The morphological diversity_of_comet_67_p_churyumov_gerasimenko

COMETARY SCIENCE

The morphological diversity ofcomet 67P/Churyumov-GerasimenkoNicolas Thomas,1* Holger Sierks,2 Cesare Barbieri,3 Philippe L. Lamy,4 Rafael Rodrigo,5

Hans Rickman,6 Detlef Koschny,7 Horst Uwe Keller,8,22 Jessica Agarwal,2

Michael F. A'Hearn,9 Francesco Angrilli,10 Anne-Therese Auger,11 M. Antonella Barucci,12

Jean-Loup Bertaux,13 Ivano Bertini,14 Sebastien Besse,7 Dennis Bodewits,9

Gabriele Cremonese,15 Vania Da Deppo,16 Björn Davidsson,17 Mariolino De Cecco,18

Stefano Debei,10 Mohamed Ramy El-Maarry,1 Francesca Ferri,14 Sonia Fornasier,12

Marco Fulle,19 Lorenza Giacomini,20 Olivier Groussin,11 Pedro J. Gutierrez,21

Carsten Güttler,2 Stubbe F. Hviid,22,2 Wing-Huen Ip,23 Laurent Jorda,24 Jörg Knollenberg,22

J.-Rainer Kramm,2 Ekkehard Kührt,22 Michael Küppers,25 Fiorangela La Forgia,3

Luisa M. Lara,21 Monica Lazzarin,3 Josè J. Lopez Moreno,21 Sara Magrin,3

Simone Marchi,26 Francesco Marzari,3 Matteo Massironi,20,14 Harald Michalik,27

Richard Moissl,25 Stefano Mottola,22 Giampiero Naletto,28,14,16 Nilda Oklay,2

Maurizio Pajola,14 Antoine Pommerol,1 Frank Preusker,22 Lola Sabau,29 Frank Scholten,22

Colin Snodgrass,30,2 Cecilia Tubiana,2 Jean-Baptiste Vincent,2 Klaus-Peter Wenzel7

Images of comet 67P/Churyumov-Gerasimenko acquired by the OSIRIS (Optical, Spectroscopicand Infrared Remote Imaging System) imaging system onboard the European SpaceAgency’s Rosetta spacecraft at scales of better than 0.8 meter per pixel show a wide varietyof different structures and textures. The data show the importance of airfall, surfacedust transport, mass wasting, and insolation weathering for cometary surface evolution, andthey offer some support for subsurface fluidization models and mass loss through theejection of large chunks of material.

The European Space Agency’s Rosetta space-craft entered a close orbit about the Jupiterfamily comet 67P/Churyumov-Gerasimenko(hereafter 67P) on 6 August 2014 at 3.60astronomical units (AU) from the Sun. It

carries OSIRIS (Optical, Spectroscopic and In-frared Remote Imaging System) (1), which hasacquired images of the surface at scales of <0.8meter per pixel in the subsequent 2 months,thereby providing large improvements in reso-lution, coverage, and data quality over previousflyby missions of cometary nuclei (2–4). Theconcept of cometary nuclei as rather uniform,pristine, protoplanetesimals that may have beensubjected to collisional processing is persistent,despite evidence of regional differences on indi-vidual objects such as comet 9P/Tempel 1 (4) andcomet 103P/Hartley 2 (5) andmajor differencesbetween objects (6). The OSIRIS observationshave revealed an irregular-shaped, processed nu-cleus surface with morphologically diverse units.Here we describe the basic morphology of thesurface and some of the structural heterogeneityevident in the highest-resolution data acquireduntil September 2014,when the cometwas 3.27AUfrom the Sun.Combining the orbit of Rosetta with OSIRIS

imaging sequences, we derived shape models ofalmost 100% of the illuminated area (~70% ofthe surface) using different reconstruction tech-niques.We used amodel obtained using a stereo-photogrammetric (SPG) approach (7–9). FigureS1 displays different shaded relief views of aglobal SPG-based shape model of ~5 to 7 m spa-

tial resolution and is discussed in detail in (10).The remaining 30% of the surface (the unillu-minated southern polar region) will receive thehighest solar fluxes around perihelion.

Regional-scale terrains

Nineteen regions have beendefined on thenucleusat this time (Fig. 1) (11). It is expected that otherswill be mapped once the Sun crosses the comet’sequator and illuminates the southernhemispherecompletely. The regions can be grouped into fivebasic categories: dust-covered terrains, brittlematerials with pits and circular structures, large-scale depressions, smooth terrains, and exposedconsolidated surfaces (table S1).

