Research Article

Impact craters on Pluto and Charon indicate a deficit of small Kuiper belt objects

See allHide authors and affiliations

Science  01 Mar 2019:
Vol. 363, Issue 6430, pp. 955-959
DOI: 10.1126/science.aap8628

Impact craters on Pluto and Charon

Collisions between Solar System bodies produce impact craters on large objects at a rate that depends on the population of impacting small bodies. Singer et al. examined impact craters on Pluto and its moon Charon. Some regions have had their impact craters erased by recent geological processes, but others appear to record 4 billion years of impacts. Because Pluto and Charon are located in the Kuiper belt, the distribution of crater sizes reflects the size distribution of impacting Kuiper belt objects (KBOs). The authors found fewer small KBOs than predicted by models of collision equilibrium, implying that some of the KBO population has been preserved since the formation of the Solar System.

Science, this issue p. 955


The flyby of Pluto and Charon by the New Horizons spacecraft provided high-resolution images of cratered surfaces embedded in the Kuiper belt, an extensive region of bodies orbiting beyond Neptune. Impact craters on Pluto and Charon were formed by collisions with other Kuiper belt objects (KBOs) with diameters from ~40 kilometers to ~300 meters, smaller than most KBOs observed directly by telescopes. We find a relative paucity of small craters ≲13 kilometers in diameter, which cannot be explained solely by geological resurfacing. This implies a deficit of small KBOs (≲1 to 2 kilometers in diameter). Some surfaces on Pluto and Charon are likely ≳4 billion years old, thus their crater records provide information on the size-frequency distribution of KBOs in the early Solar System.

The size-frequency distribution (SFD) of Kuiper belt objects (KBOs) with diameters (d) smaller than 100 km provides empirical constraints on the formation and evolution of bodies in the region beyond Neptune. The smaller bodies in the Kuiper belt are remnants of Solar System evolution and may have persisted largely unaltered for the past 4 billion years or longer [e.g., (1)]. These objects also form an impactor population, producing craters seen on the younger surfaces of outer Solar System bodies [e.g., (26)]. Constraining the primary impactor KBO size distribution can aid understanding of surface and interior evolution on these worlds. The KBO size distribution carries implications for whether comets (such as 67P/Churyumov–Gerasimenko) are primordial or substantially altered by collisions over their lifetimes [e.g., (79)], as well as for the number of small bodies discovered by telescopic campaigns [e.g., (10)].

Telescopic observations of smaller KBOs are difficult, so limited numbers <100 km in diameter have been confidently identified [e.g., (1114)]. With few direct observations, predictions of the KBO population generally extrapolate from ~100-km-diameter objects down to smaller sizes by using a variety of plausible SFDs [e.g., (6)]. Some predictions are informed by the impact crater SFDs observed on the icy satellites of Jupiter and Saturn, which reflect various combinations of impactor populations (2, 15, 16). Accretion and dynamical models of Kuiper belt formation predict either that (i) many small and midsized objects (d ≲ 100 km) form by hierarchical coagulation followed by runaway growth and collisional evolution (17), or (ii) far fewer small objects form when streaming or other aero-gravitational instabilities preferentially create larger bodies during the lifetime of the solar nebula [e.g., (1821)]. The New Horizons flyby of the Pluto system (22) provides an opportunity to characterize the KBO population at small sizes by examining impact craters in the system.

Impact crater distributions

New Horizons imaged ~40% each of the surfaces of Pluto (~7 × 106 km2) and Charon (~1.8 × 106 km2), hereafter P&C, at moderate to high resolution during the 2015 flyby; these areas are referred to as their encounter hemispheres. The observed image scales range from 76 to 850 m pixel−1 (px−1) for Pluto, and from 154 to 865 m px−1 for Charon, with the higher-resolution datasets covering smaller areas (table S1 and figs. S1 and S2). We mapped craters on each image sequence or scan of approximately consistent pixel scale independently, to allow for comparisons across image sets (23).

