Icarus 222 (2013) 424–435

Contents lists available at SciVerse ScienceDirect

Icarus

journal homepage: www.elsevier .com/ locate/ icarus

Return to Comet Tempel 1: Overview of Stardust-NExT results

J. Veverka a,⇑, K. Klaasen b, M. A’Hearn c, M. Belton d, D. Brownlee e, S. Chesley b, B. Clark f, T. Economou g,R. Farquhar h, S.F. Green i, O. Groussin j, A. Harris k, J. Kissel l, J.-Y. Li c, K. Meech m, J. Melosh n,J. Richardson n, P. Schultz o, J. Silen p, J. Sunshine c, P. Thomas a, S. Bhaskaran b, D. Bodewits c, B. Carcich a,A. Cheuvront q, T. Farnham c, S. Sackett a, D. Wellnitz c, A. Wolf b

a Cornell University, Ithaca, NY 14853, USAb Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA 91109, USAc University of Maryland, College Park, MD 20742, USAd Belton Space Exploration Initiative, Tucson, AZ 85716, USAe University of Washington, Seattle, WA 98195, USAf Space Science Institute, Boulder, CO 80301, USAg University of Chicago, Chicago, IL 60637, USAh Kinetx, Tempe, AZ 85084, USAi The Open University, Milton Keynes, MK7 6AA, UKj Laboratoire d’Astrophysique de Marseille, 13388 Marseille Cedex 13, Francek Space Science Institute, La Canada, CA 91011, USAl Max Planck Institute for Solar System Research, Katlenburg-Lindau, Germanym University of Hawaii, Honolulu, HI 96822, USAn Purdue University, Lafayette, IN 47907, USAo Brown University, Providence, RI 02412, USAp Finnish Meteorological Institute, Helsinki 00560, Finlandq Lockheed Martin, Littleton, CO 80127, USA

a r t i c l e i n f o a b s t r a c t

Article history:Available online 12 April 2012

Keywords:Comets, dustComets, nucleusComet Tempel-1Comet Wild-2

0019-1035/$ - see front matter � 2012 Elsevier Inc. Ahttp://dx.doi.org/10.1016/j.icarus.2012.03.034

⇑ Corresponding author.E-mail address: veverka@astro.cornell.edu (J. Veve

On February 14, 2011 Stardust-NExT (SN) flew by Comet Tempel 1, the target of the Deep Impact (DI)mission in 2005, obtaining dust measurements and high-resolution images of areas surrounding the2005 impact site, and extending image coverage to almost two thirds of the nucleus surface. The nucleushas an average radius of 2.83 ± 0.1 km and a uniform geometric albedo of about 6% at visible wavelengths.Local elevation differences on the nucleus reach up to 830 m. At the time of encounter the spin rate was213� per day (period = 40.6 h) and the comet was producing some 130 kg of dust per second. Some 30% ofthe nucleus is covered by smooth flow-like deposits and related materials, restricted to gravitationallows. This distribution is consistent with the view that the smooth areas represent material erupted fromthe subsurface and date from a time after the nucleus achieved its current shape. It is possible that someof these eruptions occurred after 1609 when the comet’s perihelion distance decreased from 3.5 AU to thecurrent 1.5 AU. Much of the surface displays evidence of layering: some related to the smooth flows andsome possibly dating back to the accretion of the nucleus. Pitted terrain covers approximately half thenucleus surface. The pits range up to 850 m in diameter. Due to their large number, they are unlikelyto be impact scars: rather they probably result from volatile outbursts and sublimational erosion. TheDI impact site shows a subdued depression some 50 m in diameter implying surface properties similarto those of dry, loose snow. It is possible that the 50-m depression is all that remains of an initially largercrater. In the region of overlapping DI and SN coverage most of the surface remained unchanged between2005 and 2011 in albedo, photometric properties and morphology. Significant changes took place onlyalong the edges of a prominent smooth flow estimated to be 10–15 m thick, the margins of which recededin places by up to 50 m. Coma and jet activity were lower in 2011 than in 2005. Most of the jets observedduring the SN flyby can be traced back to an apparently eroding terraced scarp. The dust instrumentsdetected bursts of impacts consistent with a process by which larger aggregates of material emitted fromthe nucleus subsequently fragment into smaller particles within the coma.

� 2012 Elsevier Inc. All rights reserved.

ll rights reserved.

rka).

Table 1bEncounter circumstances.

Deep Impact Stardust-NExT

Encounter time from perihelion �1 day +34 dayFlyby distance 500 km 178 kmFlyby speed 10.2 km/s 10.9 km/sActivity (H20) 5 � 1027 mol/s 3 � 1027 mol/s

J. Veverka et al. / Icarus 222 (2013) 424–435 425

1. Introduction

In 2007 NASA approved an extended mission for the Stardustspacecraft, which had successfully completed its mission to CometWild 2 by collecting and returning dust samples from the comet’scoma to Earth (Brownlee et al., 2004). The extended mission, Star-dust-NExT (for New Exploration of Tempel) flew by Comet Tempel1, the target of the Deep Impact (DI) mission in 2005 (A’Hearnet al., 2005), on February 14, 2011. Key parameters describingTempel 1, its orbit, and the circumstances of the Deep Impactand Stardust-NExT encounters are summarized in Table 1.

The primary goal of Stardust-NExT (SN) was to use the space-craft camera, NAVCAM, to obtain high-resolution images of the nu-cleus to

(a) Look for changes on the comet’s surface that might haveoccurred between the 2005 perihelion passage and that inJanuary 2011.

(b) To extend coverage of the surface to regions not imaged byDI.

(c) To image the impact site where Deep Impact’s impactor hitthe surface.

All three of these goals were achieved. The 2005 impact threwup so much ejecta that the cameras on the main DI spacecraftcould not see the surface in the impact area, and therefore the sizeof the crater that had been excavated could not be determined(A’Hearn et al., 2005). In addition to the NAVCAM camera, thespacecraft carried two experiments to study the comet’s dust envi-ronment: the Dust Flux Monitoring Instrument (DFMI) designed todetermine the fluence and size distribution of dust particles(Tuzzolino et al., 2003) and the Comet and Interstellar Dust Ana-lyzer (CIDA) designed to measure the elemental composition ofindividual dust grains (Kissel et al., 2004). Results obtained bythe three SN instruments are reported in this special issue of Icarus.

