Finding a biological marker in the atmosphere of an exoplanet is a major goal, but as Ignas Snellen argues in the essay below, space-based missions are not the only way to proceed. A professor of astronomy at Leiden University in The Netherlands, Dr. Snellen makes a persuasive case that technologies like high dispersion spectroscopy and high contrast imaging are at their most effective when deployed at large observatories on the ground. A team of European observers he led has already used these techniques to determine the eight-hour rotation rate of Beta Pictoris b. We’ll need carefully conceived space missions to study those parts of the spectrum inaccessible from the ground, but these will find powerful synergies with the next generation of giant Earth telescopes planned for operations in the 2020s.

by Ignas Snellen


While I was deeply involved by my PhD project, studying the active centers of distant galaxies, a real scientific revolution was unfolding in a very different field of astronomy. In the mid-1990s the first planets were found to orbit stars other then our Sun. For several years I managed to ignore it. Not impeded by any knowledge I was happy to join the many skeptics to dismiss the early results. But soon they could be ignored no more. And when the first transiting planet was found and a little later its atmosphere detected, I radically changed research field and threw myself, like many others, on exoplanet research. More than a decade later the revolution is still going strong.


Not all scientific endeavors were successful during this twenty-year period. Starting soon after the first exoplanet discoveries, enormous efforts were put in the design (and getting the political support) for a spacecraft that could detect potential biomarker gases in the atmospheres of nearby planet systems. European astronomers were concentrating on DARWIN. This mission concept was composed of four to five free-flying spacecraft carrying out high-resolution imaging using nulling interferometry, where the starlight from the different telescopes is combined in such way that it cancels out on-axis light, leaving the potential off-axis planet-light intact. After a series of studies over more than a decade, in 2007 the European Space Agency stopped all DARWIN developments – it was too difficult. Over the same time period, several versions of the Terrestrial Planet Finder (TPF) were proposed to NASA, including a nulling interferometer and a coronagraph. The latter uses a smart optical design to strongly reduce the starlight while letting any planet light pass through. Also these projects have subsequently been cancelled. Arguably an even bigger anticlimax was the Space Interferometry Mission (SIM), which was to hunt for Earth-mass planets in the habitable zones of nearby stars using astrometry. After being postponed several times, it was finally cancelled in 2010.

How pessimistic should we be?

Enormous amounts of people’s time and energy were spent on these projects, costing hundreds of millions of dollars and euros. A real pity, considering all the other exciting projects that could have been funded instead. We should set more realistic goals and learn from greatly successful missions such as the NASA Kepler mission, which was conceived and developed during that same period. A key aspect of the adoption of Kepler as a NASA space mission was the demonstration of technological readiness through ground-based experiments (by Bill Borucki and friends). A mission gets approved only if it is thought to be a guaranteed success. It is this aspect that killed Darwin and TPF, and it is this aspect that worries me about new, very smart spacecraft concepts such as the large external occulter for the New World Mission. Maybe I am just not enough of a (Centauri) dreamer.

In any case, lead times of large space missions, as the Kepler story has shown, are huge. This implies that it is highly unlikely that within the next 25 years we will have a space mission that will look for biomarker gases in the atmospheres of Earth-like planets. If I am lucky I will still be alive to see it happen. My idea is – let’s start from the ground!

The ground-based challenge

The first evidence for extraterrestrial life will come from the detection of so-called biomarkers – absorption from gases that are only expected in an exoplanet atmosphere when produced by biological processes. The prime examples of such biomarkers are oxygen and ozone, as seen in the Earth’s atmosphere. Observing these gases in exoplanet atmospheres will not be the ultimate proof of extraterrestrial life, but it will be a first step. These observations require high-precision spectral photometry, which is very challenging to do from the ground. First of all, our atmosphere absorbs and scatters light. This is a particular problem for observations of Earth-like planets, because their spectra will show absorption bands at the same wavelengths as the Earth’s atmosphere. In addition, turbulence in our atmosphere causes the light that enters ground-based telescopes to become distorted. Therefore, light does not form perfect incoming wavefronts, hampering high-precision measurements. Furthermore, when objects are observed for a longer time during a night, their light-path through the Earth atmosphere changes, as does the way starlight enters an instrument, making stability a big issue. These are the main reasons why many exoplanet enthusiasts thought that it would be impossible to ever probe exoplanet atmospheres from the ground.