Dust-covered terrains

The Ma’at and Ash regions (both blue in Fig. 1)appear to be coated with a smooth covering ofdust (12). The thickness is uncertain and variable.The source of the dust is activity that produceslow-velocity non-escaping dust particles in aform of airfall. Upward-facing surfaces in sev-eral other areas (such as Seth, Fig. 2, right, andHathor, Fig. 4) also show smooth deposits, whichare inferred to have the same source(s). The coat-ing is not sufficiently thick to completely maskthe large-scale structure of the substrate. The sur-face of 67P is almost devoid of recognizable im-pact craters. One example, whichmay be partiallyburied by airfall, leading to an estimated thick-ness for the airfall of 1 to 5 m using standardscaling laws, has been identified (fig. S2).We ex-pect that any ice chunks or particles larger than a

fewmillimeters that are emitted from the sourcewill not sublimate before falling back and couldtherefore form visually bright, highly localized icedeposits and sources of gas on these surfaces. Theimportance of low-velocity airfall deposition offluffy particles in reducing the thermal inertia shouldbe considered in future thermalmodels. Given thelikelihood of a high organic content for the airfalldeposit (e.g., cometary tholin) and exposure tosunlight, the production of a form of organic,possibly polymerized, matrix from this deposit(after the loss of any volatiles) is also conceivable.High-resolution data show evidence of dune-

like structures (fig. S3) that may be the result ofaeolian-driven surface transport of the dust (13).Images acquired in the Hapi region of aeolian

SCIENCE sciencemag.org 23 JANUARY 2015 • VOL 347 ISSUE 6220 aaa0440-1

1Physikalisches Institut, Sidlerstrasse 5, University ofBern, 3012, Bern, Switzerland. 2Max-Planck-Institut fürSonnensystemforschung, Justus-von-Liebig-Weg, 3, 37077,Göttingen, Germany. 3Department of Physics and Astronomy,University of Padova, Vicolo dell'Osservatorio 3, 35122 Padova,Italy. 4Laboratoire d'Astrophysique de Marseille, UMR 7326,CNRS, and Aix Marseille Université, 38 Rue Frédéric Joliot-Curie,13388 Marseille Cedex 13, France. 5International Space ScienceInstitute, Hallerstraße 6, 3012 Bern, Switzerland, and Centrode Astrobiología, Consejo Superior de InvestigacionesCientificas - Instituto Nacional de Tecnica Aeroespacial, 28850Torrejón de Ardoz, Madrid, Spain. 6Department of Physics andAstronomy, Uppsala University, Box 516, SE-75120 Uppsala,Sweden, and Polish Academy of Sciences Space ResearchCenter, Bartycka 18A, PL-00716 Warszawa, Poland. 7ScientificSupport Office, European Space Agency, 2201, Noordwijk,Netherlands. 8Institute for Geophysics and ExtraterrestrialPhysics, Technische Universität Braunschweig, 38106,Braunschweig, Germany. 9Department of Astronomy, Universityof Maryland, College Park, MD 20742-2421, USA. 10Departmentof Industrial Engineering, University of Padova, via Venezia 1,35131 Padova, Italy. 11Aix Marseille Université, CNRS, Laboratoired’Astrophysique de Marseille, UMR 7326, 38 rue Frédéric Joliot-Curie, 13388 Marseille, France. 12Laboratoire d’Etudes Spatialeset d’Instrumentation en Astrophysique, Observatoirede Paris, CNRS, Université Paris 06, Université Paris-Diderot,5 Place J. Janssen, 92195, Meudon, France. 13LaboratoireAtmosphères, Milieux, Observations Spatiales, CNRS/UVSQ/IPSL, 11 Boulevard d'Alembert, 78280, Guyancourt, France.14Centro di Ateneo di Studi ed Attivitá Spaziali, “GiuseppeColombo” (CISAS), University of Padova, Via Venezia 15, 35131Padova, Italy. 15Istituto Nazionale di Astrofisica, OsservatorioAstronomico di Padova, Vicolo dell'Osservatorio 5, 35122Padova, Italy. 16CNR-IFN UOS Padova LUXOR, Via Trasea, 7,35131 Padova, Italy. 17Department of Physics and Astronomy,Uppsala University, 75120 Uppsala, Sweden. 18UNITN, Universitádi Trento, Via Mesiano, 77, 38100 Trento, Italy. 19INAF,Osservatorio Astronomico, Via Tiepolo 11, 34014 Trieste, Italy.20Dipartimento di Geoscienze, University of Padova, via G.Gradenigo 6, 35131 Padova, Italy. 21Instituto de Astrofísicade Andalucía (CSIC), c/ Glorieta de la Astronomía s/n,18008 Granada, Spain. 22Deutsches Zentrum für Luft- undRaumfahrt (DLR), Institut für Planetenforschung, Asteroidenund Kometen, Rutherfordstraße 2, 12489 Berlin, Germany.23National Central University, Graduate Institute of Astronomy,300 Chung-Da Road, Chung-Li 32054, Taiwan. 24Laboratoired'Astrophysique de Marseille, 38 Rue de Frédéric Joliot-Curie,13388 Marseille Cedex 13, France. 25Science OperationsDepartment, European Space Astronomy Centre/EuropeanSpace Agency, Post Office Box 78, 28691 Villanueva de laCanada, Madrid, Spain. 26Solar System Exploration ResearchVirtual Institute, Southwest Research Institute, 1050 WalnutStreet, Suite 300, Boulder, CO 80302, USA. 27Institut fürDatentechnik und Kommunikationsnetze der TechnischeUniversität Braunschweig, Hans-Sommer-Strasse 66, 38106Braunschweig, Germany. 28University of Padova, Department ofInformation Engineering, Via Gradenigo 6/B, 35131 Padova, Italy.29Instituto Nacional de Tecnica Aeroespacial, Carretera deAjalvir, 28850 Torrejon de Ardoz (Madrid), Spain. 30Planetaryand Space Sciences, Department of Physical Sciences, TheOpen University, Milton Keynes MK7 6AA, UK.*Corresponding author. E-mail: [email protected]