We consider craters in several broad regions on P&C, roughly delineated on the basis of morphologic or albedo criteria (figs. S1 and S2). The western and eastern sides of Pluto’s encounter hemisphere are geologically different, and although terrains generally fall into broad latitudinal bands, they are not continuous across the encounter hemisphere. We divide Charon into two regions—the northern terrains (Oz Terra) and the southern Vulcan Planitia (VP; Fig. 1)—noting that some place names in this report are informal. VP is a large plain covering the southern portion of Charon’s encounter hemisphere, which appears to have been completely (or nearly completely) resurfaced through icy volcanism, likely early in Charon’s history (24). We focus on VP as it is highly favorable for crater mapping because of (i) near-terminator lighting that emphasizes topography such as crater rims; (ii) a mostly smooth initial surface, which aids in feature discrimination; and (iii) a crater areal density far from saturation (i.e., craters with their ejecta are not strongly overlapping, which could obliterate and/or obscure one another). We prefer craters on VP as the clearest representation of the impactor flux, because these craters appear little modified or interrupted by younger craters or any subsequent geologic activity (Fig. 1).

Fig. 1 Geology of Charon.

(A) Vulcan Planitia (VP); (B) high-resolution view of VP with smoother regions; (C) craters with hummocky floors; (D and E) the only two craters on VP that may have been cut by faults, or alternatively, whose formation was affected by preexisting faults; (F) partially filled craters (arrows) just north of the large chasmata separating the main expanse of VP from the northern terrains (Oz terra); and (G and H) views of small craters on VP. Arrows indicate small craters that were mapped, but only those larger than 1.4 km (10 pixels) were included in the analysis. Image resolutions and IDs (see table S1): (A) mosaic of PELR_C_LORRI (865 m px−1) and PELR_C_LEISA_HIRES (234 m px−1); (G, D, H) PELR_C_LORRI_MVIC_CA (154 m px−1); (C, D, E, F) PEMV_C_LORRI_MVIC_CA (622 m px−1). Images are shown north upwards except (B), where the orientation is indicated by the north arrow. Red rectangles indicate areas enlarged in other panels.

We quantify the crater populations and examine their geologic context to obtain information about the small-body population (KBOs) that impacted P&C. The differential number of craters per diameter (D) bin in a crater SFD is expressed as dN/dDDq (25, 26), where q is the power-law index or log-log slope. We use the symbol D to refer to crater diameters and the symbol d to refer to impactor diameters. The relative or R-plot SFD divides the differential SFD by D−3 (23) such that a differential distribution with q = −3 is a horizontal line, whereas q = −4 and −2 form lines that slope downward and upward with increasing D, respectively (Fig. 2). The R-plot helps distinguish changes in slope as a function of diameter and between crater populations. We normalize the number of craters per bin by the mapped area to give the relative density of craters per size bin (23).

Fig. 2 Regional crater size-frequency distributions.

(A) Pluto and (B) Charon SFDs shown as relative differential, or R-plots, which display crater spatial densities normalized to a D−3 distribution (see text), for areas with more than 10 craters. In (B), the highest and lowest Pluto SFDs are overlain in light gray for comparison with the Charon SFDs. Legend titles refer to the observation name and mapped region, as listed in table S1 and shown in figs. S1 and S2. For reference, the additional axis below (B) shows an approximate impactor diameter (based on water ice density for impactor and target) and approximate equivalent g-band H-magnitude, a measure of brightness used in telescopic observations of KBOs (23). Dashed lines indicate slopes of differential SFD power-law q values of −2, −3, and −4. Error bars represent Poisson statistical uncertainties (26).

SFDs for both Pluto and Charon exhibit differential slopes (q) near −3 for D ≳15 km to D ~100 km (there are few craters larger than this on either body) (Fig. 2 and table S2). However, for craters ≲15 km in diameter, the SFDs for both P&C transition to a much shallower slope. VP, our preferred crater-mapping surface (see above and discussion of geologic context below), has q ~ −1.7.

Geological processes on Pluto and Charon

Pluto has had a complex geologic history (22, 24). The differing crater densities across Pluto and the range of SFDs in Fig. 2 indicate that the geology influences crater modification and retention [as it does on Mars (27)]. However, we cannot identify any geological resurfacing that has preferentially erased Pluto’s small craters (D ≲15 km) in all areas. Geologic processes that could substantially modify craters on Pluto are [1] viscous relaxation of topography, [2] tectonics, [3] mass wasting, [4] glacial erosion by surface volatile ice (N2, CO, CH4) flow, [5] mantling by atmospheric fall-out or volatile ice deposition during Pluto’s seasonal or climatic cycles, [6] sublimation erosion, [7] subsurface convective activity, or [8] icy volcanism also known as cryovolcanism (supplementary text).