To reach Tempel 1 the spacecraft performed an Earth flyby onJanuary 14, 2009 and some dozen subsequent TCM’s or trajectorycorrection maneuvers (Fig. 1). The targeted flyby distance was200 km. The actual flyby occurred at 178 km at a speed of10.9 km/s. In addition to distant images on approach and depar-ture, 72 images were obtained within ±4 min of closest approach:the best images obtained were at 11 m/pixel.

The solar phase angle varied from 81� on approach through 15�at closest approach, reaching 98� on departure. DFMI data werecollected from 22 min before to 8 min after closest approach

Table 1aTempel 1 characteristics.

Orbit semi-major axis 3.12 AUOrbit eccentricity 0.52Orbital period 5.52 yearsPerihelion date January 12, 2011Mean radius 2.83 ± 0.1 kmSurface area 108 km2

Volume 95 km3

Densitya 400 kg m�3

Range of radii 2.21–4.00 kmSpin period (at encounter) 40.6 hSpin poleb RA = 255, DEC = 64.5Surface acceleration 0.025–0.032 cm s�2

Direction of maximum momentc 200.5 W, +82.0Average geometric albedo 0.059 (Hapke model)

0.045 (Minnaert model)

a Richardson et al. (2007).b Observations on July 2005 and February 2011.c Assuming uniform density.

(�14,000 km to +5200 km from the nucleus). CIDA collected spec-tra for 2 h around closest approach (±78,000 km).

An essential aspect of the mission was to arrive at the comet ata time when the region previously imaged by Deep Impact wouldbe visible and well illuminated. This requirement meant that thecomet’s spin state and spin rate had to be predicted with highaccuracy well before the encounter. Dynamically the optimumtime to make a time-of-arrival adjustment (TOA) was about 1 yearbefore the scheduled February 2011 flyby (Fig. 1). Accordingly, aTCM was executed on February 17, 2010 to delay the arrival atthe comet by about 8.5 h.

The magnitude of the correction was based on extensive analy-ses of the comet’s rotation behavior reaching back to 2000, basedon observations of the comet’s light curve by a worldwide networkof astronomers organized to support the Deep Impact and Star-dust-NExT missions (Meech et al., 2005 and Meech et al., 2011).The light curve observations were analyzed by two independentgroups to predict the comet’s rotation state at the time of encoun-ter (Belton et al., 2011).

2. Imaging

The primary science instrument on Stardust-NExT is the naviga-tion camera (NAVCAM) the performance and calibration of whichis described in detail by Klaasen et al. (2013). NAVCAM uses a1024 � 1024 pixel CCD detector and optics that provide a field ofview (FOV) of about 3.5� (60 lrad/pixel) (Newburn et al., 2003).The camera’s filter wheel failed early in the primary mission(Brownlee et al., 2004), so all NExT images are acquired througha broadband filter (510–760 nm). Imaging within ±4 min ofencounter (E) was restricted by available spacecraft memory to72 full-frame, compressed images. Images were taken on 8-s cen-ters outside of E ± 144 s and on 6-s centers inside this interval. Thelatter provided excellent stereo coverage of the nucleus near clos-est approach. Image data can be returned fully encoded to 14 bitsor compressed to 8 bits per pixel using an onboard look up table.The data system can support a maximum imaging rate of one fullframe of compressed data every 6 s. NAVCAM performance wasmonitored throughout the Stardust-NExT mission using standardcalibration sequences that involved imaging a variety of stars,acquiring dark current frames, etc. (Klaasen et al., 2013). A problemwith recurring camera contamination was controlled successfullyby periodic heating of the instrument using its internal electricalheaters and placing direct sunlight on the camera radiator. Noevidence of optical contamination is observed in any of theencounter images.

Repetitive imaging of the comet was carried out before andafter closest approach to monitor the light curve and coma behav-ior. NAVCAM imaging of Tempel 1 was initiated at E � 60 daysusing exposures of 10 and 20 s (the maximum commandable),but the comet was too faint to be detectable until E � 27 days.Images with good signal to noise ratios became available atE � 7 days. Imaging sets were obtained every 2 h from this pointuntil E � 2 days when approach imaging was halted to preparethe spacecraft for encounter.

Fig. 1. Summary trajectory of the Stardust-NExT mission.

426 J. Veverka et al. / Icarus 222 (2013) 424–435

The pixel scale ranged from 158 m/pixel to 11 m/pixel at closestapproach. Four of the 72 images (first, last, and those at E ± 72 s)were overexposed intentionally to allow better detection of near-nucleus jets (Farnham et al., in review). Slight saturation of thebright limb occurred in two frames (E � 33 s and E � 15 s) due tothe actual arrival time being 15 s earlier, which caused someimages to be taken at lower phase angles than planned. A sampleof nine close encounter images is shown in Fig. 2.

Departure imaging began at E + 1 day and continued throughE + 10 days to support high-time-resolution monitoring of comaactivity. The sampling rate was one image every 5 min at thebeginning of the monitoring period, and was decreased to onceevery 11 min at E + 7 days.

3. Light curve observations

The signature of the comet’s rotation was detected in the pho-tometry from E � 3 days to E + 5 days. The predictions in Beltonet al. (2011) compare well with the NExT results. The time-aver-aged photometry over the 8-day observing period yields a spin rateof 210 ± 3�/day. With the additional constraint that the sub-solarlongitude at encounter was 321.7�W, the spin rate is213.3 ± 0.8�/day. The predicted spin rate at encounter was213.5 ± 0.2�/day (Belton et al., 2011). The observed sub-solar longi-tude at encounter of 321.7�W differs from the predicted value of342 ± 29�W by only 21�. A post-encounter reanalysis of the prob-lem of predicting the expected spin state at the time of encounteris provided by Chesley et al. (2013). Chesley et al. update theconclusions of Belton et al. (2011) most of which remain un-changed. However, they provide new insights into the torques that

were active at the time of the 2005 perihelion: most of the angularacceleration of the nucleus appears to have occurred well beforeperihelion. They also find that the rotational impulse during the2011 perihelion was weaker than in 2005.

