the Very Large Telescope — will be able to detect rocky planets possessing liquid water and atmospheres that may allow life to exist. But this huge telescope will probably be unable to obtain the spectropolarimetric measurements of rocky planets that Sterzik and colleagues have obtained for Earth. This is not because the telescope is too small, but because Earth's atmosphere makes it difficult to see a faint planet close to an extremely bright star. To obtain these measurements, a sophisticated space telescope such as the planned New Worlds Mission4 is needed.

In the meantime, we must also measure the degree of circular polarization of light from planets. Light is circularly polarized when its electric field describes a corkscrew motion about its direction of propagation. It is well known that molecules produced by living organisms show only one form of handedness (homochirality). When unpolarized light is reflected by biological materials, homochirality imprints a small amount of circular polarization on the reflected light. For example, chlorophyll produces a particular spectrum in the degree of circular polarization5. Circularly polarized spectra are therefore an even better tool than linearly polarized spectra for remotely detecting extraterrestrial life — assuming that it also favours homochirality.

Sterzik and colleagues1 used a large, ground-based telescope to obtain indirect polarization spectra of Earth on two days. Such observations cannot be obtained continuously, as these large telescopes are heavily oversubscribed. Successful Earthshine observations also require specific positions of the Moon and Sun. It is therefore difficult to obtain much better measurements from the ground than those achieved by the authors. However, all of these problems could be avoided by building a small telescope on the Moon that directly and accurately monitors Earth's linear and circular polarization spectrum at all times. By learning how to extract biosignatures, including those of seasons and different types of vegetation, from such measurements, we would be much better prepared to understand future polarization measurements of exoplanets that may harbour life.

Benchmark data

Daphne M. Stam

Almost 40 years ago, using ground-based polarimetry and a numerical code, researchers6 found that Venus's bright cloud layer consists of tiny droplets of sulphuric acid. This result convincingly demonstrated the strengths of polarimetry of reflected sunlight for characterizing planetary atmospheres. Sterzik and colleagues1 now show that spectropolarimetry of Earthshine provides a benchmark for studying Earth-like exoplanets.

Today, numerical codes similar to the one used to study Venus's atmosphere6 calculate polarization spectra of various types of exoplanet. These spectra are used to optimize the design and observational strategy of instruments on future telescopes with a view to studying exoplanets. To ensure that the codes are predicting realistic spectra, given a model planetary atmosphere, they must be tested against real measurements of planets that have well-known atmospheric compositions and structures, such as Solar System planets.

But validating exoplanet codes against polarization measurements of Solar System planets is far from trivial, because the polarization signal of a planet depends strongly on how the Sun illuminates the planet. From Earth, we can observe the outer planets — Mars through to Neptune — only at times when their Earth-facing hemisphere is almost fully illuminated by the Sun (as with a full Moon). And at this geometry, their polarization signal is, unfortunately, almost zero. Only Venus and Mercury (and the Moon) can be observed from Earth at gibbous (more than half but less than full), quarter or crescent phases, when their polarization signal can be strong. Because exoplanets orbit other stars, they can be observed at phases that are favourable for polarimetry.

For exoplanets with Earth-like atmospheres and surfaces, codes can be tested against polarization spectra of Earth's sky, as measured from the ground, and against polarization measurements from the Earth-observing satellite instrument POLDER. Such local observations cannot, however, capture the rich variation in aerosol and cloud particles, cloud coverage and types of surface that make up a whole-Earth observation, which would be representative of an Earth-like exoplanet observation. Might features such as polarized rainbows, which should be typical signs of liquid-water clouds7, and the polarized glint8 of sunlight on a water surface, be extracted from a planet's spectropolarimetric signal? Or will the mixture of oceans, continents and clouds dilute them beyond detection?

Sterzik and colleagues' spectropolarimetric measurements of Earthshine1 offer an opportunity to test numerical codes for Earth-like exoplanets. The authors compare their measurements against calculated spectra9. Although the agreement in spectral features is good, there is a significant difference between the slopes of the observed and calculated spectra, especially at the redder wavelengths, where the planetary surface and clouds contribute more strongly to the polarization. The difference in slope is not unexpected, because the available calculated spectra9 represent a limited sample of surface types (vegetation and ocean) and include only one cloud type. In addition, the patchiness of the surface and clouds is not taken into account in the numerical code.

Improved spectral calculations should at least include the polarized reflection by the Sahara Desert and the scattering of sunlight by ice clouds. Indeed, the differences between the measured and best-fit calculated spectra emphasize the sensitivity of spectropolarimetry to atmospheric and surface properties, and hence the value of spectropolarimetry for (exo)planet characterization.

Future spectropolarimetry of Earthshine can reveal whether Earth's oceans flash glints when skies are clear. To confirm the presence of rainbows in Earth's polarization signal, Earthshine should be measured across a narrow range of phase angles as seen from the Moon, a difficult feat. Such measurements could be obtained by a spectropolarimeter located on the Moon that would catch the Earthshine directly. In this way, the detrimental and rather unknown contribution of the lunar reflection to the Earthshine signal would also be eliminated.