Evidence for Possible Clouds in Pluto's Present-day Atmosphere

Pluto's atmosphere has long been known to contain a variety of minor constituents (e.g., CH₄, CO) beyond the N₂ that dominates its composition (e.g., Stern et al. 2015). New Horizons has recently identified trace constituents, including C₂H₂, C₂H₄, and C₂H₆, as well as extensive haze layers (Stern et al. 2015; Gladstone et al. 2016; Young et al. 2017). Submillimeter observations by ALMA have also recently reported HCN as another atmospheric trace constituent (Lellouch et al. 2017).

Numerous planets in our solar system, including Venus, Earth, Mars, Titan, and all four of the giant planets, possess atmospheres that contain clouds, i.e., discrete atmospheric condensation structures. This said, it has long been known that Pluto's current atmosphere is not extensively cloudy at optical or infrared wavelengths. Evidence for this came primarily from the high amplitude and temporal stability of Pluto's light curve (Stern 1992). However, because no high-spatial-resolution imagery of Pluto was possible before New Horizons, it remained to be seen if clouds occurred over a small fraction of Pluto's surface area.

We, therefore, undertook cloud searches against the disk, as well as near and on Pluto's limb, using imagery obtained during the Pluto flyby. Here we report the results of these searches, as well as a new analysis of currently known species that can condense at various altitudes in Pluto's present-day atmosphere (PDA) using temperature and pressure profiles retrieved by New Horizons (Gladstone et al. 2016; Hinson et al. 2017; Young et al. 2017). As we discuss below, our searches included the entire close-encounter hemisphere of Pluto. Both the limb and the disk of Pluto itself were searched for cloud-like features. Seven bright and discrete possible cloud candidates (PCCs) were identified, all at oblique geometries near the terminators. Unfortunately, no stereo information is available to demonstrate that any of the identified candidates are in fact atmospheric features as opposed to surface features.

This paper begins in Section 2 with a discussion of the condensation conditions needed to create clouds in Pluto's PDA. Section 3 then describes our cloud searches and our findings. Section 4 discusses the common attributes of cloud candidates and points to future work to advance knowledge on this front. Section 5 summarizes our conclusions.

Atmospheric dynamics, radiation, and latent heat all play a role in the formation of clouds. The degree to which one process dominates over the others determines the nature of the clouds that are formed, e.g., fog versus cumulus versus stratus, etc.

Physical mechanisms that can trigger cloud formation derive from surface heating, flow over mountains and other topography, pressure systems, and weather fronts. Strobel et al. (1996) estimated the radiative timescale near Pluto's surface to be ∼10–15 terrestrial years. This long timescale would seem to indicate that radiation plays a small or negligible role in triggering cloud formation near Pluto's surface. Our current knowledge of winds near Pluto's surface derives primarily from general circulation models (GCMs; e.g., Toigo et al. 2015; Forget et al. 2017), which predict winds of a few m s⁻¹ near the surface induced by both flow over topography and the N₂ surface condensation/sublimation cycle. Winds of a few m s⁻¹ result in dynamical timescales of ∼10 terrestrial days. So, on Pluto today, flow over topography and the pressure/wind changes associated with the nitrogen sublimation cycle are expected to be key triggers for cloud formation.

In order to truly understand whether clouds are a limited or widespread phenomenon on Pluto, we would need longer-term temporal coverage, such as what a future orbiter mission could provide. In the current absence of a dedicated orbiter, GCMs and other models will continue to be important to make predictions.

With the knowledge we have gained from the New Horizons mission, we can examine the possibility of cloud formation in Pluto's lower atmosphere by considering whether (1) any of its atmospheric gases are abundant enough to condense and (2) cloud particle growth will proceed on a reasonable timescale.

To address the first issue, the measured abundance of each known species detected in Pluto's atmosphere can be used to calculate a condensation temperature—the temperature at which the species reaches saturation (i.e., where the partial pressure equals the saturation vapor pressure). Comparing these species condensation temperatures to Pluto's atmospheric vertical temperature profiles gives an estimate of the altitude ranges (if any) at which a given species can reach saturation. Figure 1 shows such condensation temperature curves overlaid on the New Horizons entry and exit vertical atmospheric temperature profiles (Hinson et al. 2017).