ripple structures, rocks with wind tails (fig. S4),and rocks withmoats provide further support forlocalized gas-driven transport. To initiate saltation,the surface shear stress from the gas expansionof a vent must overcome the gravitational forceand interparticle forces (14). Although the gas

densities are low, following (15), velocities on theorder of 300 m/s appear to be sufficient to move100-mm particles and may be generated by local-ized sublimation. Electrostatic levitation in com-bination with horizontal electric fields across theterminator, as proposed for the Moon and 433

Eros, may be an alternative mechanism, althoughthismay only be effective for smaller particles (16).

Brittle material

The Seth region and interfaces between Ma’atand Ash and other units (such as Imhotep) show

aaa0440-2 23 JANUARY 2015 • VOL 347 ISSUE 6220 sciencemag.org SCIENCE

Fig. 1. Regional definitions based on large-scale unit boundaries. The nomenclature for the regions used here is also given. Images were acquired byOSIRIS (top left, image no. WAC_2014-09-05T02.29.12; bottom left, WAC_2014-09-05T06.29.13; right, NAC_2014-08-05T23.19.14).

Fig. 2. (Left) The Ash-Seth boundary shows evidence of collapse with talus at the base (positions A and B). Fracturing to produce crevices (C) is alsoobserved, with deflation of the surface (D and E) appearing as a possible precursor (NAC_2014-08-07T18.37.34.552Z_ID30_1397549700_F22). (Right) Obliqueview of a circular depression in Seth, which appears to have been partially eroded. The face (F) appears rather uniform in texture, with some evidence of linearfeatures. Note also the evidence of surface deflation (G) and that the horizontal slope appears dust-covered, whereas the vertical face appears “clean”(NAC_2014-09-02T21.44.22.575Z_ID10_1397549800_F22).

CATCHING A COMET

evidence of a morphologically distinct, consoli-dated brittle material (17). There is evidence inmany places of fracturing and collapse of thismaterial as if it is being undercut by masswasting of a stratum below (Fig. 2, left). Wherecollapse or disruption of the material has oc-curred, debris is frequently observed.We infer that this brittle material (18) is

present everywhere underneath the airfall de-posits in Ma’at and Ash. The evidence of col-lapse combined with the low gravity reinforcesthe concept of cometary material having lowtensile strength. Simple mechanics has beenused to estimate the strength necessary to main-tain some overhangs, resulting in values locallyof <20 Pa (19).The Seth region is dominated by a series of

circular, flat-floored, steep-walled depressions.In many cases, the Hapi facing wall is absent(Fig. 2, right). In some examples, the flat floorshold debris, presumably from collapse of thewalls of the depressions. The flat floors, whichare all almost orthogonal to the local gravityvector, all show evidence of the dust depositionseen in Ash and Ma’at, whereas the sides thatare roughly parallel to the potential gradientare relatively clean. The largest example (Fig. 2,right) is 650 m in diameter and (if originallycircular) has lost around one-third of its form,although there is little evidence of talus at itsbase, suggesting that the cut surface is composedof more-consolidated material. This surface alsoshows some faint horizontal and vertical linea-tion. There is some similarity to the circular de-pressions seen on 81P/Wild 2 (3), suggesting acommon production mechanism such as subli-mation or the collapse of internal voids. However,the quasicircular nature requires an organizationthat is not necessarily expected in either mecha-

nism. The Seth region also contains a pit chain(10). Similar pits are seen in the Ma’at region.

Large-scale depression structures

Hatmehit is a circular depression, 1.0 km in di-ameter, on the head of the nucleus. The structureis rather shallow and hence very unlike the bowl-shaped impact structures commonly seen onother bodies. This may not rule out a collisionorigin, but the details of the creation of the rimand of the final surface distribution of fluidizedejecta require modeling in a parameter regimerarely explored numerically up to this point (20).The head and the body of the nucleus have

one large irregular-shaped depression struc-ture each. The one on the body (Aten) is moredistinctive than that on the head (Nut). Aten issurrounded by the brittle material of Ash, whichis coated to a large extent by airfall. However,Aten shows no evidence of a similar dust coating.Furthermore, there is no evidence of the brittlematerial within Aten, with the interior being cov-ered with boulders. At the edges of the depres-sion, disrupted consolidated brittle material isseen. We suggest that Aten has been formed byone or more major mass loss events removingairfall and the mantling brittle material. It is im-portant to recognize that the volume of the Atendepression as a whole is large (~0.12 km3 fromthe shape model). We estimate that 67P can losearound 3 to 5 × 109 kg per orbit in its currentorbit via sublimation. Assuming the density tobe 440 kg/m3 (10), the production of Aten via asublimation mechanism alone would have re-quired ~10 to 20 orbits at this production rate.