No evidence was found in the densely sampled departure pho-tometry of the pseudo-periodic mini-outbursts seen in the DeepImpact approach photometry in 2005 (Belton et al., 2008), perhapsbecause the SN encounter occurred 34 days post-perihelion whilethe DI encounter took place 1 day before perihelion. The comet’sactivity is known to peak some 60 days pre-perihelion (Schleicher,2007).

4. Pole orientation and shape

The spin period of Tempel 1 changes with time in a complexfashion (Belton et al., 2011) but is approximately 40 h. Combiningimaging data from the Deep Impact and Stardust-NExT flybysyields an improved estimate of the spin axis orientation (Thomaset al., 2013). Approximately 480 manually selected control pointswere used to solve for both the pole orientation and for thebody-centered coordinates of a shape model. The solution em-ployed data from four cameras on three spacecraft (Deep ImpactFlyby and Impactor craft and Stardust-NExT). The effective spinpole orientation is found to be RA = 255�, DEC = 64.5�, a solutionnearly 16� different from the Deep Impact estimation of 293.8�,72.6� (Thomas et al., 2007). Some of this displacement is undoubt-edly due to the larger uncertainty in the original Deep Impact solu-tion (�±5�) and some could be due to precession. Belton et al.(2011) estimated that precession should, at most, amount to�1�/perihelion passage. Ambiguities of limb coordinates were

Fig. 2. A sample of Stardust-NExT near-encounter images. Inbound images at upper left to outbound at lower right (n30033–n30041). Range/phase angle: (top row) 352 km/52�; 297, 46�; 249, 38�; (middle row) 209/26�; 185/15�; 182/22�; (bottom row) 202/38�; 239/51�; 286/60�.

J. Veverka et al. / Icarus 222 (2013) 424–435 427

probably the major cause of the poor spin pole solution in Thomaset al. (2007). The new solution is free of such defects.

The shape model is well constrained by the combined Deep Im-pact and Stardust-NExT images, which cover two-thirds of the sur-face, and by limb silhouettes from the 2005 data. The mean radiusderived (2.83 ± 0.1) km is slightly smaller than the value derived byThomas et al. (2007) from DI data alone, but the resulting change inestimated volume is insignificant (Table 1). The value for the co-met’s mean density estimated by Richardson et al. (2007) at 400(+600, �200) kg m�3 remains unchanged. No estimate of the co-met’s mass could be obtained from the SN flyby.

Chesley et al. (2013) show that the improved shape model andrevised pole orientation provide significantly improved fits toaccumulated light curve data and that precession must be negligi-bly small. Thomas et al. (2013) note that the pole direction iswithin a few degrees of the maximum moment axis based on theshape model assuming a homogeneous density distribution withinthe nucleus.

5. Nucleus photometry

Stardust-NExT images of Tempel 1 cover phase angles from 81�on approach, down to 15� at closest approach, increasing to 98� ondeparture. The photometric properties of the nucleus modeledfrom NExT images agree closely with those derived by Li et al.(2007) from Deep Impact data. Comparison of the two image setsobtained one-comet year apart reveals no significant photometricchanges. Other than small variations (at the 10% level) in albedo,the surface is photometrically homogeneous. The average clear-fil-ter geometric albedo is 0.059 ± 0.009; the error bar quoted includesthe uncertainty in extrapolating to zero phase. Outbound imagesshow isolated brighter spots that have albedos only 25% greater

than surrounding areas. Significantly, no ‘‘bright albedo patches’’similar to those reported by Sunshine et al. (2006) and shown tobe associated with exposed water ice are seen in the areas imagedby NExT for the first time. The bright patches reported by Sunshineet al. had albedoes up to four times higher than surrounding areas.NExT images do not cover the region where these patches wereobserved in 2005.

6. Surface morphology

The approximately one third of the comet’s surface imaged athigh resolution by Deep Impact in 2005 revealed a geologically di-verse surface consisting of three major units (Thomas et al., 2007):extensive regions of layered terrains displaying varying degrees oferosion; smooth areas with preserved flow-like characteristics sug-gestive of down-slope movement of material that had erupted ontothe surface; and regions characterized by generally rimless, crater-like pits.

Fig. 3 shows the expanded coverage of the surface obtained byStardust-NExT added to the Deep Impact observations. Approxi-mately two thirds of the comet’s surface has been imaged betweenthe two missions. Morphologically, Tempel 1’s surface can be di-vided broadly into two types of terrain: pitted terrains and smoothterrains. The expanded coverage reveals additional areas of smoothterrain and demonstrates that this terrain occurs preferentially ingravitational lows on the nucleus (see below).

6.1. Pitted terrain

A significant fraction of the newly imaged portion of Tempel 1’ssurface is covered by a relatively rough pitted terrain somewhatsimilar to that on Wild 2 (Brownlee et al., 2004), and unlike

Fig. 3. Maps of Tempel 1. (A) Global mosaic of combined DI and SN coverage. The region around 270�E was imaged only by DI. Common coverage (DI and SN) extends oversouthern and equatorial latitudes between about 200 and 40�E. The remainder is new coverage obtained by SN. (B) Gravitational topography of the nucleus from Thomas et al.(2013). Topography in some of the blank areas in the mosaic was constrained by limb profiles. The maximum topographic variation is about 830 m. (C) Sketch map ofprominent surface features. Four areas of smooth terrain (S1–S4) are indicated. Pits marked by circles of scaled diameters. Dark dots show mesa-like forms.

428 J. Veverka et al. / Icarus 222 (2013) 424–435

anything seen on the surfaces of Comet Halley (Keller et al., 1988),Borrelly (Soderblom et al., 2004) or Hartley 2 (A’Hearn et al., 2011).Early discussions as to whether these pits are entirely endogenic(representing areas in which volatiles are sublimating) or whetherat least some of them are impact scars modified by later sublima-tional erosion had proved inconclusive (Thomas et al., 2007). Thesedepressions vary widely in size, morphological sharpness, and cir-cularity suggesting a broad range of development and age. In termsof sharpness, the depressions range from almost circular pits todepressions that have irregularly scalloped walls. In many cases,on both Wild 2 and Tempel 1, such depressions have flat floors,suggestive of possible layering of the surface or ongoing floordeposition. In no cases are raised rims detectable. A typical contactrelationship between the higher, rougher pitted terrain and thelower, smoother flow-like deposits is shown in Fig. 4.