The technique

Work over the last decade has shown that one particular ground-based technique – high dispersion spectroscopy (HDS) – is very suitable for detecting absorption features in exoplanet atmospheres. The dispersion of a spectrograph is a measure of the ‘spreading’ of different wavelengths into a spectrum of the celestial object. Space telescopes, such as the Hubble Space Telescope (HST), Spitzer, and the future James Webb (JWST) have instruments on board that are capable of low to medium dispersion spectroscopy, where the incoming light can be measured at typically 1/100th to 1/1000th of a wavelength. With HDS, precisions of 1/100,000th of a wavelength are reached – hence about two orders of magnitude higher than from space. For two reasons this can practically only be done from the ground: 1) the physical size of a spectrograph scales with its dispersion, meaning that HDS instruments are generally too big to launch to space. 2) At high dispersion the light is spread very thinly, requiring a lot of photons to do it right, hence a large telescope. For example, the hot Jupiter tau Bootis b required 3 nights on the 8m Very Large Telescope to measure carbon monoxide in its atmosphere. Scaling this to the HST (pretending it would have an HDS instrument) it would have cost on the order of 200 hours of observing time – more than was spent on the Hubble Deep Field. Hence, HDS is the sole domain of ground-based telescopes.

The high dispersion is key to overcome the challenges that arise from observing through the Earth’s atmosphere. At a dispersion of 1/100,000th of a wavelength, HDS measurements are sensitive to Doppler effects due to the orbital motion of the planet. E.g. the Earth moves with nearly 30 km/sec around the Sun, while hot Jupiters have velocities of 150 km/sec or more. This means that during an observation, the radial component of the orbital velocity of a planet can change by tens of km/sec. While this makes absorption features from the planet move in wavelength, any Earth-atmospheric and stellar absorption lines remain stationary. Clever data analysis techniques can filter out all the stationary components of a time-sequence of spectra, while the moving planet signal is preserved. Ultimately, the signal from numerous individual planet lines can be added up together to boost the planet signal using the cross-correlation technique – weighing the contribution from each line by its expected strength.

Screenshot from 2015-09-11 08:59:46

Image: Illustration of the HDS technique, with the moving planet lines in purple.

So why does this work? Although the Earth atmosphere has a profound influence on the observed spectrum, the absorption and scattering processes are well behaved on scales of 1/100,000th of a wavelength and can be calibrated out. The signal of the planet can be preserved, even if variations in the Earth atmospheres are many orders of magnitude larger. In this way starlight reflected off a planet’s atmosphere can be probed, but also a planet’s transmission spectrum – when a planet crosses the face of a star and starlight filters through its atmosphere. In addition, a planet’s direct thermal emission spectrum can be observed. This is particularly powerful in the infrared. And it works well! In the optical, absorption from sodium has been found in the transmission spectra of several exoplanets. In the near-infrared, carbon monoxide and water vapor have been seen in both the transmission spectra as well as thermal emission spectra of several hot Jupiters – on par with the best observations from space. In the next two years new instruments will come online (such as CRIRES+ and ESPRESSO on the VLT) that will take this significantly further – allowing a complete inventory of the spectroscopically active molecules in the upper atmospheres of hot Jupiters, and extending this research to significantly cooler and smaller planets.

One step beyond

There is more. The HDS technique makes no attempt to spatially separate the planet light from that of the much brighter star – it is only filtered out using its spectral features. Hot Jupiters are much too close to their parent stars to be able to see them separately anyway. However, planets in wider orbits can also be directly imaged, using high-contrast imaging (HCI) techniques (also in combination with coronography). This technique is really starting to flourish using modern adaptive optics in which atmospheric turbulence is compensated by fast-moving deformable mirrors. A few dozen planets have already been discovered using HCI, and new imagers like SPHERE on the VLT and GPI on Gemini, which came online last year, hold a great promise. What I am very excited about is that HDS combined with HCI (let’s call it HDS+HCI) can be even more powerful. While HDS is completely dominated by noise from the host star, HCI strongly reduces the starlight at the planet position – increasing the sensitivity of the spectral separation technique used by HDS by orders of magnitude. Last year we showed the power of HDS+HCI by for the first time measuring the spin velocity of an extrasolar planet, showing beta Pictoris b to have a length of day of 8 hours. [For more on this work, see Night and Day on ? Pictoris b].