Smooth terrains

There are three areas characterized by extreme-ly smooth material with no obvious impacts or

circular flat-floored depressions and a paucityof boulders (Imhotep, Anubis, and Hapi). TheImhotep region is morphologically remark-able. It is dominated by a smooth surface thatcovers an area of >0.7 km2 (Fig. 3). At its mar-gins, the smooth material gives the impressionthat it is layered. Unlike the airfall deposits, thesmoothmaterial appears to be enclosed bymore-consolidated material that surrounds it [con-solidated cometary material (CCM) (21)]. Thesmooth material thins to one side and gives wayto a terrain dominated by circular filled and un-filled structures. The filling material appears sim-ilar to the smooth material (Fig. 3, position D).On comet 9P/Tempel 1, the presence of pits ex-pressed primarily as topographic and albedorings with probable fill on the inside—a fewhaving distinct flat floors—has been noted (22).The terrain on 67P has some similarities, and wenote that a subsurface fluidization productionmechanism has been proposed for 9P/Tempel 1(23). However, the sharp-sided nature of the ex-ternal surfaces of these circular structures maynot be consistent with this mechanism in its sim-plest form, and other mechanisms [e.g., for-mations such as sun cups or dirt cones (24)]should be considered. Imhotep is the only re-gion on the illuminated surface with these typesof structures.The smooth material of Imhotep is surrounded

on three sides by fractured CCM, which has beensubject to mass wasting with debris produc-tion. However, to the south, there is a pronouncedraised circular structure, 650 m in diameter,which also contains smooth material that ispartially eroded. The external surface of the cir-cular structure is irregularly fractured. Smoothterrain is seen at several topographic levels inthe region (25). TheAnubis region appears similar

SCIENCE sciencemag.org 23 JANUARY 2015 • VOL 347 ISSUE 6220 aaa0440-3

Fig. 3. Part of the Imhotepregion.The area shows smoothterrain (A), layering of this materialat its margin (B), smooth materialon topographically higher surfaces(C), circular structures possiblyrelated to those seen on 9P/Tempel 1(D), and layered consolidatedmaterial (E) rising toward the650–m-diameter raised semicircularstructure (G). Fracturing of theconsolidated material is evidentthroughout [e.g., position (F)](NAC_2014-09-05T06.31.16.575Z_ID10_1397549600_F22).

to Imhotep, but no circular structures are evi-dent in the early data.The Hapi region in the neck was the first

unit to become visibly active (emitting dust) inOSIRIS data. The surface reflectance propertiesat visible wavelengths are mostly dominated bydust rather than the high (blue) reflectances ex-pected of ices. (We discuss high-reflectivity parti-

cle “clusters” below.) The region is at least 2.2 kmlong, roughly 0.8 km wide, and forms a partialring of ~140° around the nucleus, leading to itsnecklike appearance. A row of boulders is presentalong part of Hapi’s length (Fig. 4, bottom right);strong dust emission from the vicinity of theseboulders has been observed but not accuratelylocalized.

Unlike Imhotep, the smooth material formingHapi is piled up against the faces of Hathor andSeth in several places, suggesting that particlesare falling or falling back from the faces. This mayallow future studies of the coefficient of restitu-tion. There is also what appears to be a smallaeolian ripple field, and several boulders appearto have wind tails, indicating again that aeoliandust transport may be of significance (fig. S4).These observations suggest that Hapi is funda-mentally different in character from the Imhotepand Anubis regions.

Exposed consolidated surfaces

Several regions on 67P give a rocky (CCM) ap-pearance, although it is important to recognizethat the bulk densities of the materials are prob-ably factors of 5 to 10 lower than those of ter-restrial silicates (10). The regions included in thiscategory are listed in table S1. Hathor mostlyconsists of a 900-m-high “cliff” that rises up fromthe smooth Hapi region (Fig. 4). It is character-ized by a set of aligned linear features, which runvertically upward for much of the height of thecliff, and by roughly perpendicular linear featuresaligned with small terraces, which might suggestinner layering. The cliff is roughly parallel to thelocal gravitational acceleration (fig. S1), indicat-ing consolidatedmaterial of greater strength thanthe brittle material, with aligned weaknesses.Of further interest within the Hathor region

is the presence of an unlineated alcove, 250 to300mhigh, with unusual surface properties. TheCCM appears to contain distinct locally brightspots <10 m in dimension (Fig. 5, left). Theseare seen at all observing geometries and not onadjacent terrain (fig. S6). At the base of theCCM (on the surface of Hapi), there is evidenceof mass wasting. Within the talus, there are 2- to5–m-sized structures that are ~20% brighter inthe OSIRIS orange filter (lcentral = 649 nm) thanthe surroundings and are appreciably bluer thanthe surrounding material. This may be evidenceof more volatile-rich material being embeddedin the CCM and would imply that not only isthe surface of Hapi active (10) but also that theCCM has the potential to be active. Processesshould be considered that involve the sublima-tion of freshly exposed material that has fallenonto the surface of Hapi from its surroundings.Anuket borders Hathor and is also a consolid-