Zoom In Zoom Out Reset image size

The abundances used to construct the saturation profiles in Figure 1 were taken from Gladstone et al. (2016) and Young et al. (2017) for CH₄, C₂H₄, C₂H₆, and C₂H₂ and Lellouch et al. (2017) for CO and HCN. As Figure 1 reveals, only C₂H₆, C₂H₂, and HCN reach saturation for both temperature profiles. In contrast, CH₄ and C₂H₄ require the colder temperatures at the location of the entry profile to reach saturation. In all cases, once saturation is reached, the vapor remains in a supersaturated state down to Pluto's surface. Note that, although CO is more volatile than the various hydrocarbon species and HCN, its abundance is too low for it to condense in Pluto's atmosphere.

We now discuss the second issue noted above: cloud particle growth. Rannou & Durry (2009; hereafter RD09) examined the nucleation of N₂, CO, and CH₄ in Pluto's atmosphere using pre–New Horizons occultation temperature profiles and estimated abundances. They concluded that CH₄ reaches saturation at the lowest altitudes, followed at successively higher altitudes by CO and N₂. However, their assumed CO abundance was much higher than was later observed by ALMA (Lellouch et al. 2017). Lellouch et al. (2009; hereafter L09) also discussed the supersaturation of CH₄ using stellar occultation data. Their estimate of vapor pressures 30% above saturation (i.e., a supersaturation of 0.3) near the surface is consistent with the profiles shown in Figure 1. Neither RD09 nor L09, however, modeled the microphysics of the cloud formation process, which we consider next.

Nucleation is the initial step in the formation of clouds. We calculate nucleation rates using the classical heterogeneous nucleation theory equations of Pruppacher & Klett (1997). Heterogeneous nucleation requires the presence of particles to act as condensation nuclei. We assume that the haze observed by New Horizons supplies these condensation nuclei, which Gladstone et al. (2016) estimated at a number density of ∼0.8 cm⁻³ near the surface.

Because haze is seen in the PDA, heterogeneous nucleation is a more plausible means to initiate cloud formation than homogeneous nucleation (which implies no preexisting condensation nuclei). Both processes involve overcoming an energy barrier, but the presence of preexisting condensation nuclei reduces the amount of energy needed as a function of the "contact parameter." This contact parameter describes the relationship between the condensing species and the substrate (i.e., the condensation nuclei) and so is a function of the composition of both. The contact parameter is determined from lab measurements that determine the saturation value at which the nucleation rate reaches unity, i.e., the critical saturation for nucleation. Unfortunately, very few such measurements exist for nonterrestrial condensing species, but important exceptions for our purposes include methane, ethane, and butane nucleation onto a number of substrates, including Titan tholins (Curtis et al. 2005, 2008).

Barth (2017) described how these measurements were used to estimate a critical (i.e., cloud-forming) relative humidity for a number of hydrocarbon and nitrile species in Titan's atmosphere. We follow that same methodology here, using a critical relative humidity of 110% for methane and choosing a mid-range value of 130% for all other species.

Once a stable cloud particle is formed, it can grow in a supersaturated environment through condensation. Because condensation is directly correlated to the atmospheric temperature, growth proceeds more slowly at colder temperatures.

The cold temperatures near Pluto's surface make condensation very inefficient there, with characteristic timescales to grow a 2 μm radius particle of order ∼10⁶–10⁹ terrestrial years for all condensing species we considered except methane. This means that all nucleation and growth by condensation on any (nonmethane) cloud particles in Pluto's atmosphere must be occurring at higher altitudes (e.g., ∼8–10 km above the surface at the entry site) where temperatures are >∼60 K and the characteristic condensation timescales are short compared to a Pluto year.

