≡ Menu

Extracting Exoplanet Topography from Transit Data

How do we go from seeing an exoplanet as a dip on a light curve or even a single pixel on an image to a richly textured world, with oceans, continents and, perhaps, life? We’ve got a long way to go in this effort, but we’re already having success at studying exoplanet atmospheres, with the real prospect of delving into planets as small as the Earth around nearby red dwarfs in the near future. Atmospheric detection and analysis can help us in the search for biosignatures.

But I was surprised when reading a recent paper to realize just how many proposals are out there to analyze planetary surfaces pending the development of next-generation technologies. Back in 2010, for example, I wrote about Tyler Robinson (University of Washington), who was working on how we might detect the glint of exo-oceans (see Light Off Distant Oceans for more on Robinson’s work). And Robinson’s ideas are joined by numerous other approaches. I won’t go into detail on any of these, but l do want to illustrate the range of possibilities here:

  • Exomoon detection (see Sartoretti & Schneider, 1999, or the Hunt for Exomoons with Kepler and papers from David Kipping);
  • Planetary oblateness — i.e., having an equatorial diameter greater than the distance between poles (Seager & Hui 2002; Carter & Winn 2010);
  • Light from alien cities (Loeb & Turner 2012);
  • Plant pigments (Berdyugina et al 2016);
  • Industrial pollution (Lin, Gonzalez Abad, & Loeb 2014);
  • Circumplanetary rings (Arnold & Schneider 2006).

I’ve pulled this list with references out of a paper suggesting yet another target, the surface topography of exoplanets. The work of graduate student Moiya McTier and David Kipping (Columbia University), the paper points out that while many of these effects are beyond the reach of current equipment, they are nonetheless valuable in pushing the limits of exoplanet characterization and helping us understand what technologies we will need going forward.

So is it really possible to detect surface features like mountains, trenches and craters on a distant exoplanet? McTier and Kipping make the case that we can draw conclusions about a planetary surface through what they call its ‘bumpiness,’ which should show up in a planetary transit as a scattering in the light curve produced as its silhouette gradually changes (assuming, of course, that we are dealing with a rotating planet in transit). We would obtain not the image of a specific mountain or other surface feature but a general analysis of overall topography.

The paper’s method is to model planetary transits for known bodies — the Earth, the Moon, Mars, Venus, Mercury — to see what it would take to tease out such a signature. We have ample elevation data for rocky planets in our Solar System. Using this information, we can model what would happen if one of them transited a nearby white dwarf. The researchers used thse values to find a general relationship between bumpiness and transit depth scatter.

In terms of bumpiness, the paper argues:

…the definition should encode the planet’s radius. An Everest-sized mountain on an otherwise featureless Mercury provides more contrast to the average planet radius than an Everest on an otherwise featureless Earth, and should result in a higher bumpiness value.

What we are after here is what the paper calls “an assessment of global average features,” one that incorporates the largest feature on a planet (an enormous mountain, for example) but also includes the contribution to the lightcurve scatter produced by all the planet’s features.

Mars, because of its small size and low surface gravity, turns out to be the bumpiest of these planets. A Mars-sized planet orbiting a white dwarf in its habitable zone proves to be an optimal situation for detecting bumpiness. Why white dwarfs? We learn that even huge ground-based telescopes planned for future decades such as the Extremely Large Telescope and Colossus would be unable to detect bumpiness on planets around stars like the Sun or M-dwarfs because of astrophysical noise and the limitations of the instruments. False positives through pulsations on the star’s surface, for example, can likewise appear as extra scatter in the light curve.

White dwarfs, on the other hand, appear unlikely to have convective star spots, but even if they do occur, McTier and Kipping argue that they can be detected and filtered out. Orbiting moons could similarly cause variations in the transit depth that could be mistaken for topography, but here the signature of the exomoons shows up just outside the ingress and egress points in the transit curve, unlike the topographical signature, which appears only in the in-transit data.

It turns out that the largest of our next generation of big telescopes would be able to work with a white dwarf planet, which the paper models as orbiting at 0.01 AU in the center of the star’s habitable zone. If we assume a mass typical of such stars (0.6 M), we get an orbital period of just over 11 hours. 20 hours of observing time covering some 400 transits with a telescope like the 74-meter combined aperture of Colossus should be able to detect topography.

Image: This is Figure 8 from the paper, showing an oceanless Earth transiting a white dwarf. The caption: Top: Transits of a dry Earth with features (in red) and an idealized spherical Earth (in black) in front of a .01R white dwarf with noise of 20 ppm added (20σ detection). The exaggerated silhouettes of Earth at different rotational phases are shown in brown. Middle: Zoomed-in frame of the bottom of the light curve in the top panel. Bottom: Residual plot showing the difference between the realistic and idealized transits. Grey shadows show the error bars on the residuals equal to 50 ppm. Dashed lines are to illustrate that residuals deviate from 0ppm only inside the transit. Credit: McTier & Kipping.