Viscous relaxation [1] preferentially erases larger craters except in unusual circumstances (25, 28). Viscous relaxation, tectonics, mass wasting, and glacial activity [1 to 4] rarely completely erase craters, and there is no geomorphic evidence for these processes modifying craters on Pluto, which would produce a range of degradation states. There is evidence that processes [5 to 8] have modified craters (or at least terrains) on Pluto and could potentially erase more small craters than large ones. Of these four processes, there is the most evidence for mantling, or covering of a terrain with a new layer of material. This is most clearly observed in Pluto’s northern terrains (Lowell Regio, fig. S3A), where partially filled larger craters and general softening of topography imply that smaller craters may be completely buried (24). Some smooth terrains are present in Pluto’s dark equatorial band (Cthulhu, fig. S3B) and may indicate localized in-filling. However, the topography in Cthulhu is generally not softened to the same degree as in Lowell, and Cthulhu does not show as many smooth-floored, nearly filled craters as seen in the north. On Pluto, cryovolcanism has possibly occurred in specific regions (e.g., Wright Mons), convection in Sputnik Planitia’s nitrogen-ice-rich plains, and sublimation erosion in other regions (e.g., Piri Planitia and the “bladed” terrain); these processes appear to have removed or buried all craters in their respective regions, large and small (24, 29) (fig. S3, D to G).

On Charon’s VP, the main resurfacing process that has affected large surface areas is cryovolcanism. Charon has no detectable atmosphere (30), and there is no evidence for past glacial activity involving volatile ices (or even their presence on the surface), as there is on Pluto. Volcanic and tectonic activity has occurred since the emplacement of the main VP unit, but there is little evidence that this later activity has embayed or directly modified individual craters (Fig. 1). That is, none of the craters on the main VP area are partially filled, embayed, or breached with smooth, flat floors [in the manner seen on the lunar mare (31)]. A few partially filled craters occur north of the large tectonic scarps that separate the main area of VP from Oz Terra (Fig. 1F); this embayment may have been contemporaneous with the emplacement of the VP to the south or could represent a later episode. Some large craters on VP have hummocky floors (Fig. 1C). At the available image resolutions, these hummocks could be viscous flow features, or they may be landslide material, but they appear similar to mass movement (e.g., landslides and debris flows) and collapse features seen in crater floors on other worlds [e.g., the Moon; fig. S4 (31)]. A few smooth patches (~50 to 100 km across) with even fewer craters are found in VP itself, suggesting possible resurfacing by flows after the main resurfacing event(s) (Fig. 1A and fig. S5). However, these smooth areas are generally devoid of large craters (with only a few smaller craters throughout), and thus do not indicate that small craters were preferentially erased in these areas.

Overall, the break in logarithmic slope and deficit of smaller craters on P&C appear to be a primary characteristic of the impacting body population because (i) no process or combination of processes appear to have preferentially erased small craters across all areas of P&C, and (ii) the SFD break in slope or turnover occurs at approximately the same diameter (D ~10 to 15 km) on both P&C and across varied terrains or subsets of terrains (fig. S6). The same or similar shallow slope is seen for equivalent (scaled) crater sizes on young terrains on Jupiter’s moon Europa and on (relatively) younger impact basins on Ganymede and Callisto (2) (fig. S7).

With regard to the last point, the icy satellites of Jupiter and Saturn are thought to be cratered by objects ultimately derived from the Kuiper belt (2). Although young surfaces on Europa and Ganymede were by definition resurfaced in the past, the existing craters on these younger surfaces show few signs of geological disruption, and craters with diameters ≳2 km are thought to represent the primary impact flux at Jupiter and Saturn (2, 15, 32, 33). The shallow SFD slope found at small diameters on some of these icy satellites also implies a dearth of small comets or other KBOs (2) and is thus consistent with our results.

The surfaces of P&C do not exhibit any obvious secondary craters, which are smaller craters formed by ejected fragments from a primary impact. Any dispersed (as opposed to clustered and thus obvious) secondary craters, or craters from collisionally derived debris within the Pluto system, would tend to increase the number of small craters relative to large ones and lead to a steeper crater SFD (3436). These populations thus do not appear to play a role at the scale of our measurements, because we see no morphological or spatial patterns that would indicate their presence, and we observe shallow SFD slopes. If impactors internal to the P&C system do contribute, this would require the primary, heliocentric, small KBO impactor population SFD slope to be even shallower to match the observed crater SFDs.