We simulated cloud particle formation and evolution in Pluto's atmosphere using the Pluto module of the Community Aerosol and Radiation Model for Atmospheres (CARMA; Turco et al. 1979; Barth 2017). A Pluto atmosphere was created using the temperature–pressure (T–P) profiles from Hinson et al. (2017); these are shown as the black T–P profiles in Figure 1. CARMA's aerosol model simulates nucleation, condensation, evaporation, sedimentation, and coagulation of atmospheric particles. Particles are represented in discrete size bins in a column of atmosphere partitioned into layers.

As will be seen in Section 3, all of the suspected cloud features detected in New Horizons imagery are near Pluto's surface, so we focused our CARMA simulations on altitudes below 25 km. Such simulations were conducted for CH₄, C₂H₂, C₂H₄, C₂H₆, and HCN using atmospheric columns representing both the entry and exit sounding sites (entry only for CH₄ and C₂H₄). Most species were added to the model using the vapor pressure and latent heat references shown in Barth (2017, Table 2); however, the CH₄ vapor pressure equations and latent heat were calculated from Moses et al. (1992), and the C₂H₄ vapor pressures were calculated from Brown & Ziegler (1980) and Lara et al. (1996) for latent heat of sublimation. In all cases, the vertical grid extended from the surface to 25 km altitude with a grid spacing of 500 m. Haze particles were initialized in each altitude layer in the model using a log-normal size distribution (typical for atmospheric aerosols) consisting of 35 size bins that double in mass with each step; the number of particles peaks in size at 0.1 μm radius (see also Gladstone et al. 2016), as shown in Figure 2. The particles have a fractal shape with a monomer radius of 50 nm and a fractal dimension of two. The haze particles are meant to be representative of the aerosols observed in Pluto's atmosphere but are not themselves microphysically modeled. Variation of the parameters listed above does not alter the conclusions with regard to ice particle formation described below. The haze particles were then subjected to vertical transport (sedimentation and eddy diffusion), including a flux into the top layer to maintain a constant supply of condensation nuclei. All particles in the model can fall out the bottom layer to the surface.

Zoom In Zoom Out Reset image size

In all of our simulations, cloud formation occurred near and below the condensation altitudes shown in Figure 1. Figure 3 shows the results of our cloud formation modeling calculations and compares cloud-top heights (the knee in each profile) and species abundances for the two temperature cases. For both entry and exit, HCN nucleation occurs at the highest altitudes. The similarity in entry and exit HCN number density profiles is due to the similar slope in the two temperature profiles at these altitudes. Although C₂H₄ nucleation does appear to occur as indicated in the entry simulation, the process is inefficient, and there are too few cloud particles to contribute to any observed cloud opacity. The CH₄ and C₂H₂ nucleation produces a similar total number of particles; however, as described below, CH₄ particles grow to be the larger of the two types. Nucleation of C₂H₆ creates by far the most cloud particles and so would be the dominant constituent in any visible cloud.

Zoom In Zoom Out Reset image size

More insight into the processes controlling the relative abundances of cloud particles shown in Figure 3 can be gained from looking at the nucleation rates and condensation timescales for each species. To facilitate this comparison, temperature will be used as a proxy for altitude (the mapping between temperature and altitude above Pluto's surface is shown in Figure 3). In the case of CH₄, nucleation rates are high even for the cold temperatures near Pluto's ∼40 K surface. In contrast, C₂H₄ nucleation is inefficient (i.e., slow) for cold temperatures, and so it does not provide a significant source of cloud particles. The vapor pressure of CH₄ at 40–50 K is also a factor of ∼10¹⁰ times higher than that of both C₂H₆ and C₂H₄, resulting in a much more rapid growth timescale and allowing particles to grow to radii greater than 10 μm (with a peak in their size distribution near 4 μm). The decrease in the CH₄ density profile near the surface is due to precipitation of larger CH₄ particles, whose fall timescale becomes shorter than their formation timescale. (Fall timescales for ice particles over a distance of 2 km are larger than 1 terrestrial day for 1 μm particles, about 12 hr for 2 μm particles, 6 hr for 4 μm particles, and less than 3 hr for 10 μm particles.) At 60–70 K, the nucleation rate for C₂H₆ is greater than that of CH₄ at 40 K; thus, overall there are significantly more ethane than methane cloud particles formed (for ethane, saturation is reached on Pluto at an altitude level where both temperature profiles are ∼70 K; for CH₄, it is reached on Pluto only at 40 K).