Surface features should tell us a good deal about a planet’s composition. From the paper:

…a detection of bumpiness could lead to constraints on a planet’s internal processes. Mountain ranges like the Himalayas on Earth form from the movement and collision of tectonic plates (Allen 2008). Large volcanoes like Olympic Mons on Mars form from the uninterrupted buildup of lava from internal heating sources. A high-bumpiness planet is likely to have such internal processes, with the highest bumpiness values resulting from a combination of low surface gravity, volcanism, and a lack of tectonic plate movement. Truly low-bumpiness planets are less likely to have these internal processes. On such planets, surface features are likely caused by external factors like asteroid bombardment.

I like the phrase the authors use in closing the paper, referring to their mission “of adding texture to worlds outside our own.” Texture indeed, for we are beginning to move into the realm of deeper planetary analysis, like a painter gradually applying detail to the roughest of sketches. Because of the magnitude of the challenge, we are coming at the question of exoplanet characterization from numerous different directions, as the list at the beginning of this post suggests. Synergies between their methods will be key to exoplanet surface discoveries.

The paper is McTier and Kipping, “Finding Mountains with Molehills: The Detectability of Exotopography,” accepted at Monthly Notices of the Royal Astronomical Society (preprint).


Comments on this entry are closed.

  • Gary Wilson March 7, 2018, 13:13

    As well as doing control analyses of our own rocky planets and the moon, could the authors have a look at the very best extrasolar planet transits we have available already? Would it be possible to tease out any real data?

  • Tom Mazanec March 7, 2018, 13:39

    Since white dwarfs evolve from red giants, would a white dwarf have a planet at 0.01 AU?

    • andy March 9, 2018, 14:48

      There are several possible ways this could occur. A pre-existing planet might get scattered into a highly-eccentric, star-grazing orbit and get tidally circularised close to the star (a variant of this might have led to the asteroid debris around WD 1145+017). As the star loses mass during the late stages of its evolution, the orbits of any planets would expand, which may lead to system instabilities that make such a scenario more likely to occur. It may also be possible for a sufficiently-massive gas giant to survive being engulfed and end up on a close-in orbit.

      Another scenario is that a white dwarf acquires a new protoplanetary disc from matter lost by a companion star (a possible example would be Mira B), though there is the risk that the white dwarf would undergo nova explosions due to the accreted matter which may disrupt planet formation.

  • Harry R Ray March 7, 2018, 14:47

    Luger et al propose to do EXACTLY THE SAME THING using the same telescope(s) as McTier would, but ONLY in the TRAPPIST-1 system by observing planet-planet transits while NEITHER OF THE TWO PLANETS ARE TRANSITING THEIR STAR, but ONLY if such events occur. I have proposed in an earlier comment, that if TRAPPIST-1d were to transit either TRAPPIST-1b or TRAPPIST-1c WHILE BOTH PLANETS WERE ALSO TRANSITING TRAPPIST-1, ginormus storms could be detected on TRAPPIST-1d (assuming that b and c are not oblate)WITH CURRENT TELESCOPES! I wonder if Seager/Hui or Carter/Winn mentioned anything in their papers about a planet being oblate on ONLY ONE SIDE. If TRAPPIST-1d were to have ALL of the following properties: ONE: An atmosphere with a surface pressure similar to Earth’s. TWO: A temperature of 120 degrees farenheit at the terminator. THREE: Stean produced from a global ocean boiling at the stellar point. And, finally; FOUR: A powerful east -to-west(or west-to-east)jet stream, steam may be FORCED OVER hundred kilometer high HYPERCANES at the terminator, INSTANTLY FREEZE in the upper stratosphere, and the ice condensates plummet to the ocean surface the same way rocks and ash do after they have been expelled into the lower stratosphere by volcanoes. This would make TRAPPIST-1d look like it were oblate ON ONLY ONE SIDE. A dedicated search program to observe TRAPPIST-1d ONLY when it is transiting WITH either TRAPPIST-1b or TRAPPIST-1c AND NO OTHER PLANETS, might pick this up, if we get INCREDIBLY LUCKY and get a planet-planet transit that has ALL of TRAPPIST-1d INSIDE b or c!

    • Michael Fidler March 8, 2018, 8:37

      Remember when this was done with Pluto and Charon to map the albedo features on Pluto?