Relative to nucleus size, the depressions on Wild 2 are substan-tially more prominent than those on Tempel 1. The largest well-de-fined depressions on either comet are about 800 m in diameter.(The mean radii of Wild 2 and Tempel 1 are 2.1 and 2.8 km, respec-tively.) Between diameters of 800 m and 100 m the size distribu-tion of depressions is almost the same on the two cometsfollowing a �2 slope on a cumulative log–log plot (Thomas et al.,2013). In this size range there are approximately 10 times more

depressions per unit area on the surface of Wild 2 than on Tempel1, evidently the result of the absence of extensive smooth regionson Wild 2. Wild 2 appears to have many more flat-floored pits thandoes Tempel 1. It is unclear whether this reflects the influence oflayering on Wild 2 or more loose debris on the floors of pits on thatcomet.

The origin of the pitted terrain is discussed at length by Beltonet al. (2013). Based on the work of Duncan et al. (2004), Beltonet al. argue that given the calculated impact rates on a typical Jupi-ter Family comet during its lifetime there are far too many pits onTempel 1 to be accounted for by impacts. They suggest that mostpits originate as sources of material ejected into the coma during‘‘mini-outbursts.’’ They estimate that a mini-outburst of 1 mag in-volves the removal of enough surface material to produce a pit30 m across. Such pits could be enlarged by repeated outburstsfrom the same location, or possibly by sublimation of volatilesfrom the pit walls and floor.

Belton et al. (2013) use the improved shape model and spin axisorientation derived by Thomas et al. (2013) to update estimatedsource locations of repetitive mini-outbursts observed by the DIspacecraft (Belton et al., 2008; A’Hearn et al., 2005) and confirmtheir earlier finding that these locations are restricted to thecomet’s pitted terrain. It is noteworthy that no significant albedo

Fig. 4. (Top): SN picture n30038 showing the contact between pitted terrain (topright) and smooth areas on the nucleus (bottom left). (Bottom): Close-up stereo pairof contact area. The terraced scarp is approximately 50 m high. Isolated mesaswithin the smooth region are approximately 20 m high.

J. Veverka et al. / Icarus 222 (2013) 424–435 429

variations occur in the pitted terrain (Li et al., in review) and thatno localized albedo markings around the edges or within the inte-riors of pits can be identified at the resolutions available (generallyabout 20 m/pixel).

6.2. Smooth terrain

The combined DI and SN coverage shows that smooth terrainscover about 30% of the surface and that they are located in thethree prominent gravitational lows on the nucleus (Thomas et al.,2013). Such locations are consistent with the interpretation thatthese are deposits of material erupted from the subsurface andaccumulated in the lowest areas. A specific model based on therelease of highly volatile materials such as CO2 and CO in the inte-rior followed by the fluidization of overlying material has beendeveloped by Belton and Melosh (2009).

Two of the smooth regions were identified in the Deep Impactimages (S1 and S2 in Fig. 3). The better imaged of the smooth areas(S2) showed flow-like characteristics suggesting that it resultedfrom the eruption of mobile materials onto the surface. This flow,S2, the northern edge of which occurs in the area of the DI impactsite, was also imaged by Stardust-NExT (Fig. 5).

A third, more extensive and more complex area of smooth ter-rain (S3 in Fig. 3) is found in the expanded Stardust-NExT coveragein Fig. 5. This region shows evidence of several generations of

deposition and erosion. Part of its margin is a topographic terraceabout 50 m high showing up to five benches, possibly layers, eachof which is about 10 m thick. Mesa-like forms suggest removal ofmaterials over much of this area.

To the west is a distinct smooth area (S4 in Fig. 3) resembling aplateau with well-defined digitate margins. It displays the smooth-est surface on this face of the comet and it appears to be the youn-gest flow in this area. Unlike the flows on the DI side, the interior ofthis flow is pitted. Based on shadow measurements within the pits,this flow is about 4–16 m thick. This thickness is similar to that offlow S2, which is about 10–15 m thick at its distal margin.

A detailed discussion of the characteristics of the four areas ofsmooth terrain on Tempel 1 is provided by Thomas et al. (2013).In summary, these areas occupy some 30% of the nucleus surfacerestricted to gravitational lows and to varying degrees displayevidence of flow-like morphology. At least one area shows strongevidence of multiple flow events, and all display evidence of activeerosion. While it is impossible to date these features, they must berelatively recent. An interesting speculation is that some or all maybe evidence of increased activity following the reduction ofTempel’s perihelion distance from 3.5 to the current 1.5 AU whichoccurred in 1609 according to Yeomans et al. (2005).

6.3. Layering

Layers of varying thickness are conspicuous in both the DI andStardust-NExT images.

The layers may be of two distinct types: thinner layers 10–15 mthick associated with the smooth flows that cover the low lyingportions of the surface and the thicker layers (50–100 m thick) thatare being exposed in places by erosion.

The thicker layers on the DI side of the comet were mapped anddescribed by Thomas et al. (2007). One published interpretation ofthese thicker layers is that they date to the time of accretion of thenucleus: the TALPS model of Belton et al. (2007). Recent numericalsimulations of this process are discussed by Thomas et al. (2013).The elevated area on the equator centered near 45E (Fig. 6, left)may consist of a stack of layers. If so, even for layers as thick as50–100 m, a considerable number of individual layers would beneeded to make up the 830 m of relief of this elevated area(Fig. 6, right).

Scarp morphology at boundaries varies considerably across thesurface (Fig. 7). Some scarp edges are sharp, some concave, someterraced, and others scalloped indicating different styles of erosionand implying differences in composition and/or texture.