Image: HDS+HCI observations of beta Pictoris b.

The giants are coming

Both the US and Europe are building a new generation of telescopes that can truly be called giants. The Giant Magellan Telescope (GMT) will consist of six 8.4m mirrors, equivalent of one 24.5m diameter telescope. The Thirty Meter Telescope (TMT) will be as large as the name suggests, while the European Extremely Large Telescope (E-ELT) will be the largest with an effective diameter of 39m. All three projects are in a race with each other and hope to be fully operational in the mid-2020s.

Size is everything in this game – in particular for HDS and HDS+HCI observations. HDS benefits from the number of photons that can be collected, which scales with the diameter squared. Taking into account also other effects, the E-ELT will be >100 times faster than the VLT (in particular using the first-light instrument METIS, and HIRES). This will bring us near the range needed to target molecular oxygen in the atmospheres of Earth-like planets that transit nearby red dwarf stars. We have to be somewhat lucky for such nearby transiting systems to exist, but simulations show that the smaller host star makes the transmission signal of molecular oxygen from an Earth-size planet similar to the carbon monoxide signals we already have detected in hot Jupiter atmospheres – it is just that the systems will be much fainter than tau Bootis requiring the significantly bigger telescopes. The technology is already here, but it is all about collecting enough photons. This could also be solved in a different way if even the ELTs turn out not to be large enough. HDS observations of bright stars do not require precisely shaped mirrors and this could be achieved by arrays of low-precision light collectors, but this is something for the more distant future.


Image: Artist impression of the E-ELT – ready in 2024! (credit: ESO).

Even more promising are the high-contrast imaging capabilities of the future ELTs. Bigger telescopes not only collect more photons, but also see sharper. This makes their capability to see faint planets in the glare of bright stars scale with telescope size up to the fifth power, making the E-ELT more than a 1000 times faster than the VLT. Excitingly, rocky planets in the habitable zones of nearby planets become within reach. Again, simulations show that their thermal emission can be detected around the nearest stars, while HDS+HCI at optical wavelengths can target their reflectance spectra, possibly even including molecular oxygen signatures.

Realistic space missions

Whatever happens with space-based exoplanet astronomy, ground-based telescopes will push their way forward towards characterizing Earth-like planets. This does not mean there is no need for space missions. First of all, I have not done justice to the fantastic, groundbreaking exoplanet science the JWST is going to provide. Secondly, a series of transit missions, TESS from NASA (launch 2017), and CHEOPS and PLATO from ESA (Launch 2018 & 2024), will discover all nearby transiting planet systems, a crucial prerequisite for much of the science discussed here.

Above all, ground-based measurement will not be able to provide a complete picture of a planet’s atmosphere – simply because large parts of the planet’s spectrum are not accessible from the ground. This will mean that the ultimate proof for extraterrestrial life will likely have to come from a space mission type DARWIN or TPF. Imagine how a ground-based detection of say water in an Earth-like atmosphere would open up political possibilities, but the right timing for such missions is of upmost importance. Aiming too high and too early means that lots of time and money will be wasted, at the expense of progress in exoplanet science. It is good to dream, but we should not forget to stay realistic.

Further reading

Snellen et al. (2013) Astrophysical Journal 764, 182: Finding Extraterrestrial Life Using Ground-based High-dispersion Spectroscopy (

Snellen et al. (2014), Nature 509, 63: Fast spin of the young extrasolar planet beta Pictoris b (

Snellen et al. (2015), Astronomy & Astrophysics 576, 59: Combining high-dispersion spectroscopy with high contrast imaging: Probing rocky planets around our nearest neighbors (