Research Article

Dunes on Pluto

See allHide authors and affiliations

Science  01 Jun 2018:
Vol. 360, Issue 6392, pp. 992-997
DOI: 10.1126/science.aao2975

Methane ice dunes on Pluto

Wind-blown sand or ice dunes are known on Earth, Mars, Venus, Titan, and comet 67P/Churyumov-Gerasimenko. Telfer et al. used images taken by the New Horizons spacecraft to identify dunes in the Sputnik Planitia region on Pluto (see the Perspective by Hayes). Modeling shows that these dunes could be formed by sand-sized grains of solid methane ice transported in typical Pluto winds. The methane grains could have been lofted into the atmosphere by the melting of surrounding nitrogen ice or blown down from nearby mountains. Understanding how dunes form under Pluto conditions will help with interpreting similar features found elsewhere in the solar system.

Science, this issue p. 992; see also p. 960


The surface of Pluto is more geologically diverse and dynamic than had been expected, but the role of its tenuous atmosphere in shaping the landscape remains unclear. We describe observations from the New Horizons spacecraft of regularly spaced, linear ridges whose morphology, distribution, and orientation are consistent with being transverse dunes. These are located close to mountainous regions and are orthogonal to nearby wind streaks. We demonstrate that the wavelength of the dunes (~0.4 to 1 kilometer) is best explained by the deposition of sand-sized (~200 to ~300 micrometer) particles of methane ice in moderate winds (<10 meters per second). The undisturbed morphology of the dunes, and relationships with the underlying convective glacial ice, imply that the dunes have formed in the very recent geological past.

Dunes require a supply of particulate material on a surface and a fluid boundary layer to entrain the grains (i.e., wind, for dunes on a planet’s surface). They have been identified in some surprising locations: Contrary to predictions (1), Saturn’s moon Titan has a broad belt of linear dunes encircling its equatorial latitudes (2), and despite the lack of a persistent atmosphere, eolian landforms (i.e., those related to wind) have also been suggested to occur on comet 67P/Churyumov-Gerasimenko (3). On 14 July 2015, NASA’s New Horizons spacecraft flew past Pluto, which provided spectral data and imagery of the surface at resolutions as detailed as 80 m/pixel (4). The combination of Pluto’s low gravity (0.62 m s−1, or Embedded Image that of Earth), sparse atmosphere [1 Pa (5)], extreme cold [~45 K (5)], and surface composition [N2, CO, H2O, and CH4 ices (6)] made pre-encounter predictions of surface processes challenging. However, pre-encounter speculation included that eolian processes, and potentially dunes, might be found on Pluto (7), because, despite the relatively thin atmosphere, the winds could possibly sustain saltation (i.e., particle movement by ballistic hops) in the current surface conditions. We examined images from the Long Range Reconnaissance Imager (LORRI) instrument (8) on New Horizons, taken during the probe’s closest approach to Pluto, to search for landforms with the morphological and distributional characteristics of dunes. We also searched spectroscopic data from the Multispectral Visible Imaging Camera [MVIC (9)] for evidence of sufficient sand-sized ice particles to form dunes, and discuss how sublimation may play a role in lofting these particles, enabling them to be saltated into dunes.

Observations from New Horizons

The surface of Pluto, as revealed by New Horizons, is diverse in its range of landforms, composition, and age (4, 10). One of the largest features, Sputnik Planitia (SP), is a plain of N2, CO, and CH4 ice [(6) and fig. S1] that extends across Pluto’s tropics and at its widest point covers 30° of longitude (Fig. 1A). Polygonal features on the surface of SP, tens of kilometers across and bounded by trenches up to 100 m deep (Fig. 1, B and C) have been interpreted as the result of thermally driven, convective overturning of the ice (11, 12), which, together with the uncratered surface of SP (4), suggests a geologically young [<500 thousand years (ka)] (11, 12) and active surface. Much of the western edge of the ice is bounded by the Al-Idrisi Montes (AIM), a mountainous region with relief of up to 5 km. On the SP plain bordering these mountains, distinct, regularly spaced, linear ridges are evident within a belt of ~75 km from the mountain margin (Fig. 2A). They have positive relief as evident from shadows consistent with the mountains. The ridges show pronounced spatial regularity (~0.4 to 1 km wavelength), substantial length/width ratios (sometimes >20 km length), consistent shape along these lengths, and the presence of merging and bifurcation junctions (Figs. 1, C and D, and 2, D and E). These junctions are approximately evenly spread between 47 north-facing bifurcations and 42 south-facing splits, and there is no clear spatial patterning to the direction of junctions. Farther from the mountain margin, toward the southeast, the ridges become more widely spaced and generally larger, while still in isolated fields or patches. Dark streaks are also found across the surface of the ice, typically behind topographic obstacles, and have been interpreted as wind streaks (4). These features indicate that there are loose particles near and on the surface, as the streaks are thought to result from the deposition of suspended, fine particles in the lee of obstacles to wind flow (4, 5, 13) (Fig. 1, C and E).