7. Impact site

Three contributions to this special issue deal with the DI impactsite. Wellnitz et al. update information on the location of the im-pact site, while Richardson and Melosh and Schultz et al. offercomplementary interpretations of SN images of the impact area.

Previous analyses of the DI impact event were published byA’Hearn et al. (2005) and by Richardson et al. (2007), and Schultzet al. (2007), among others. A major result based on observationsof the ejecta plume was the estimate of surface gravity and henceof the mean density (400 kg/m3) of the nucleus made by Richard-son et al. As is well known, the DI spacecraft did not succeed inimaging the crater made by the impactor (see for example, A’Hearnet al., 2005).

Wellnitz et al. (in review) re-analyze DI data related to the loca-tion of the impact site, particularly the nested sequence of imagesfrom the impactor on its way to its collision with the surface. Theseresults, consistent with previous analyses, locate the impact site tobetter than 100 m on the surface (Fig. 8), and probably to about

Fig. 5. (Top left): SN image n30034 of smooth flow S2 which was imaged by both SN and DI. (Top right): Smooth flows S3 and S4 imaged only by SN, n30040. (Bottom panels):Correlation between topographic lows and location of smooth regions. On Tempel 1 smooth terrains are associated with low regions on the nucleus, consistent with theinterpretation that these terrains originate as flows that are erupted onto the surface. Corresponding to each image in the top row is a topographic map (bottom row) withboundaries of smooth regions outlined in RED. Elevations color-coded as in Fig. 3. See also Thomas et al. (2013).

Fig. 6. (Left): Stardust-NExT image n30036 showing lineations suggestive of large scale layering in the left hand side of the image. (Right): Shape model showing elevationdifferences in this area (red = high and blue = low as in Fig. 3) and the trend of the layers (dashed lines). The maximum elevation difference is 830 m.

430 J. Veverka et al. / Icarus 222 (2013) 424–435

25 m (see also Richardson and Melosh, in review; Schultz et al., inreview).

Richardson and Melosh (2013) identify a shallow depressionwithin tens of meters of the calculated impact site as the DI crater.The feature is 4 ± 1 pixels across corresponding to a diameter of50 ± 12 m. On the basis of this identification they place constraintson the material properties of the surface at the impact site. Theyfind that the Earth-analog material with most similarity to Tempel

1 material in terms of density, strength, and scaling parameters is alightly packed, mountain snow.

The observed crater size and the scaling parameters can also beapplied to the excavation model developed in Richardson et al.(2007). This model leads to a DI impact total ejected mass of1.2 � 106 kg (5.4 � 105–2.6 � 106 kg), with 60 ± 20% of this massejected at greater than the comet’s escape velocity of 1.44 m/s.An ejecta blanket thickness can be estimated, using the technique

Fig. 7. Variety of scarp morphology observed on Tempel 1 including a terraced scarp, a scalloped depression wall, a sharply bounded flow edge and a concave eroding scarp.These differences in erosion style suggest differences in composition and/or texture.

Fig. 8. DI impact site. Pre-impact (left) and post-impact (center and right) images of the DI impact area. The SN image is the highest resolution (scale about 12 m/pixel) post-impact view of the impact area. The arrows at bottom indicate the directions of incident sunlight in the DI and SN views. The cluster of arrows in the rightmost view point tothe impact site.

J. Veverka et al. / Icarus 222 (2013) 424–435 431

described in Richardson (2009), which gives a relatively thin ejectablanket thickness at the final crater rim of 2.7 mm (0.5–13.3 mm).At the time of excavation, a significant fraction (30–60%) of thisejecta blanket would have been water ice (Lisse et al., 2006), whichwould have sublimated over a period of hours.

A complementary analysis of the impact site is provided bySchultz et al. (in review). Schultz et al. also identify a 50 m ‘‘dim-ple’’ at the impact site, but in addition point out a possible partialouter rim some 150 m wide that appears to overlap an area inwhich a tiny mound is seen in the DI images, but not in the SN cov-erage. Schultz et al. consider the possibility that the DI crater mayhave originally been larger but was significantly modified by post-impact collapse or by erosion and mass wasting over the 5.5 yearsseparating the DI and SN encounters.

8. Surface changes

A major objective of the Stardust-NExT imaging experimentwas to document surface changes that were expected to occur be-tween the 2005 and 2011 perihelion passages. Lisse et al. (2005,2006) determined that the comet loses some 109 kg of materialper perihelion passage due to sublimation, corresponding to about

1 m of surface material lost from active areas if it is assumed thatabout 10% of the surface is active.

The Stardust-NExT images cover about 20 km2 of territory pre-viously imaged by Deep Impact. The overlap region extends fromthe area of the impact site and includes the northern extent ofthe southern-most smooth flow discovered by Deep Impact. Allow-ing for the somewhat different viewing geometry (about 50� emis-sion angle for DI, compared to about 30� for Stardust-NExT) theoverlap area looks remarkably unchanged (Fig. 9). There are noobvious changes in albedo on scales of a hundred meters or more.The larger dark circular makings near the impact site appear un-changed. Several small albedo spots in the region have disap-peared, and others have appeared. A few may have changed incontrast and extent. Given the small linear scale of these featuresand the limited resolution of the images (the best have a pixel scaleof 11 m/pixel), it is impossible to determine whether we are seeingdifferences due to modifications of surface texture, surface slope,or amount of exposed water ice rather than the effects of slight dif-ferences in illumination and viewing conditions between the twoimage sets.

The only significant changes in morphology occur along thebounding scarp of smooth flow S2, which at least in two placeshas receded by up to 50 m between 2005 and 2011. Thomas

Fig. 9. Changes in the margins of flow S2 between 2005 and 2011. Left is Stardust-NExT image n30035, right from Deep Impact ITS image 173727747. Images have been mapprojected and further warped to reliable common points. Some bright albedo spots have both been added and disappeared, and changed in contrast and extent. The scarp atthe bottom has lost two roughly triangular segments (arrows) of maximum horizontal extent �50 m. Other areas may have changed as well, but limited resolution anddifferences in lighting and viewing geometry restrict the confidence of such conclusions.