Implications for Kuiper belt populations

The craters on Charon’s VP have a break in the slope (or turnover) close to 13 km (fig. S8). Similar breaks are seen in regions on Pluto (Fig. 2), but we adopt the VP turnover diameter for the remainder of the analysis because we consider the craters on VP to be the best representation of the primary impactor population (as discussed above). A crater diameter of ~13 km on Charon corresponds to an impactor diameter (d) of ~1 to 2 km (23). The crater data from VP indicate a shallow slope of q ~ −1.7 ± 0.3 for craters from ~1 to 13 km in diameter (using the mean and median value of the weighted slope fits in table S2 and the largest uncertainty estimate), corresponding to impactors from ~100 m to 1 km in diameter. We will refer to this break in slope at d ~1- to 2-km KBOs as the “elbow” as there is already a slope break known for d ~100-km KBOs (11, 3741) called the “knee.” There are too few large (>100-km diameter) craters on P&C to observe the knee slope change itself in the crater population, although the dearth of such craters is consistent with the steepening of the SFD at large d.

The elbow is located at an absolute g-band magnitude Hg ~18, and the knee at Hg ~9, assuming a geometric g-band albedo of 0.05 to convert between size and magnitude estimates (23). The crater SFD slopes (q) translate to Hg differential magnitude distribution slopes (α) via q = −(5α +1) (6). The crater SFD slopes thus equate to α of ~0.4 ± 0.04 and ~0.15 ± 0.04, respectively, for (uniform albedo) impactors larger and smaller than the elbow. The Outer Solar Systems Origins Survey has reported α = 0.4 (q ~ −3) for KBOs with d < 100 km (42). However, Earth-based observational searches have not yet probed sufficient numbers of small KBOs to test the additional elbow break at d ~1 to 2 km (or H ~18).

The craters on P&C represent impacts from a mix of different Kuiper belt dynamical subpopulations, with four contributing the majority of impactors: the stirred component of dynamically cold classical KBOs, the dynamically hot classical KBOs, other 3:2 Neptune-resonant KBOs (also known as plutinos), and the outer classical KBOs (including objects whose orbits are detached from Neptune interaction) (6, 15, 43). Figure 3 compares the SFDs from Pluto and Charon with predictions constructed from several impactor flux models for the outer solar system (2, 6, 15, 43) and one model based on a collisional evolution of the asteroid belt (44, 45). The G16 knee model (43) uses Earth-based telescopic constraints to set the SFD slope and number for Kuiper belt objects larger than ~100 km in diameter, and bends to a shallower slope (for all subpopulations) below that size (6, 11, 12). Although the absolute impactor flux levels carry substantial uncertainty (6, 43), the three different age predictions shown in Fig. 3A for the knee model suggest that the surface of Charon is ancient, close to 4 billion years (Ga) (24).

Fig. 3 Charon crater SFDs compared to impactor population models.

(A) Charon crater data (symbols, same as in Fig. 2), along with predictions for crater populations from two impactor flux models S13 (17) and G16 (6, 43) (curves), shown for three different age surfaces. The empirical crater saturation density as calculated in G16 (6, 43) is also shown for reference, where typically the density of craters does not increase because subsequent impacts erase previous craters while forming new ones. (B) The same crater data shown with three more models of impactor flux, Z03 (2), G16 (6, 43), and BD15 (15), all for a 4-Ga-old surface. The knee model from (A) is included for reference, as are two different scalings of a collisional evolution model of the asteroid belt (23, 44, 45). The crater data do not match the slopes predicted from collisional evolution models (in which collisions between KBOs would produce copious fragments) and do not resemble the asteroid belt at small sizes. The Z03 model, based on craters on the young surfaces of Europa and Ganymede (2), is the closest match to the Charon craters (D <10 km) over the entire data range.

In Fig. 3B, the Z03 model is based on crater distributions on the icy satellites of Jupiter. The Z03 4-Ga prediction (2) follows the overall shape of the Charon crater SFDs and is similar in overall crater density to the knee model between 20 km ≲ D ≲ 100 km (2). The BD15 model (15) predicts crater densities about an order of magnitude below the knee model (for 4 Ga), but the −2 slope motivated by the icy Jovian satellite craters is again more similar to that seen for P&C’s smaller craters. The P&C crater data are inconsistent with the S13 model (Fig. 3A), which is calibrated using two stellar occultations by small (d ~0.5 km) KBOs (46, 47), after conversion to equivalent Charon crater densities (43). The slope from the S13 collisional evolution model is steep and does not consistently match the P&C crater densities over any diameter range [for surface ages of 100 million years (Ma), 1 Ga, or 4 Ga; see Fig. 3A].