For both C₂H₆ and C₂H₂, nucleation and further growth through condensation are only efficient at temperatures above ∼50–60 K. Because the temperature gradient is steep near the altitude where they begin forming cloud particles, these processes only occur within a narrow altitude range of a few kilometers. For the case of HCN, nucleation is only efficient in the 0–25 km altitude domain modeled because the saturation level is reached at the much higher temperature of 100 K, which only occurs at higher altitudes. Condensation growth of HCN is inefficient, with timescales of ∼10⁹ s or greater, even with its higher level of supersaturation. The size of the HCN particles is thus controlled by the size of the condensation nuclei present, i.e., HCN will form a cloud particle but will not grow further by condensation.

Another factor that can limit the number of cloud particles formed is the number of haze particles available to serve as condensation nuclei. The haze particle population was checked for each of the above cases to see if the number used (0.8 cm⁻³) limited the number of cloud particles. The only species that showed a noticeable effect on reducing the number of haze particles was C₂H₆. However, a subsequent test with an increased number of haze particles only showed modest changes in the C₂H₆ cloud particle abundance, so the major limiting factor for the number of cloud particles formed is the effect of cold temperatures on the nucleation rate.

These simulations show the altitudes where cloud particles can be expected to be found in Pluto's lower atmosphere and can be used to estimate the relative abundances of each species. Our column (i.e., 1D) microphysics model is not designed to explicitly simulate the atmospheric dynamical motions that contribute to optically thick, horizontally extended clouds. However, some conclusions can be drawn through modeling a perturbation of the system. For example, adding a 1 m s⁻¹ updraft in the bottom 5 km of the C₂H₆ entry profile increases the (geometric limit) optical depth at visible wavelengths from 10⁻⁴ to ∼0.5, and a similar test of CH₄ results in τ ∼ 0.05 (from 10⁻⁶ in the unperturbed case). Thus, small perturbations to the observed atmospheric column can promote formation of optically thick clouds.

While vertical motion is less well constrained, there is certainly topography to induce an upslope flow, so the updraft simulations described above are not unreasonable. We also note that at least two of the cloud candidates described in Section 3 appear to be in the vicinity of mountainous terrain. Finally, we note that Forget et al. (2017) concluded that CH₄ ice clouds could form at altitudes within 1 km of the surface with horizontal extents >∼100 km (the only volatiles they considered, CO and N₂, did not atmospherically condense in their GCM). Their conclusions on the potential presence of CH₄ clouds were based on supersaturation, but, combined with our CARMA simulations and calculations of nucleation and growth timescales, we see that localized cloud formation is possible in the PDA.

3.1. Selection of Cloud Search Images

3.2. Selection of Cloud Candidates

Of all the images we examined, we found that only a few contained possible clouds or cloud-like features. These PCC features were identified as having three qualitative criteria: (i) bright features relative to the surrounding terrain, (ii) cloud-like "fuzziness" or a diffuse feature appearance, and (iii) a discrete areal extent. Only the cloud features detected by these criteria by multiple authors were designated as PCCs. This yielded the seven features shown in Figure 4. We note that several of these PCCs were identified in multiple images. No other cloud-like features that meet all three of the criteria given above were identified by multiple authors.

Zoom In Zoom Out Reset image size

3.2.1. PCC Properties, Locations, and Local Times

Each of our seven PCCs was detected in high-phase imagery and seen near the limb of Pluto. Most appeared as bright or elongated features (2, 3, 5, 6, and 7), but two of the candidates (1 and 4) appeared instead as clusters of bright spots. Horizontal extents of the PCCs—i.e., their dimension parallel to Pluto's surface—varied, with candidate 1 being less than 10 km long, while the others ranged from 30 to 70 km, as shown in Table 2. We found that the linear PCCs tended to have similar narrow aspect ratios and were generally located in or around ridges or other topography. Because no stereo imagery is available for any of the PCCs, we could not estimate PCC heights or prove that they are located above the surface; however, the fact that all of our PCCs appear to be low-lying is in accord with the predictions of the cloud condensation models discussed in Section 2, which suggest formation at altitudes below ∼10 km.