Fig. 10. Jets on the sunward side of the nucleus. The coma structure has beenenhanced by dividing out a 5� rotational average to improve the contrast of the jets.The nucleus is inset using a separate stretch. The shadow of the nucleus against thebackground coma can be seen on the right of the image. The Sun is toward the left,circular structures in the coma are artifacts created by pixelization at low signallevels.

432 J. Veverka et al. / Icarus 222 (2013) 424–435

et al. (2013) estimate that this flow is some 8–15 m thick and thatthe total loss of material may amount to some 3 � 105 m3 or about2 � 108 kg.

Thomas et al. provide a revised estimate of the mass loss per or-bit of Tempel 1. Their value of 7–14 � 109 kg is about a factor of 10higher than previous estimates. Thus the observed material lossfrom the boundary of S2 accounts for about 2% of the total massloss suggesting that mass loss is highly concentrated to specificareas on this comet: the 2% coming from about 0.01% of the surfacearea. More detailed discussions of mass loss from Tempel 1 areprovided by Thomas et al. (2013) and by Farnham et al. (in review).In this context it is interesting to note that the sources of most jetsobserved during the SN encounter can be traced to sources alongan apparently eroding terraced scarp (see below and Farnham, inreview). This terraced scarp occurs on the face of the comet imagedonly by SN, making it impossible to document any possiblechanges between 2005 and 2011.

9. Coma and jet activity

Coma and jet activity during the SN encounter is described byFarnham et al. (in review). The activity level of Comet Tempel 1was lower during Stardust-NExT than it was during Deep Impactbecause the encounter occurred 34 days after perihelion (com-pared to a day before perihelion for DI), and possibly because ofa steady decline in water production reported in recent apparitions(Schleicher, 2007). Strong seasonal and diurnal variations suggestthat the gaseous activity of Tempel 1 is confined to a limited num-ber of active areas.

The Stardust-NExT images show fewer coma features than wereobserved by Deep Impact. A subset of the encounter images(n30025–n30036) reveal a number of small, well-defined jets,highlighted against the dark background at the edge of the nucleusas they move over the horizon (Fig. 10). These jets appear to orig-inate in the vicinity of +30� latitude and 30–90� longitude. Most ofthe jet sources appear to originate at or close to the prominent ter-raced scarp (Fig. 4) and indicate that this scarp is actively eroding.The spacecraft passed over this area, but preliminary analyses ofDFMI data indicate that it did not fly through any individual jets.Active areas have been identified previously at similar, low north-ern latitudes (Farnham, 2009; Schleicher, 2007; Vasundhara, 2009;Belton, 2010; Vincent et al., 2010). Farnham et al. estimate the totaldust production to be 130 kg/s at the time of the SN flyby.

10. Dust environment: DFMI measurements

Results from the Dust Flux Monitor Instrument (DFMI) are sum-marized by Economou et al. (in review). The instrument measuresparticle impacts using two kinds of sensors—one based on polyvi-nylidene fluoride (PVDF) thin films, the other on acoustic detectors(Tuzzolino et al., 2003). The PVDF sensors comprise two circularfilms of 20 cm2 and 200 cm2, with four different mass thresholdseach. The two acoustic sensors, with two mass thresholds each,are mounted on the front and second protective shields of thespacecraft (with sensitive area 0.3 m2 and 0.7 m2 respectively).Particles reaching the second shield have to penetrate the frontshield. At the higher encounter speed of 10.9 km/s compared tothe 6.1 km/s at Wild 2, the mass sensitivity of DFMI sensors

J. Veverka et al. / Icarus 222 (2013) 424–435 433

increased by approximately a factor of between 2 to 12 (dependingon the sensor and mass channel) providing sensitivity to dust par-ticles in the mass range from �3 � 10�15 to >10�6 kg. The PVDFsensors accumulated impact counts in 0.1-s time bins. The deriva-tion of impact counts from the acoustic sensor signals is more com-plex, with time bins between 0.1 and 1 s (Tuzzolino et al., 2003;Green et al., 2004).

DFMI was powered on 22 min (�14,000 km) before closest ap-proach. The first particle detection occurred at 4300 km; 90% ofthe particle events were observed within 300 km of the nucleus.The DFMI data (Fig. 11, top) show clearly that dust is not emitteduniformly into the coma but occurs in clusters. This observationis consistent with the conclusions from Wild 2 (Tuzzolino et al.,2004) that larger aggregates of material are emitted sporadicallyfrom the cometary nucleus and undergo significant fragmentationinto smaller particles in the coma (Clark et al., 2004). Bursts of upto 1000 impacts of particles a few microns in diameter in 0.1 swere detected at Wild 2 as the spacecraft flew through expanding

Fig. 11. Dust measurements at Tempel 1. Top: Counting rates for all DFMI counters as arefer to mass thresholds of the 20-cm2 PVDF sensor. m1–m2 are thresholds for the 200-sensors. Bottom: Cumulative mass distribution of dust particles registered by DFMI in thdistribution index of a = 0.65, where the number of particles of mass greater than m, N(

clouds of fragments (Green et al., 2004). As was the case for Wild 2,it appears that at Tempel 1 the steady emission of small (micron-sized) particles from active areas is of secondary importance com-pared to the emission of aggregates. Additional evidence for emis-sion of aggregates has recently been obtained from images ofHartley 2 from the EPOXI mission (A’Hearn et al., 2011). Possibleevidence of aggregate fragmentation within the coma of CometHalley is discussed by Simpson et al. (1987).

The mass distribution of particles observed at Tempel 1 isshown in Fig. 11, bottom. As for Comet Wild 2 and Comet Halley,the total dust mass is dominated by the larger particles. As ex-pected for a dust coma dominated by fragmenting dust aggregates,the mass distribution is highly variable along the trajectory.

11. Dust composition: CIDA measurements

The Cometary and Interstellar Dust Analyzer is a time-of-flightmass spectrometer designed to analyze the chemical composition

function of time from the closest approach. The flyby speed was 10.9 km/s. m1–m3cm2 PVDF sensor. The other thresholds shown (AC1–AC4) are those of the acoustice inner coma during the encounter with Comet Tempel 1. The best fit overall mass>m) = km�a, is somewhat lower than that found for Wild 2 by the same detectors.