ated surface but shows no evidence of the linearfeatures seen in Hathor. Hathor borders the neck,whereas Anuket is part of the external surfaceof the head (fig. S1). Wide-angle camera views(fig. S5) show that the contact between Anuketand Ma’at is associated with a scarp (see alsofig. S1). This implies that material similar to thatof Anuket’s surface extends underneath the dust-covered brittle material of Ma’at, although alter-natives based on preferential erosionmay also beviable. It further suggests that processes withinthe neck are leading to erosion of the Anuketmaterial that forms the Hathor region. Thebrighter material in the Hathor region suggeststhat we are witnessing this erosion and conse-quently that Hathor is showing us the internal

aaa0440-4 23 JANUARY 2015 • VOL 347 ISSUE 6220 sciencemag.org SCIENCE

Fig. 5. (Left) Color composite of an alcove at the Hathor-Anuket boundary. The surface of the con-solidatedmaterial in the alcove contains distinct spots of brighter material independent of viewing geometry.On the floor beneath the consolidated material, material 20% brighter and bluer than the surroundings canalso be seen (e.g., positionsAandB).Central wavelengths of the filters used toproduce the compositewereR(882 nm), G (700 nm), and B (481 nm). (Sequence acquired at 15:42:17 on 21 August 2014.) (Right) Bright,highly reflectivemeter-sized boulders at the Babi-Khepry boundary adjacent tomantlematerial (C). Note alsothe brittle fracture and collapse at position D (NAC_2014-09-03T06.44.22.578Z_ID10_1397549400_F22).

Fig. 4. The Hathor structure seen from different perspectives. (Top right) The position of Hathor inrelation to other major regions. (Lower right) An oblique view of the Hathor face and the interface with theHapi region. Notice the boulders strewn along the longaxis of theHapi region. (Bottom left)Mappingof thelineaments seen in the Hathor face. (Top left) Identification of the contacts of the heavily lineated terrain.

CATCHING A COMET

structure of the head. This emphasizes the im-portance of trying to understand how the ob-served linear features in Hathor were produced.Anuket contains some notable cracks in its sur-face, which are roughly parallel to the neck (fig.S7). The location of these features within theneck region between the comet’s twomajor lobessuggests that theymay be the result of rotational-or activity-induced stresses in that particularpart of the comet. Their importance for the fu-ture evolution of the nucleus is at present un-clear (10).On the body, the Aker region contains CCM

and a set of tectonic fractures >200m in length(Fig. 6, left). The entire Aker region and muchof the neighboring Khepry region contain lin-ear features. Brittle material appears to be com-pletely absent. However, both Khepry and Akershow extremely smooth patches (50 to 100 m indimension) of material that are in local topo-graphic lows and surrounded by rougher mate-rial (Fig. 6, right). Boulders within their boundsare rare. These bear considerable similarity toponded material observed on asteroid 433 Eros

(26) in that they are smooth, flat, and of higherreflectance than the surroundings. For Eros,mechanisms such as electrostatic levitation offine particles, seismic shaking, and disaggrega-tion of local boulders have been proposed, with acurrent preference for electrostatic levitation (27).

Small-scale features

Local fracturing

Most of the consolidated materials on 67P areextensively fractured (e.g., Fig. 7 and Fig. 3, posi-tion F). Insolation weathering resulting from ei-ther thermal fatigue or thermal shock seems tobe a viable mechanism, based on terrestrial studies(28), given the huge temperature ranges and tem-poral gradients likely to be experienced by thesurface materials over diurnal and orbital timescales (29). However, fracturing is related to thethermal expansion coefficient of thematerial, whichis itself a function of porosity. The more porousnature of cometary materials, as compared toterrestrial materials, will therefore reduce the mag-nitude of the stresses. Conversely, the inferred low

tensile strength of cometary materials undoubtedlyenhances their propensity to fracture.With thermal fatigue, failure occurs along

the preexisting lines of weakness and does notgenerate new failure planes (28). Thermal shock,however, can produce the seemingly randomfracturing observed over most of the nucleus. Aclear exception is the observed vertical jointingin material that appears to have been upliftedafter the creation of a circular depression inthe Ash region (Fig. 7, right).

Evidence for block/chunkmovement and loss

Inspection of Fig. 7, left, also indicates thatthis particular fractured surface was once cov-ered by a blanket of material, a fewmeters thick,which has been lost with little or no residualtalus. This is reminiscent of the Aten depression.The Maftet region has several irregular-shapedpits, 10 to 20 m deep and 100 to 150 m in scale.On the surrounding terrain, there are blocks ofsimilar size (fig. S8). A possible interpretation ofthese observations is that gas pressure, buildingin a subsurface pocket (30), overcomes the over-burden and the cohesive strength of thematerial.A block or chunk is then ejected by the expand-ing gas. In the case of Fig. 7, left, and Aten, suf-ficient energy was apparently available to ejectthematerial from the area completely. ForMaftet,insufficient energy was available, and the blockwas therefore deposited on adjacent terrains. Theoverburden that the mechanism has to overcomeis in all cases only a few tens of pascals at most,considering the observed thickness combinedwith the gravitational acceleration. Observed iso-lated boulders may be the result of a similar pro-cess. Further evidence of volatile activity in theMaftet region (a possible flow) can be seen infig. S9, which offers some additional supportfor subsurface fluidization models.Given the low thermal inertia inferred for

cometary nuclei (31) and 67P in particular (32),it seems probable that the driving volatile is ei-ther CO or CO2 but almost certainly not H2O[based on thermodynamic properties (33)]. Clear-ly, models with a highly insulating airfall dustlayer above a slightly more conductive consoli-dated layer should be developed.