Fig. 1 New Horizons flyby imagery of landforms attributed to eolian origins.

All images are unrectified, and thus all scales are approximate. Color-composite MVIC images are shown here for context; dune identification was performed on grayscale LORRI images (shown below). (A) Overview of Pluto centered on ~25° latitude, ~165° longitude, showing the locations of images (B) and (E) and Fig. 3A and fig. S3 (47). (B) The spatial context for SP and the AIM mountains to the west (48). Insets (C) and (D) show details of the highly regular spatial patterning, which we attribute to eolian dune formation, and two newly identified wind streaks (arrows x), along the margins of the SP/AIM border. Here the dunes show characteristic bifurcations (arrows y) and a superposition with SP’s polygonal patterning (arrow z), suggesting a youthful age for these features (49). (E) Two further wind streaks on the surface (x’), downwind of the Coleta de Dados Colles (4). These wind streaks, farther from the SP/AIM margin, are oriented differently than those close to the icefield’s edge and are still roughly orthogonal to the dunes there (50).

Fig. 2 Identified features.

(A) Dunes (black lines) at the margins of western Sputnik Planitia. Prominent wind streaks are marked with orange lines. (B) Radial plot of the orientation of the dunes (n = 331), and the direction orthogonal to the wind implied by the wind streaks close to the SP/AIM margin (orange dashed line; n = 4; arithmetic mean, Embedded Image = 203°). Because the dunes have a distinct shift in orientation (fig. S1), the distribution of dunes in the three patches closest to the wind streaks within the dunefield [outlined in dashed green in (A)] has been separately highlighted on the radial plot, in green. These have a mean orientation of 204° (n = 77), highlighted by the dashed green line. The dark blue line indicates the mean trend of the border of SP and the Al-Idrisi Montes in this area (194°). (C) Frequency of dune spacings in clusters close to [red line representing dunes within the red dashed line of (A)] and far from [green line representing dunes within the green dashed line of (A)] the icefield/mountain interface. Dunes farther from the mountains are more widely spaced (Embedded Image = 700 m) than those close to the mountains (Embedded Image = 560 m). (D) Detail of the image interpretation process of the highest-resolution swath, showing linear ridges, which sometimes bifurcate but are otherwise notable for their regularity. (E) The same image with ridge lines highlighted.

We have identified 357 pale-colored, linear ridges on SP adjacent to the AIM (Fig. 2, A to C), as well as six darker wind streaks in addition to the seven previously identified (4). The ridges closest to the SP/AIM mountain front are oriented approximately parallel with it, and ridges farther to the southeast shift orientation clockwise by ~30° over a distance of ~75 km (Fig. 2, A and B); the ridges farther from the SP/AIM margin are significantly (Mann-Whitney U = −7.41; p < 0.0001) more widely spaced (Fig. 2C). Beyond the ~75-km-wide belt in which the linear ridges are predominantly found, the morphology of the surface changes, with preferential alignment of the ridges gradually disappearing (fig. S2), until the landscape is dominated by weakly aligned or unaligned, but still regularly dispersed, pits likely caused by sublimation of the ice (14). Wind streaks adjacent to the SP/AIM border are perpendicular to the ridges and mimic the shift in orientation shown by the ridges (Fig. 3, A and B). Streaks within the zone in which the ridges are found (i.e., <75 km from the SP/AIM border) are geographically (i.e., clockwise from north) oriented 113 ± 4° (1 standard deviation, σ, with sample number n = 4), while more distant wind streaks are oriented significantly [(heteroscedastic Student’s t = 9.912; p < 0.001 (Fig. 3B)] differently at 153 ± 10° (1σ, n = 9).

Fig. 3 The western margin of Sputnik Planitia.