434 J. Veverka et al. / Icarus 222 (2013) 424–435

of dust grains impacting the instrument’s silver target (Kisselet al., 2003). Typical encounter speeds are sufficient to ionize afraction of each impacting dust grain. This instrument was usedat the flyby of Wild 2 in 2004 (Kissel et al., 2004) and is similarto those used during encounters with Comet Halley in 1986(Kissel et al., 1986a, 1986b). Previous measurements were carriedout using positive ions, but later studies showed that importantinformation on the organic component of dust grains can beobtained from negative ions (Krueger et al., 2004). Accordingly,during the encounter with Tempel 1, CIDA was operated in thenegative ion mode on the inbound leg and switched to its posi-tive ion mode at closest approach. The instrument operated fromE � 2 h to E + 2 h (±78,000 km of the nucleus). CIDA recorded atotal of 80 spectra, 46 in the negative ion mode and 34 in the po-sitive mode. Prominent peaks due to H� (mass number 1) andCN� (mass number 26) are evident in the data. Many spectrashow long tails at high mass numbers indicative of the presenceof complex molecules.

12. Conclusion

What have we learned about comet nuclei from the flybys ofBorrelly, Wild 2, Tempel 1, and Hartley 2?

Possibly the most important fact is these nuclei show significantmorphological diversity—we are far from knowing what a ‘‘typical’’Jupiter family comet looks like. While significant portions of thesurfaces of Wild 2 and Tempel 1 are pitted, this does not appearto be the case for Borrelly and Hartley 2. Furthermore, the pitson Wild 2 and Tempel 1, while they display a similar size—fre-quency distribution, are morphologically distinct: it is very easyto distinguish images of pitted terrain on Wild 2 from those ofTempel 1.

The smooth flows that cover some thirty percent of the surfaceof Tempel 1 seem to be totally absent on Wild 2. Suggestions of apossibly similar feature occur in images of Borrelly (Soderblomet al., 2004) but due to the limited resolution of the Borrelly cover-age, a definitive identification cannot be made. It is true that asmooth region is observed on Hartley 2, but the context in whichit occurs makes it unlikely that it is the result of an eruptive eventsuch as those implicated in the formation of smooth flows in Tem-pel 1.

While some evidence of layering has been noted for Borrelly(Britt et al., 2004) as well as for Wild 2 and Hartley 2 (Thomaset al., 2013), only in the case of Tempel 1 is evidence of layeringdramatically evident on global scales. The common occurrence ofscarps on all four nuclei is consistent with some degree of layeringof the surfaces and suggests that scarp erosion, most likely drivenby volatile sublimation, is a common erosion process on theseobjects.

It is interesting that in spite of marked morphological differ-ence, the surfaces of these comets are uniformly black with verylow albedos and generally very restricted exposures of water ice.

The five comet nuclei explored by spacecraft to date (Halley,Borrelly, Wild 2, Hartley 2, and Tempel 1) show widely differentsurface characteristics indicating that processes by which cometnuclei evolve are varied and complex. Preserved on the surface ofTempel 1 is evidence of an astonishingly wide range of geologicalprocesses making Tempel 1 arguably the most geologically inter-esting and puzzling among the so-called ‘‘primitive bodies’’ (com-ets and asteroids) explored by spacecraft so far. The geologiccomplexity and the knowledge about the surface properties(occurrence of smooth, easily sampled areas) provided by Deep Im-pact and Stardust-NExT make Tempel 1 an ideal target for futuresample return missions.

13. Post script

Having successfully completed all aspects of the original Star-dust and of the Stardust-NExT missions, the spacecraft was shutdown on March 24, 2011. The spacecraft is in a 1.5-year solar orbitand will not come closer to Earth than 1.7 million km for at least100 years.

Acknowledgments

We thank our two reviewers for very helpful comments andsuggestions.

Stardust-NExT was supported by NASA through its DiscoveryProgram. The Science Team expresses its thanks and acknowledgesits debt to the Project Management and Navigation Teams at the JetPropulsion Laboratory, to the Deep Space Network (DSN), and tothe Spacecraft Team at Lockheed Martin Aerospace (LMA) in Den-ver. We record our special thanks to the world-wide network ofobservers for providing crucial observations of Tempel 1 to supportthe determination of the appropriate time-of-arrival at the comet.Part of the research described was carried out at JPL under contractwith NASA. O. Groussin’s participation in the project was sup-ported by the Centre Nationale d’Etudes Spatiales (CNES).

References

A’Hearn, M.F. et al., 2005. Deep Impact: Excavating Comet Tempel 1. Science 310,258–264.

A’Hearn, M.F. et al., 2011. EPOXI at Comet Hartley 2. Science 332, 1396–1400.Belton, M.J.S., 2010. Cometary activity, active areas, and a mechanism for collimated

outflow on 1P, 9P, 19P, and 81P. Icarus 210, 881–897.Belton, M.J.S., Melosh, J., 2009. Fluidization and multiphase transport of particulate

cometary material as an explanation of the smooth terrains and repetitiveoutbursts on 9P/Tempel 1. Icarus 200, 280–291.

Belton, M.J.S. et al., 2008. The internal structure of Jupiter family cometary nucleifrom Deep Impact observations: The Talps or ‘‘Layered Pile’’ model. Icarus 187,332–344.

Belton, M.J.S., Feldman, P.D., A’Hearn, M.F., 2008. Cometary cryo-volcanism: Sourceregions and a model for the UT 2005 June 14 and other mini-outbursts onComet 9P/Tempel 1. Icarus 198, 189–207.

Belton, M.J.S. et al., 2011. Stardust-NExT, Deep Impact, and the accelerating spin of9P/Tempel 1. Icarus 213, 345–368.

Belton et al., 2013. The origin of pits on 9P/Tempel 1 and the geologic signature ofoutbursts in Stardust-NExT images. Icarus 222, 477–486.

Britt, D.T., Boice, D.C., Buratti, B.J., Campins, H., Nelson, R.M., Oberst, J., Sandel, B.R.,Stern, S.A., Soderblom, L.A., Thomas, N., 2004. The morphology and surfaceprocesses of Comet 119/P Borrelly. Icarus 167, 45–53.