High-reflectivity particle clusters

Atmore than 10 places on the surface, there areclusters of bright, mostly unresolved particles(e.g., Fig. 5, right). Particle sizes can reach severalmeters in diameter. The unresolved individualparticles should be in the 10 cm to 1 m size rangein order to have sufficient cross-sectional area(and therefore be sufficiently bright) to stand outagainst the background lower-reflectance mate-rial. If these clusters are bright because they areice-rich, then they must also be new/fresh, as weexpect most, if not all, surface ices to be removedduring perihelion passage. This would indicatethat mass wasting of the brittle material seen inFig. 5, right, is active in the current epoch, andchanges in surface appearance are expected atthis position in the coming months.

SCIENCE sciencemag.org 23 JANUARY 2015 • VOL 347 ISSUE 6220 aaa0440-5

Fig. 6. (Left) Tectonic feature in the Aker region (NAC_2014-09-18T08.07.20.370Z_ID10_1397549000_F22).(Right) Ponded deposits in the Khepry region (NAC_2014-09-16T16.24.25.334Z_ID10_1397549500_F22).

Fig. 7. (Left) An exposed spur at the interface to Hapi, showing extensive irregular fracturing (A).Thecracking on the spur stops abruptly at a step (extending from position B) that seems to indicate a disruptionof a mantling layer (NAC_2014-09-12T05.09.04.388Z_ID10_1397549600_F22). (Right) Circular depressionthat appears to have produced uplift with vertical fracture structures (position C). Note also the evidence ofcollapse and talus generation (position D) (NAC_2014-08-08T04.20.34.577Z_ID30_1397549900_F22).

Similarly, there are strong brightness differences(up to a factor of 10) between the smooth terrainof Imhotep and some of the surfaces of bouldersnearby that are not readily explained by illumina-tion. The spectral ratio between these bright sur-faces and the surroundings indicates that they arerelatively blue and therefore likely to be ice-rich.

Conclusions

Although we are still in a relatively early phaseof the mission, the following conclusions ap-pear to be robust. The surface shows a widerange of different surface textures, which pointtoward a variety of processes. The presence ofairfall, dune/ripple-like structures, wind tails,and smooth depressions with ponded dust sug-gests that surface dust transport is of majorimportance in defining the uppermost surfacelayer in many regions. We suggest that this leadsnaturally to low surface thermal inertia and lowbut non-negligible gas production rates frommost of the surface by the redistribution of icychunks. Mass wasting of a consolidated brittle,possibly layered, material below this is perva-sive and is enhanced by undercutting of thislayer through erosion of its substrate in places.Fracturing of the surface is seen at all scales. Inspecific regions, the structural coherence of thesurface material has led to extensive aligned lin-eament sets over several hundreds ofmeters.Whenviewed in relation to the observed mass wasting,it suggestsmajor heterogeneity in tensile strength,which may also imply spatially variable thermalinertia (34). The observed, more random fractur-ing may result from insolation weathering andspecifically from thermal shock. There is evidenceto support surface erosion through the loss oflarge chunks of material (up to 10 to 100 m inscale), indicating that it may be amajormass lossprocess for the nucleus. The buildup of supervol-atile (e.g., CO and/or CO2) subsurface gas pocketsto pressures exceeding the overburden and cohe-sive strength appears to be a plausible mecha-nism that could also explain evidence of flows.Small-scale high-reflectivity materials have beenexposed in the surfaces of consolidated (“rocky”)materials, at the bases of consolidated structures,and adjacent to disintegrating brittle material.There are spectral indications that these are ice-rich nodules, which we expect to sublimate as thecomet moves toward perihelion in the comingmonths. A key outstanding question is whetherall the observed diverse large-scale morphologycan be explained by insolation alone or whetherphysical and/or chemical inhomogeneity (includ-ing large-scale layering) is required.