(A) Transverse dunes are shown in black, the margin of the icefield and neighboring Al-Idrisi Montes to the northwest in blue, wind streaks close to this margin in orange, and further wind streaks farther from the mountains in yellow. There is an orientation shift between the two sets of wind streaks, matching the correlation between the distance to the margin of the icefield and mountains, and the orientation of the transverse dunes [(B); wind streaks in orange]. We interpret this as topography and/or surface composition influencing regional wind regimes.

Interpretation as dunes

The ridges found on western SP have morphological similarities to dunes (Figs. 1, C and D, and 4, A to C). In addition to analog similarities, we argue that these landforms are most consistent with an initial eolian depositional origin (i.e., dunes) on the grounds that (i) a depositional origin is favored by the superimposition of many of the dunes on the trenches bordering SP’s convective cells (Figs. 1D and 2, D and E); (ii) the distribution of the dunes with pattern coarsening (enlarging toward the southeast), away from the mountains (Fig. 4C), is characteristic of dunefields; (iii) their orientation, and systematic regional changes to this orientation, are more readily explained by the wind regime than variations in incoming solar radiation; (iv) the presence of pronounced wind streaks, orthogonal to the dunes, demonstrates the potential efficacy of Pluto’s winds; (v) their location, on a methane- and nitrogen-dominated ice cap adjacent to mountains, is where the strongest winds and a supply of sediment might be expected; and (vi) their differing morphologies and undeformed regular alignment differ from the randomly aligned, shallow pits that border on the dune regions of SP (Fig. 4E and fig. S2) and the deeply incised, discrete, aligned pits that can be found toward SP’s southern and eastern margins (Fig. 4D). These pit-like features are morphologically distinct from the dune-like ridges farther north near the AIM that we discuss here (Fig. 4, A and D). To test this hypothesis, we use a model (15) to examine the saltation of sand-sized (in this case, ~200 to 300 μm) particles on Pluto. Once initiated, the model indicates that saltation can be sustained even under the low (Earth-like; 1 to 10 m s−1) wind speeds predicted at the surface today (16). However, the model also suggests that an additional process may be necessary to initially loft particles (15). This can be accomplished by sublimation, which is capable of lofting particles, and we model this process to find that particles can be entrained. This function of sublimation is in addition to the role sublimation may play in eroding mature dunes to more altered forms, which is also discussed in more detail below. Thus, under the current conditions, if there are sufficiently noncohesive sand-sized particulates on the surface of Pluto, we should expect to find dunes.

Fig. 4 Analogs and comparison with sublimation features.

(A) Details of the dunes on western Sputnik Planitia, centered on 34.35° 159.84° (location shown in Fig. 1). (B) Analogous terrestrial transverse dunes of the Taklamakan Desert, western China [Image credit: image CNRS/SPOT, DigitalGlobe and courtesy of Google Earth], and (C) the same location down-sampled to a relative resolution similar to that of the Pluto dunes (i.e., ~5 to 10 pixels per crest-crest spacing). (D) The aligned and distorted sublimation features abundant on southern and eastern SP (image centered on −4.78° 189.48°) and (E) weakly aligned to randomly oriented, shallow sublimation pits. (F) An example of a landscape revealing both eolian and sublimation-derived landforms at Mars’ southern polar ice cap from the Mars Reconnaissance Orbiter reveals both dark eolian bedforms (dunes and ripples), as well as sublimation pits developing in the underlying CO2 ice. [Image credit: NASA/JPL/University of Arizona, ESP_014342_0930_RED]

Terrestrial and planetary dunes that form straight ridges can occur either perpendicular to the wind, forming transverse dunes, or parallel to the net local wind regime, forming longitudinal (or linear) dunes, and regional variation in the alignment of such dunes on Earth is typically associated with meso- or large-scale atmospheric patterns (17). Wind streaks are well known on Venus and Mars, are present even under the tenuous atmosphere of Triton, and are generally considered to represent the wind direction (13). The presence of pronounced wind streaks (Fig. 1C) within the dunefield, very nearly orthogonal to the dune trends, suggests that the observed dunes are transverse forms (Fig. 3A). The transverse nature of the dunes is further supported by the lack of consistency in bifurcation orientation; within dunefields oriented parallel to net sediment-transporting winds, such defects tend to cluster in terms of their orientation (18). The transverse orientation also shows that these ridges cannot be sastrugi (erosional snow ridges that form parallel to net winds) (19), or other erosional features analogous to yardangs (wind-carved ridges). The dunes in the northwestern portion of SP/AIM (Fig. 4A) are even more regularly spaced and parallel than many transverse dunes on Earth. Possible explanations for this include a highly consistent wind regime, lack of topographic deflection of winds, or a smooth substrate.