Brownlee, D.E. et al., 2004. The surface of young Jupiter family Comet Wild 2: Viewfrom the Stardust spacecraft. Science 304, 1764–1769.

Chesley et al., 2013. An Updated Rotation Model for Comet 9P/Tempel 1. Icarus 222,516–525.

Clark, B.C. et al., 2004. Release and fragmentation of aggregates to produceheterogeneous, lumpy coma streams. J. Geophys. Res. 109. http://dx.doi.org/10.1029/2004JE002319.

Duncan, T.L., Levision, H., Dones, L., 2004. Dynamical evolution of ecliptic comets.In: Festou, M., Keller, H.U., Weaver, H.A. (Eds.), Comets II. University of ArizonaPress, Tucson, pp. 139–204.

Economou et al. Icarus, in review.Farnham, T.L., 2009. Coma morphology of Jupiter-family comets. Planet. Space Sci.

57, 1192–1217.Farnham. Icarus, in review.Farnham et al. Icarus, in review.Green, S.F. et al., 2004. The dust mass distribution of Comet 81P/Wild 2. J. Geophys.

Res. 109. http://dx.doi.org/10.1029/2004JE002318.Keller, H.U., Kramm, R., Thomas, N., 1988. Surface features on the nucleus of Comet

Halley. Nature 331, 227–230.Kissel, J. et al., 1986a. Composition of Comet Halley dust particles from VEGA

observations. Nature 321, 280–282.Kissel, J. et al., 1986b. Composition of Comet Halley dust particles from GIOTTO

observations. Nature 321, 336–337.Kissel, J. et al., 2003. Cometary and Interstellar Dust Analyzer for Comet Wild 2. J.

Geophys. Res. 108 (E10), 8114. http://dx.doi.org/10.1029/2003JE002091.Kissel, J. et al., 2004. The Cometary and Interstellar Dust Analyzer at Comet 81P/

Wild 2. Science 304, 1774–1776.Klaasen et al., 2013. Stardust-NExT NAVCAM Calibration and Performance. Icarus

222, 436–452.

J. Veverka et al. / Icarus 222 (2013) 424–435 435

Krueger, F.R. et al., 2004. Quinone derivatives composing ‘interstellar grains’ due toboth polarity ions detected by the ‘Cometary and Interstellar Dust Analyser’(CIDA) onboard the spacecraft STARDUST. Rapid Commun. Mass Spectrom. 18,103–111.

Li, J.-Y. et al., 2007. Deep Impact photometry of Comet 9P/Tempel 1. Icarus 187,41–55.

Li et al. Icarus, in review.Lisse, C.M. et al., 2005. The coma of Comet 9P/Tempel 1. Space Sci. Rev. 117, 161–

192.Lisse, C.M. et al., 2006. Spitzer spectral observations of the Deep Impact ejecta.

Science 313, 635–640.Meech, K.J. et al., 2005. Deep Impact: Observations from a worldwide Earth-based

campaign. Science 310, 265–269.Meech, K.J. et al., 2011. Deep Impact, Stardust-NExT and the behavior of Comet

9P/Tempel 1 from 1997–2010. Icarus 213, 323–344.Newburn, R.L., Bhaskaran, S., Duxbury, T.C., Fraschetti, G., Radey, T., Schwochert, M.,

2003. Stardust imaging camera. J. Geophys. Res. 108 (E10), 8116.Richardson, J.E., 2009. Cratering saturation and equilibrium: A new model looks at

an old problem. Icarus 204, 697–715.Richardson, J.E. et al., 2007. A ballistics analysis of the Deep Impact ejecta plume:

Determining Comet Tempel 1’s gravity, mass, and density. Icarus 190, 357–390.Richard, Melosh, 2013. An examination of the Deep Impact collision site on comet

Tempel 1 via StardustûNExT: placing further constraints on cometary surfaceproperties Icarus 222, 492–501.

Schleicher, D.G., 2007. Deep Impact’s target 9P/Tempel 1 at multiple apparitions:Seasonal and secular variations in gas and dust production. Icarus 191, 322–338.

Schultz, P.H. et al., 2007. The Deep Impact oblique impact cratering experiment.Icarus 190, 295–333.

Schultz et al. Icarus, in review.Simpson, J.A., Rabinowitz, D., Tuzzolino, A.J., Ksanformality, L.V., Sagdeev, R.Z., 1987.

The dust coma of Comet P/Halley: Measurements on the Vega-1 and Vega-2spacecraft. Astron. Astrophys. 187, 742–752.

Soderblom, L.A. et al., 2004. Imaging Borrelly. Icarus 167, 4–15.Sunshine, J. et al., 2006. Exposed water ice deposits on the surface of Comet Tempel

1. Science 311, 1453–1455.Thomas, P.C. et al., 2007. The shape, topography, and geology of Tempel 1 from Deep

Impact observations. Icarus 187, 4–15.Thomas et al., 2013. The Nucleus of Comet 9P/Tempel 1: Shape and Geology from

Two Flybys. Icarus 222, 453–466.Tuzzolino, A.J. et al., 2003. Dust Flux Monitor Instrument for the Stardust mission to

Comet Wild 2. J. Geophys. Res. 108. http://dx.doi.org/10.1029/2003JE002086.Tuzzolino, A.J. et al., 2004. Dust measurements in the coma of Comet 81P/Wild 2 by

the Dust Flux Monitor Instrument. Science 304, 1776–1780.Vasundhara, R., 2009. Investigations of pre-Deep Impact morphology of the dust

coma of Comet Tempel 1. Icarus 204, 194–208.Vincent, J.B., Boehnhardt, H., Lara, L.M., 2010. A numerical model of cometary dust

structures. Application to Comet 9P/Tempel 1. Astron. Astrophys. 512, A60,http://dx.doi.org/10.1051/0004-6361/200913418.

Wellnitz et al. Icarus, in review.Yeomans, D.K., Giorgini, J.D., Chesley, S.R., 2005. The history and dynamics of Comet

9P/Tempel 1. Space Sci. Rev. 117, 123–135.