REFERENCES AND NOTES

1. H. U. Keller et al., OSIRIS - The scientific camera systemonboard Rosetta. Space Sci. Rev. 128, 433–506 (2007).doi: 10.1007/s11214-006-9128-4

2. H. U. Keller et al., Comet P/Halley's nucleus and its activity.Astron. Astrophys. 187, 807–823 (1987).

3. D. E. Brownlee et al., Surface of young Jupiter family comet 81P/Wild2: View from the Stardust Spacecraft. Science 304, 1764–1769(2004). doi: 10.1126/science.1097899; pmid: 15205524

4. P. C. Thomas et al., The shape, topography, and geology ofTempel 1 from Deep Impact observations. Icarus 187, 4–15(2007). doi: 10.1016/j.icarus.2006.12.013

5. P. C. Thomas et al., Shape, density, and geology of the nucleusof Comet 103P/Hartley 2. Icarus 222, 550–558 (2013).doi: 10.1016/j.icarus.2012.05.034

6. A. Basilevsky, H. U. Keller, Comet nuclei: Morphology andimplied processes of surface modification. Planet. Space Sci.54, 808–829 (2006). doi: 10.1016/j.pss.2006.05.001

7. F. Preusker et al., The northern hemisphere of asteroid (21)Lutetia - topography and orthoimages from Rosetta OSIRISNAC image data. Planet. Space Sci. 66, 54–63 (2012).doi: 10.1016/j.pss.2012.01.008

8. R. W. Gaskell et al., Characterizing and navigating small bodieswith imaging data. Meteorit. Planet. Sci. 43, 1049–1061(2008). doi: 10.1111/j.1945-5100.2008.tb00692.x

9. A shape model using stereophotoclinometry (SPC) has also beengenerated. The SPC and SPG techniques are complementary, anddetailed comparisons, outside the scope of this report, will bemade at a later date. The exclusive use of SPG here is solely forinternal consistency and simplicity and should not be interpretedas a choice of one approach over the other.

10. H. Sierks et al., Science 347, aaa1044 (2015).11. Although it is unusual for planetary geomorphology works, we do

not use a latitude/longitude system for the nucleus at this timebecause of the ambiguity that would result when using a simple(latitude/longitude) system. An approach to solving the cartographicproblem of the highly irregular shape is currently being formulated.

12. The sizes of the particles making up the surface layer are notknown at this time and are formally constrained only by theresolution limit (~1 m). The use of the word “dust” should notbe interpreted as implying a specific size range.

13. A. F. Cheng, C. M. Lisse, M. A'Hearn, Surface geomorphology ofJupiter Family Comets: A geologic process perspective. Icarus222, 808–817 (2013). doi: 10.1016/j.icarus.2012.10.004

14. R. Greeley, J. D. Iversen, Wind as a Geological Process on Earth,Mars, Venus and Titan (Cambridge Planetary Science Series,Vol. 4. Cambridge Univ. Press, Cambridge, 1985).

15. Y. Shao, H. Lu, A simple expression for wind erosion thresholdfriction velocity. J. Geophys. Res. 105, 22,437–22,443(2000). doi: 10.1029/2000JD900304

16. A. R. Poppe, M. Piquette, A. Likhanskii, M. Horányi, The effectof surface topography on the lunar photoelectron sheath andelectrostatic dust transport. Icarus 221, 135–146 (2012).doi: 10.1016/j.icarus.2012.07.018

17. Dust forms a thin veneer over the brittle material in the Ma’atand Ash regions. We here avoid the use of the word “mantle”because of the strong geological connotations and theimplications of a compositionally different surface layer forwhich we have no evidence at this time. As noted in the text,layering is evident at several scales, but the origin (primordialor evolutionary) remains open.

18. We use the term “brittle material” to imply a rock that breaks/fractures with a high ratio of fine to coarse fragments. It is aweakly consolidated material that displays fracturing but morereadily crumbles and creates debris deposits. Fracturescontained in more-consolidated material (e.g., in the Akerregion) show no evidence of fine fragments.

19. Assuming a rectangular cross-section, the tensile strength ofan overhang can be estimated using s = 3 r g cos a l2/h,where r is the density, g is the gravitational acceleration atthe overhang, a is the angle between gravitational accelerationand the normal of the overhang, l is the length of theoverhang, and h is its height. For larger, already collapsed,overhangs of ~200 m in length (e.g., at the Ash-Imhotepinterface), we estimate values of s ≤ 200 Pa. Smaller (~10 m)overhangs are seen (with talus below in some cases whereprevious collapse has occurred), which suggest that collapseoccurs at 10 to 20 Pa.

20. We are aware of some work in this field by (35) andunpublished calculations by M. Jutzi, which may be ofsignificance for the future.

21. Although the morphology of many surfaces on 67P suggests arock-like texture, we have avoided using that term herebecause of the expected very low density of the surfaces whencompared to terrestrial rocks. We have adopted the term“consolidated cometary material” instead (CCM).

22. P. Thomas et al., The nucleus of Comet 9P/Tempel 1: Shapeand geology from two flybys. Icarus 222, 453–466 (2013).doi: 10.1016/j.icarus.2012.02.037

23. M. J. S. Belton, H. J. Melosh, Fluidization and multiphasetransport of particulate cometary material as an explanation ofthe smooth terrains and repetitive outbursts on 9P/Tempel 1.Icarus 200, 280–291 (2009). doi: 10.1016/j.icarus.2008.11.012

24. M. D. Betterton, Theory of structure formation in snowfieldsmotivated by penitentes, suncups, and dirt cones. Phys. Rev.