Conditions for the formation of dunes and sublimation pits

The existence of dunes on the surface of Pluto requires three necessary criteria to be met. First, there must be a fluid atmosphere of sufficient density to make eolian transport possible. Second, there must be a granular material of a size and density, and with sufficiently low cohesion, that it can be entrained by winds. On Earth, this role is typically played by sand-sized mineral grains of a variety of compositions, including snow and ice. Third, given the high wind speeds needed to lift surface particles against cohesion forces (20) (Fig. 5), a specific mechanism must exist to loft large quantities of ice particles into the atmosphere where they are available for eolian transport. The presence of these criteria alone is necessary but not sufficient to identify the surface features as dunes. To justify our interpretation of these features as being dunes, we also examine the conditions required for the other most likely candidate: aligned sublimation pits.

Fig. 5 Minimal threshold wind speeds.

The wind speeds required for initiation (Uft, orange line) and continuation (Ut, black line) of saltation on Pluto, at a reference height of 10 m above the soil, were computed for different values of the average particle diameter (15). The dashed horizontal line indicates maximum likely wind speeds at Pluto’s surface.


The orientations of the dunes and the wind streaks change locally, and consistently; in the case of the dunes, over a distance on the order of 10 to 102 km. This implies that the topography and/or surface composition has influenced the local wind regime, as was anticipated (21). These orientations are consistent with sublimation-driven and topographic mechanisms for the horizontal displacement of the atmosphere, as winds are generated by a gravity-driven flow toward lower regions. Modeling of Pluto’s current atmosphere suggests that surface winds on the order of 1 to 10 ms−1 are possible; they should be strongest where there are topographic gradients and when driven by sublimation of surface ices by sunlight (15). The location of the dunes at the western margins of the SP and AIM should thus be among the windiest locations on the known regions of Pluto. As with Earth and Mars, once grain transport along the surface of Pluto has begun, increased efficacy of grain-splash (the ejection of new particles due to grains in saltation colliding with the ground) promotes a hysteretic effect that further sustains sediment flux (22, 23). We use a numerical model (15) to demonstrate that despite the high wind speeds needed for initial eolian entrainment, eolian transport can, once established, be sustained with wind speeds of ~10 m s−1 (Fig. 5).

Sediment supply

Although terrestrial dunes are typically associated with quartz, basalt, or gypsum sand, other materials can form the grains for dune development. Snow dunes of a very large scale are observed in the center of the Antarctic continent (24); on Titan, it is generally assumed to be atmosphere-derived organics, perhaps initially tholins, which form the equatorial belt of giant dunes (25). Whereas tholins are thought to form the dark patches of Pluto’s equatorial regions (6), the dunes evident on SP are light in color and are thus not formed from the same complex, organic, photochemically derived haze seen in Pluto’s atmosphere (4). The most likely candidates are thus N2 and CH4 ices. The surface of SP has generally been interpreted as predominantly composed of N2 ice (4, 5), just as solidified nitrogen snows are believed to account for Triton’s ice-covered surface (26, 27). The zone in which the ridges occur is coincident with the latitudes in which net N2 condensation occurs over the course of a Pluto year (16, 28) (fig. S3). However, recent analyses suggest that the composition may be a more complex mix of N2, CH4, and CO ices (29). Our analysis of data from the MVIC instrument, using a CH4 filter (15), suggests that the location of the ridges and streaks coincides with a region of enhanced CH4 ice content (fig. S4). To the west of SP, the Enrique Montes in Cthulhu Macula (CM) have been shown to be capped with methane, presumably as the result of condensation or precipitation (29, 30). CH4 ice retains hardness and rigidity under Pluto surface conditions, which is ideal for saltation and dune formation, whereas N2 ice is likely to be softer. These constraints lead us to conclude that the dunes are formed predominantly of grains of methane ice, though we do not rule out that there could also be a nitrogen ice component. The presence of transverse forms, indicating sediment-rich local conditions, as opposed to more sediment-starved isolated barchans (discrete, crescentic dunes), suggests that locally, the sediment supply to this region of SP must be, or must have been, abundant. Given the strength of the color and boundary delineations of methane in the AIM (fig. S3), the methane ice may be quite thick and perhaps similar to valley glaciers in these isolated regions. If such high-altitude methane snowpack is a regular, seasonal occurrence, this may be a substantial reservoir from which to derive the abundant sand across the northwestern surface of SP required to form these transverse dunes.