      • Harry R Ray March 9, 2018, 10:34

        And those albedo features matched up PERFECTLY with now well known topological features on the dwarf planet. This proves that if you are VERY CAREFUL, you CAN break through the “noise” issues mentioned below and get accurate data.

        • Aanta March 9, 2018, 13:36

          Sorry, that is not correct.
          For Pluto and Charon the scientists knew exactly the positions.
          More importantly they did look toward know dark space with known stars.
          For exoplanet the signal is much weaker, and faint background objects might not be known.
          Much stronger by several magnitudes is the primary star with activity of several kinds that change the amount of light.
          So my belief is firm it cannot be done with current instruments, I would be all happy to be proven wrong.

          • Michael C. Fidler March 10, 2018, 2:19

            Binocular telescope can null out the star, that was before AO so Hubble did it, now much better scopes and new technics . Flare from star can be used as a flashbulb to light up both planets in a echo of original flare and timing of transits as in Trappist 1 system is well understood. A good UV flare on the side of the star facing away from earth and you have one hell of a contrast ratio, might see reflections off oceans and details in colors and albedo, maybe even reflections from cities!!!

            • Michael C. Fidler March 10, 2018, 2:49

              DWARF PLANET.

              I can still remember when the first spotted Pluto’s moon Charon in 1978, it was big new!

              Good article on history of Pluto and the observations of surface features, could give the young one’s some idea as to how quickly the technology has improved and how much faster it is still changing!!!


  • Thomas Hair March 7, 2018, 14:50

    This kind of pseudo-analysis seems little better than throwing chicken bones on the floor and trying to discern the will of the gods. Until we have much better instruments people are going to see what they want to see, just ask Giovanni Schiaparelli.

    • Andrei March 7, 2018, 18:33

      I tend to agree, any features will be lost in noise.
      The study of the planets themselves are pushing the tech quite far.
      But all hope is not lost. A planet with a very large albedo difference might be possible to detect.

    • J. Jason Wentworth March 8, 2018, 17:23

      *Nods* I disagree with the contention by McTier and Kipping that we couldn’t be fooled by exomoons (particularly where “long inter-transit interval” exoplanets are concerned). If a planet with a small moon–especially a small, irregular, asteroidal one similar to Phobos or Deimos–transited its star just as its moon was protruding a little beyond the planet’s limb (from our distant point of view), it would “look” just like a mountain.

  • Alex Tolley March 7, 2018, 16:25

    It is extraordinary to realize that just 2 decades ago transits were hard. More recently, exomoons were possible. With the upcoming instruments, under certain conditions, even the “bumpiness” of the surface topography is possible to measure.

    I wonder whether a similar approach can be taken from the reflected light outside the transit. In that situation, the varying albedo of continents and ocean might be detectable, adding geographic information to the topography’s “bumpiness”.

  • Geoffrey Hillend March 7, 2018, 17:12

    The light curve might also include a change in the spectra especially if one gets an angle when the oceans are reflecting or there is more land mass reflecting. I am hoping at some point telescopes will become powerful enough to differentiate the exoplanet spectra of plants, sea and land. Infra red spectroscopy can do this but the signal has to be large and bright enough. These would change quickly due to a fast rotation, so we should be able to at least get some of their spectra if it was too difficult to detect bumpiness which would be supported by the light compared to a non bumpy part of the planet.

  • Robin Datta March 7, 2018, 17:46

    Amazing and commendable how the faintest of additional inormation is squeezed and teased out of what to most would apper as data that has been exploited to its maximum.

  • Charlie March 8, 2018, 13:17

    look, this is pushing detection into unverifiable limits …

  • ljk March 8, 2018, 18:22

    JHU Performs First Lab Simulation of Exoplanet Atmospheric Chemistry

    March 7, 2018

    Scientists have conducted the first lab experiments on haze formation in simulated exoplanet atmospheres, an important step for understanding upcoming observations of planets outside the solar system with the James Webb Space Telescope.

    The simulations are necessary to establish models of the atmospheres of far-distant worlds, models that can be used to look for signs of life outside the solar system. Results of the studies appeared this week in Nature Astronomy.

    “One of the reasons why we’re starting to do this work is to understand if having a haze layer on these planets would make them more or less habitable,” said the paper’s lead author, Sarah Hörst, assistant professor of Earth and planetary sciences at the Johns Hopkins University.

    Full article here:


  • Harry R Ray March 9, 2018, 10:39

    We will soon be adding “Clarke Exobelts” to the “Exomoon detection…Planetary oblateness…Light from alien cities…Plant pigments…Industrial pollution…Circumplanetary rings” list posted above when the ArXiv preprint is published in a journal.