The KBO SFD derived from P&C is not consistent with a population in traditional collisional equilibrium where destructive collisions would produce an average SFD slope of q ~ −3.5 below some scale, with undulations or waves superimposed (23). Collisional models can match the asteroid belt SFD (48) and have been applied to explaining the Kuiper belt knee (49). It is possible the KBO population may have reached a collisional equilibrium state over a limited size range (the slopes for craters between ~10- and 100-km diameter are consistent with such collisional models), but with smaller impactors (d < 1 km) lost by some other process(es). If small objects have a low internal strength, some collision models can produce shallow slopes by destroying objects smaller than 1 km (50). Stirring by Neptune can also affect the predicted SFD structure (51). Thus, it is possible that different KBO structural properties or dynamical influences may result in collisional outcomes different from those in the asteroid belt. Nevertheless, the differential q ~ −1.7 slope over at least one order of magnitude in crater size is difficult to reproduce in models.

Alternatively, the SFD shape that we see on P&C may represent a more primordial (less collisionally evolved) KBO impactor population, possibly a remnant accretional size distribution (23), which would constrain early Solar System processes. If the shallower slope for small KBOs is primordial, it may be more consistent with dynamical models of gravitational instabilities causing rapid growth to larger objects (7, 1820). These models produce fewer small bodies, which could also result in fewer collisions overall, leading to a lower production rate of collisional fragments. However, collisions of even a few large objects would produce a collisional tail of small bodies and cause the overall SFD to evolve toward an equilibrium slope (q ~ −3–4) (e.g., 48). Models of streaming instabilities (52) followed by gravitational collapse produce size distributions with q ~ −2.8 down to D ~80 km (the smallest objects numerically resolved) (20, 21, 53). Gravitational collapse should become less efficient for smaller-mass clumped regions at some characteristic scale or scales, which could potentially explain the shallow SFD slopes (q values) that we observe (54). There may be more than one combination of processes that can produce the observed KBO SFD. If there are fewer small objects in the outer Solar System, the probability of collisions is lower, and comets such as 67P/Churyumov–Gerasimenko are less likely to be affected by many catastrophic collisions throughout their lifetimes (7, 8) (supplementary text). By providing constraints on a size range of KBOs not currently accessible by telescopes, the New Horizons crater data can help discriminate between models of accretion, evolution, and/or emplacement of the Kuiper belt.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S8

Tables S1 to S3

Data Files S1 to S3

References (55105)

References and Notes

  1. Materials and methods are available in the supplementary materials.
Acknowledgments: We thank the entire New Horizons team for their hard work leading to a successful Pluto system flyby. We thank three anonymous reviewers for helpful comments that improved this manuscript. We also thank M. Kirchhoff, K. Zahnle, B. Bottke, S. Marchi, S. Jacobson, D. Nesvorný, A. Schreiber, A. Youdin, and J. Simon for sharing their data and/or engaging in discussions. Funding: New Horizons team members gratefully acknowledge funding from the NASA New Horizons Project. B.G. acknowledges support of the Natural Sciences and Engineering Research Council of Canada. S.G. acknowledges support from NASA NEOO grant NNX14AM98G to Las Cumbres Observatory. Author contributions: K.N.S conducted the crater size-frequency analysis, coordinated the research, and co-wrote the paper. W.B.M., B.G., S.G., and E.B.B. contributed ideas and co-wrote the paper. S.A.S., A.H.P, S.J.R., P.M.S., W.M.G., V.J.B., R.A.B., R.P.B., H.A.W., L.A.Y., J.R.S., J.J.K., J.M.M, A.M.Z., C.B.O., T.R.L, C.M.L., and K.E. are New Horizons team members who contributed to the development of the paper through discussion and text suggestions, as well as contributing to the success of the New Horizons encounter with Pluto that enabled the data presented in this paper. Competing interests: There are no competing interests to declare. Data and materials availability: The data points in Figs. 2 and 3 are provided in data file S3. The data needed to derive the points in Figs. 2 and 3 are provided in data files S1 and S2. The specific New Horizons images used are listed in table S1 and archived in the Planetary Data System (PDS) Small Bodies Node at (LORRI) and (MVIC).

Stay Connected to Science

Navigate This Article