Grain size

Credible sediment sizes are required for dune formation under the likely eolian regime. The grain sizes proposed for nitrogen ices (e.g., on Triton) have varied from micrometer (31, 32) to meter scale (33). We develop a method (15) to approximately constrain average grain size (d) and formative wind speed (U) from the mean crest-to-crest distance, or wavelength (λ), of the transverse dunes. For eolian dunes, the relevant length scale controlling this wavelength is the saturation length (Lsat) of the sediment flux, which is the distance needed by the flux to adapt to a change in local flow conditions. By combining theory (34), which predicts Lsat as a function of wind speed and attributes of sediment and atmosphere, with a mathematical model (35) for λ as a function of Lsat and U, we obtain the values of d and U that are consistent with λ. These values are shown in fig. S5, for λ ≈ 700 m and λ ≈ 560 m, which correspond to the transverse dunes far from and near to the mountainous area of Fig. 1, respectively. Given that expected formative wind speeds on Pluto are not larger than 10 m s−1 (16), fig. S5 implies that grain size does not exceed ~370 μm and is most probably in the range between 210 and 310 μm. The spectral response of the MVIC CH4 filter offers an additional constraint on the possible grain sizes observed, as Hapke modeling of the scattering within a granular medium provides a grain size–dependent control on the equivalent width of the absorption band. We find (15) that the observed response is consistent with a granular medium of ~200 to 300 μm.


Although eolian transport can be maintained under Pluto’s current wind regime, the speeds necessary for initial entrainment are orders of magnitude greater than those believed to be present at Pluto’s surface (Fig. 5). An additional process is thus likely to be necessary to initiate eolian activity. In Sputnik Planitia, this process may be related to the intense, solar-driven sublimation of surface ices that injects more than 103 m3 m−2 of gas into the atmosphere every afternoon [figure 8 of (16)]. When sunlight penetrates through, the upper layers of semitransparent ice particles are lofted, sometimes at high vertical velocities, due to a mechanism often referred to as a solid-state greenhouse (36). Therefore, initial entrainment of ice grains that eventually form dunes may result from sublimating subsurface ice, as has been observed in the thin atmospheres of Mars’ northern polar region (37) and proposed for comet 67P/Churyumov-Gerasimenko (38, 39) and Triton (24, 25). Modeling (15) suggests that subsurface N2 sublimation under Pluto surface conditions is capable of lofting even the densest candidate particles [N2 ice, at 1030 kg m−3, is denser than CH4 ice, at 494 kg m−3 (40)] with sizes ~200 μm, even at 0.1 Pa; within the range of both Pluto’s atmospheric pressure and the solid N2 vapor pressure. Surface ices of mixed composition offer an additional potential mechanism for facilitating grain lofting. At the nitrogen frost point temperature of 63 K (11), pure methane ice particles mixed with nitrogen should not sublimate at all. As methane particles are slightly heated by the Sun, they should enable the sublimation of the nitrogen ice that they touch and thus be readily lofted into the atmosphere. Similar processes have recently been suggested for the migration of tholin deposits on the surface of Pluto (41). Past periods of higher atmospheric pressures, which have been suggested (42), could facilitate initial entrainment due to increased efficacy of eolian processes.