E Stat. Nonlin. Soft Matter Phys. 63, 056129 (2001).doi: 10.1103/PhysRevE.63.056129; pmid: 11414983

25. Some smooth areas are also seen in the adjacent Khepryregion (fig. S10), again at different gravitational heights. Thisargues against a single flow or a single source of fluidizedmaterial being responsible. More complex emplacementmechanisms seem to be required.

26. M. S. Robinson, P. C. Thomas, J. Veverka, S. Murchie,B. Carcich, The nature of ponded deposits on Eros.Nature 413, 396–400 (2001). doi: 10.1038/35096518;pmid: 11574881

27. J. H. Roberts, L. M. Prockter, O. S. Barnouin, C. M. Ernst,E. Kahn, R. W. Gaskell, Origin and Flatness of Ponds on Asteroid433 Eros [American Geophysical Union (AGU) Fall MeetingAbstracts, AGU, San Francisco, CA, 2013], abstr. P23E-1831.

28. K. Hall, C. E. Thorn, Thermal fatigue and thermal shock inbedrock: An attempt to unravel the geomorphic processes andproducts. Geomorphology 206, 1–13 (2014). doi: 10.1016/j.geomorph.2013.09.022

29. Crack formation through a thermal contraction mechanism infrozen terrestrial soils is discussed in (36).

30. N. H. Samarasinha, A model for the breakup of Comet LINEAR(C/1999 S4). Icarus 154, 540–544 (2001). doi: 10.1006/icar.2001.6685

31. O. Groussin et al., The temperature, thermal inertia, roughnessand color of the nuclei of Comets 103P/Hartley 2 and9P/Tempel 1. Icarus 222, 580–594 (2013). doi: 10.1016/j.icarus.2012.10.003

32. P. L. Lamy et al., Spitzer Space Telescope observations of thenucleus of comet 67P/Churyumov-Gerasimenko. Astron.Astrophys. 489, 777–785 (2008). doi: 10.1051/0004-6361:200809514

33. W. F. Huebner, J. Benkho, M.-T. Capria, A. Coradini, C. De Sanctis,R. Orosei, D. Prialnik, Heat and Gas Diffusion in Comet Nuclei[International Space Science Institute (ISSI) Scientific Report,ISSI, Bern Switzerland, 2006], report SR-004.

34. B. Davidsson et al., Thermal inertia and surface roughness ofComet 9P/Tempel 1. Icarus 224, 154–171 (2013). doi: 10.1016/j.icarus.2013.02.008

35. D. Korycansky, E. Asphaug, Low-speed impacts between rubblepiles modeled as collections of polyhedra, 2. Icarus 204,316–329 (2009). doi: 10.1016/j.icarus.2009.06.006

36. A. C. Maloof, J. B. Kellogg, A. M. Anders, Neoproterozoic sandwedges: Crack formation in frozen soils under diurnal forcingduring a snowball Earth. Earth Planet. Sci. Lett. 204, 1–15(2002). doi: 10.1016/S0012-821X(02)00960-3

ACKNOWLEDGMENTS

OSIRIS was built by a consortium of the Max-Planck-Institut fürSonnensystemforschung, in Göttingen, Germany; CentroInterdipartimentale di Studi e Attività Spaziali–University of Padova,Italy; the Laboratoire d’Astrophysique de Marseille, France; theInstituto de Astrofísica de Andalucia, Consejo Superior deInvestigaciones Cientificas, Granada, Spain; the Research andScientific Support Department of the European Space Agency(ESA), Noordwijk, Netherlands; the Instituto Nacional de TécnicaAeroespacial, Madrid, Spain; the Universidad Politećhnica deMadrid, Spain; the Department of Physics and Astronomy ofUppsala University, Sweden; and the Institut für Datentechnik undKommunikationsnetze der Technischen Universität Braunschweig,Germany. The support of the national funding agencies of Germany(Deutschen Zentrums für Luft- und Raumfahrt), France (CentreNational d’Etudes Spatiales), Italy (Agenzia Spaziale Italiana), Spain(Ministerio de Educación, Cultura y Deporte), Sweden (SwedishNational Space Board; grant no. 74/10:2), and the ESA TechnicalDirectorate is gratefully acknowledged. H.R. was also supportedbygrant no. 2011/01/B/ST9/05442 of the Polish National ScienceCenter. W.-H.I acknowledges the Ministry of Science andTechnology, Taiwan (grant no. NSC 101-2111-M-008-016). M.F.A.acknowledges NASA funding through Jet Propulsion Laboratorycontract no. 1267923. We thank the ESA teams at European SpaceAstronomy Centre, European Space Operations Centre, andEuropean Space Research and Technology Centre for their work insupport of the Rosetta mission. The data will be placed in ESA’sPlanetary Sciences Archive after the proprietary period and areavailable on request until that time.

SUPPLEMENTARY MATERIALS

www.sciencemag.org/content/347/6220/aaa0440/suppl/DC1Figs. S1 to S11Table S1

10 October 2014; accepted 30 December 201410.1126/science.aaa0440

aaa0440-6 23 JANUARY 2015 • VOL 347 ISSUE 6220 sciencemag.org SCIENCE

CATCHING A COMET