Table 2.  Cloud Candidate Properties, Sizes, and Brightness

PCC	Image ID	General Description	Long Dimension (km)	I/F
1	PELR_P_MULTI_DEP_LONG_1	Bright spots	<10	∼0.2
2	PELR_P_MULTI_DEP_LONG_1	Linear	∼55	∼0.3
3	PEMV_01_P_CharonLight	Linear	∼70	∼0.1
4	PEMV_01_P_CharonLight	Bright spots	∼30	∼0.1
5	PELR_P_LORRI_ALICE_DEP_3	Linear	∼30	∼0.1
6	PELR_P_LORRI_ALICE_DEP_3	Linear	∼65	∼0.1
7	PELR_PC_MULTI_DEP_LONG_2_01	Linear	∼55	∼0.1

Download table as:  ASCIITypeset image

Estimated PCC brightness (measured as I/F, where I is the scattered intensity and F is the incident solar flux) is also provided in Table 2. Pluto's global haze layers have similar brightness, i.e., I/F ∼ 0.2–0.3 in LORRI images made at similar phase angles of 165°–169° (Gladstone et al. 2016). Gladstone et al. (2016) used the blue color and high forward-scattering behavior of the hazes to estimate their line-of-sight optical depth to be τ ∼ 0.16 and their number density to be n ∼ 0.8 particles cm⁻³. The fact that we detected the PCCs only in high-phase-angle images is consistent with their being preferentially forward scattering. However, because the PCCs were only observed in forward scattering, we cannot independently verify this. If the PCCs have a composition and particle size distribution similar to that of the hazes, then their similar I/Fs may indicate a similar optical depth and number density.

We also note that all seven PCCs were seen close to the day–night terminator. The two PCCs identified in MVIC images (3 and 4) were at locations within the encounter hemisphere, where high-resolution coverage is available for comparison at lower phase angles during approach. However, all five LORRI PCCs (1, 2, 5, 6, and 7) were located on the nonencounter hemisphere, which has much poorer low-phase spatial coverage. For all of the PCCs, their local time of day can be estimated based on their position on Pluto and the time the imagery was captured. The latitudes, longitudes, and local times are provided for each PCC in Table 3. Figure 5 shows their locations on a base map of Pluto, as well as the day–night terminator position near the time the images were acquired.

Zoom In Zoom Out Reset image size

Table 3.  Cloud Candidate Locations and Local Times on Pluto

PCC	Image ID	UTC (DOY H:M:S)	Lat (deg)	E Long (deg)	Local Time (H:M)
1	PELR_P_MULTI_DEP_LONG_1	195 15:41:50	+17.7	009.5	4:44 am
2	PELR_P_MULTI_DEP_LONG_1	195 15:41:50	+27.6	359.5	4:03 am
3	PEMV_P_CharonLight	195 12:11:24	+2.5	205.0	5:12 pm
4	PEMV_P_CharonLight	195 12:11:24	−19.0	174.0	3:08 pm
5	PELR_P_LORRI_ALICE_DEP_3	196 00:03:20	+05.6	005.8	5:47 am
6	PELR_P_LORRI_ALICE_DEP_3	196 00:03:20	−09.8	021.2	6:49 am
7	PELR_PC_MULTI_DEP_LONG_2_01	195 19:04:45	−07.8	033.5	6:51 am

Download table as:  ASCIITypeset image

The seven PCCs spanned latitudes from 27.6 N to 19.0 S, with no clear preference for either the summer or winter hemisphere. We note that, due to the high tilt of Pluto's rotational axis, the day–night terminator is located at only low- to mid-latitudes, and therefore we cannot rule out similar clouds at higher latitudes. However, cloud formation near the terminator would be consistent with the changing solar flux there that might promote cloud growth, as shown in our Section 2 results.