The landscape of SP contains evidence of sublimation-driven landforms (4, 7, 14), and this process is important in shaping parts of Pluto’s surface. We consider whether the landforms described here are more consistent with origins attributable to eolian or sublimation processes. Locally, sublimation pits are deeply incised and may align to form linear troughs up to tens of kilometers long and up to ~1 km deep (4, 7). Frequently, and especially toward the southern and eastern margins of SP, any alignment is subsequently heavily deformed, presumably driven by glacial flow and convective overturning. Analog landforms on Earth are provided by sublimation-driven textures of snow and ice surfaces: ablation hollows (suncups) and penitentes (14, 43). On Earth, ablation hollows on snow may become aligned to leave ridges (penitentes), which align themselves to within ±30° of east-west (i.e., the annual mean net Sun path) (44). Although the orientation of any penitentes on Pluto is likely to be more complex and seasonally dependent, they are only likely to form wavelengths in excess of ~1000 m (45). It is possible that sublimation has acted upon already formed dunes in some regions (Fig. 2A). In polar regions on Earth, wind-driven snow or ice grains can produce dunes, which then become hardened by sintering and begin to undergo modification by wind and sublimation processes, thus changing from depositional to erosional landforms (44, 45). Given the tendency of ices to sinter together under the right conditions, this could also happen in the CH4 or N2 ices of Pluto’s dunes. Sublimation erosion of Pluto’s dunes may enlarge and round the areas between the dunes, and sharpen the dune crests while preserving the overall dune orientation and spacing. This morphology may be seen just at the resolution limit in the features farthest from the mountains in Fig. 2A (enlarged view in fig. S2). This is supported by modeling (15) of the net accumulation of ices across Pluto’s surface during the past two (Earth) centuries (fig. S4), which suggests that for the past ~30 Earth years, the dunefield has been experiencing net sublimation. Some of these features may have progressed so far toward being erosional that we have not identified them as dunes (fig. S2).


An upper limit on the age of the dunes, which sit atop the ice of the western margins of SP, is imposed by the recycling rate of the upper surface of the convectional cells within the ice (i.e., <500 ka) (11, 12). This overturning of the substrate, inferred from the complete absence of identified craters on SP, provides an age constraint for superficial landforms that is not available for dunes on other solar system bodies and implies a geologically and/or geomorphologically active surface (4, 10, 46). Surface features, undistorted by the convectional overturning within the ice, must be much younger than the time scales of convection, and therefore closer to the time scales of Pluto’s strong seasons (i.e., terrestrial decades to centuries). Further evidence that the dunes form on a time scale substantially shorter than that of the convection is suggested by the superposition of the dunes over the depressions at the cell margins (Fig. 1G).

Summary and conclusions

We have presented evidence that the highlands adjacent to SP accumulate methane. The ridged, dune-like landforms nearby, and accompanying wind streaks, are rich in methane relative to their underlying substrate. Although the wind speeds needed for eolian entrainment are higher than the likely wind speeds present on the surface, sublimation provides a credible mechanism for lofting grains. Numerical sediment transport and spectral modeling suggest that these methane grains are ~200 to 300 μm. Our models suggest that eolian transport is highly effective under Pluto surface conditions once initiated. An ample sediment supply appears to be available from a seasonally abundant snowpack in the adjacent mountains. The result is the formation of transverse dunes, as we identify in the images from New Horizons. The orientation of the dunes perpendicular to the wind is supported by the local topography and surface, and accompanying wind streaks. The presence of these dunes indicates an active atmosphere that produces geologically young landforms.

Supplementary Materials

The New Horizons Geology, Geophysics and Imaging Science Theme Team

Materials and Methods

Figs. S1 to S5

References (5169)

References and Notes

  1. Materials and methods are available as supplementary materials.
Acknowledgments: We thank everyone involved, from concept to data retrieval, with the New Horizons mission. This research has made use of the USGS Integrated Software for Imagers and Spectrometers (ISIS). Funding: E.J.R.P. thanks the German Research Foundation for grant RI2497/3-1. All New Horizons team member authors are funded by the NASA New Horizons Project. Author contributions: M.W.T. conducted the spatial analysis and image interpretation, coordinated the research, and co-wrote the paper. E.J.R.P. developed and conducted the numerical modeling and co-wrote the paper. J.R. coordinated the research and co-wrote the paper. R.A.B. produced and provided LORRI mosaicking. T.B. and F.F. provided data on surface/atmosphere exchanges. F.N. performed calculations on the effectiveness of sublimation modeling. W.M.G. conducted the Hapke modeling. J.M.M., S.A.S., and J.S. contributed to the manuscript. T.R.L. produced and provided LORRI mosaicking. R.P.B. and A.M.E. provided circulation model data. H.A.W., C.B.O., L.A.Y., and K.E. are project scientists and contributed to the manuscript. K.R. provided discussion of ideas. Competing interests: There are no competing interests to declare. Data and materials availability: The LORRI data are archived in the Planetary Data System (PDS) Small Bodies Node at MVIC data are available via the PDS at

Stay Connected to Science

Navigate This Article