3.3. Comparison with Low-Phase Images

PCCs 3 and 4 provide a unique opportunity for further analysis. This is because their locations were also imaged in lower-phase, higher-resolution LORRI and MVIC images. These images allowed for (i) examination of the PCC region in different lighting conditions and spatial resolutions and (ii) context to assess the effects of topography and surface albedo variation in the area.

Figure 6 shows the result of georeferencing for PCC 3, including both the original MVIC image and three panels of georeferenced information. While the cloud candidate is not seen in the low-phase imagery (Figure 6, bottom right), there does appear to be an association with topography suggesting either a possible surface explanation for the observed high-phase brightening in that area (instead of a cloud) or a correlation between the formation of PCC 3 and the local topography. Figure 7 shows the result of georeferencing for PCC 4, including the original MVIC image and three panels of georeferenced information. This candidate is found in a region with scattered patches of bright surface albedo, and thus this PCC could be either a cloud or a surface feature; we cannot confirm/refute either possibility.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

As Figures 6 and 7 reveal, georeferencing of PCCs 3 and 4 suggests a correlation of each feature with a topographic depression that has a locally bright and smooth floor. Our analysis was inconclusive, however, as to whether these PCCs were simply caused by variability of the local surface scattering at high phase angles or  due to scattered light from low-lying clouds.

The seven identified PCCs share several common attributes, which we point out and briefly discuss here. These include:

• 1.  
  Small size. Both direct image analysis and georeferencing techniques confirm that all the PCCs are relatively small in size (tens of km maximum dimension).
• 2.  
  Low altitude. All of our PCCs appear to lie near (within a few km) of the surface. While their altitudes above the surface were not well constrained due to a lack of stereo imaging coverage at their locations, they additionally all appear to be low-lying relative to local topography, which would also be consistent with the predictions of the cloud condensation models discussed in Section 2.
• 3.  
  Early/late local time of day. All of the PCCs lie near local dawn or dusk. This is consistent with formation in cooler conditions where condensation is preferred but could also be an observational selection effect resulting from the more favorable observing geometry near the planet's limb. A future orbiter mission should be able to distinguish between these possibilities.
• 4.  
  Oblique detection geometry. Finally, we note that all of the PCCs we identified were detected obliquely. Such detections may be the result of a selection effect in which enhanced path length may facilitate detection, indicating that additional PCCs may have escaped detection at more normal observing geometries.

We have used the New Horizons image data set to search for discrete cloud features on Pluto. Only seven PCCs have been identified; all are small features. None can be absolutely confirmed to be a cloud, but all share attributes that are consistent with possible clouds.

Our more formal conclusions are as follows:

• 1.  
  At the current epoch, Pluto's atmosphere is almost entirely or entirely free of clouds.
• 2.  
  The seven PCCs we identified cannot be confirmed as discrete clouds because none are in regions where (i) stereo imaging, (ii) surface-lying shadows, or (iii) limb-viewing geometry can be used to definitively demonstrate that they are aboveground features.
• 3.  
  Nonetheless, all seven PCCs share a range of common features (see Section 4) that circumstantially bolster the case that some or all may be low-lying clouds.
• 4.  
  No clouds were detected as discrete features above the limb, at backscattering (low-phase-angle) geometries, or away from Pluto's terminator. Because hazes were detected to altitudes as high as 220 km, we believe clouds could also have been detected to such altitudes.
• 5.  
  The cloud condensation modeling that we performed using vertical 1D models indicates that our PCCs cannot be due to N₂ or CO condensation but could be the result of condensation of any of several atmospheric trace species, including HCN, CH₄, or any of several known light hydrocarbons.

Finally, we speculate that Pluto's seasonal and "super-seasonal" (Binzel et al. 2017; Earle et al. 2017; Stern et al. 2017) variations in cloud abundances, areal extents, and other attributes may be significant. Further modeling in this area is recommended. Specifically, a Pluto orbiter would be extremely useful to search for clouds more thoroughly in time and space than was possible during the brief reconnaissance flyby by New Horizons.

We thank NASA for financial support of the New Horizons project and the entire New Horizons mission team for making the results of the flyby possible. We thank Ms. Cindy Conrad for her editorial assistance with this manuscript. We also thank an anonymous referee for many helpful comments.