Cassini/VIMS observes rough surfaces on Titan’s Punga Mare in specular reflection
- 3.5k Downloads
Cassini/VIMS high-phase specular observations of Titan’s north pole during the T85 flyby show evidence for isolated patches of rough liquid surface within the boundaries of the sea Punga Mare. The roughness shows typical slopes of 6°±1°. These rough areas could be either wet mudflats or a wavy sea. Because of their large areal extent, patchy geographic distribution, and uniform appearance at low phase, we prefer a waves interpretation. Applying theoretical wave calculations based on Titan conditions our slope determination allows us to infer winds of 0.76±0.09 m/s and significant wave heights ofcm at the time and locations of the observation. If correct, these would represent the first waves seen on Titan’s seas, and also the first extraterrestrial sea-surface waves in general.
Saturn’s moon Titan posesses the only known open surface liquids beyond Earth . Those liquids take the form of lakes made primarily of methane, ethane, and dissolved nitrogen . The bulk of the volume of Titan’s liquids occurs near the north pole ,, though isolated lakes have also been observed near the south pole at Ontario Lacus –, possibly near the equator , and southern mid-latitude (Sionascaig Lacus) (Vixie G, Barnes JW, Jackson B, Wilson P: Two temperate lakes on Titan. Icarus, submitted).
The existence of extraterrestrial lacustrine environments allows for the possibility of waves. In theory, expanses of liquid acted upon by sufficient winds ought to show the formation of waves on Titan as they do on Earth. The wind velocity needed to generate such waves, along with the resultant wave frequencies, will necessarily be affected by Titan’s alien gravity, atmospheric density, and liquid viscosity/surface tension (which are in turn a function of composition and temperature). Theoretical calculations – predict that the first waves to be incited on Titan when the winds break the threshold should occur with wavelengths between 2.8 cm and 3.2 cm. Hayes et al.  show that these waves should be capillary-gravity waves — ones for which surface tension and gravity both contribute to the restoring force. Initial laboratory experiments with kerosene in a wind tunnel  showed that waves on hydrocarbons are both larger than waves on water and form at lower wind speeds, at least under Earth gravity and atmospheric conditions.
Despite concerted efforts, however, Cassini has thus far not detected any waves. Brown et al.  showed that near-infrared spectral determinations of the reflectivity of Ontario Lacus were consistent with a “smooth” surface. Wye et al.  used direct reflection of Cassini’s RADAR off Ontario Lacus to constrain the surface roughness to be less than 3 mm(!). Barnes et al.  used a time-resolved specular reflection across north polar Jingpo Lacus to constrain wave angles at that time to be less than 0.15°. All of these observations are consistent with Titan lakes that are as flat as a millpond at the time of the observations: entirely wave-free .
These nondetections notwithstanding, there is indirect evidence to support the hypothesis that waves do form on Titan’s lakes and seas. Wall et al.  claimed geomorphological evidence for waves, suggesting that the eastern shore of Ontario Lacus represents a beach formed by wave-deposited sediments. Lorenz et al.  explored possible explanations for why no waves have been seen, suggesting a seasonal effect, i.e. that the winds above Titan’s lakes and seas were too low at the time of the observations to initiate wave formation. Calculations suggest that the threshold wind speed for wave formation under Titan conditions might be between 0.4 m/s and 0.8 m/s ,. General Circulation Models (GCMs) generally predict that the winds near Titan’s north pole should have been rather quiescent until late northern spring, consistent with the low-wind explanation for Titan’s lack of waves thus far ,.
In this paper, we report evidence for waves on Titan’s northern sea, Punga Mare. In the ‘Observation’ section we describe the Cassini Visual and Infrared Mapping Spectrometer (VIMS) data that we interpret to show the waves in specular reflection. We describe our model for simulating the appearance of roughness-driven specular reflections away from the specular point in the ‘Model’ section. In the ‘Analysis’ section we apply that model to the VIMS observations to derive wave properties. Then, in the ‘Discussion’ section, we consider the implications of the discovered waves, before concluding.
The bright lakes and seas do not all show the same measured I/F. There are three reasons for this. The first is that at 5 μm Titan’s atmosphere is optically thin. So if you were standing on the surface in a boat on one of these lakes you would see that the sky was brighter near the horizon than at the zenith. Because lower emission angles specularly reflect a portion of the sky nearer the zenith, those areas necessarily show a dimmer specular sky reflection. The second reason is that the efficiency of the specular reflection decreases as the emission angle decreases . Finally, the T85 view encompasses the terminator, meaning that past 90° incidence angle there is no direct flux at all in the lower atmosphere.
However all three of these effects are continuous and would not lead to any liquid-filled areas differing significantly in brightness from neighboring areas in a discontinuous manner. In particular, bright specular sky reflections seen in Ligeia Mare and Kraken Mare are continuous, without spurious brighter or darker areas within the seas. Kraken does have a darker area corresponding to the island Mayda Insula. Smaller lakes Sparrow Lacus, Waikare Lacus, and Muggel Lacus each are detectably brighter than their surroundings. Neagh Lacus is beyond the terminator.
In particular, these spots’ spectra are not consistent with an isolated patch of fog or other atmospheric aerosols above the lake — such a patch would be expected to show significant signal at wavelengths shorter than 5 μm as well –. While lakes and seas can reflect the image of background clouds, VIMS imaging in the wings of the 2 micron window show no evidence of cloud activity in the area at the time of the T85 observation. Hence while we cannot be certain that the spurious 5-micron flux derives from surface specular reflection, all of our tests are consistent with a surface specular phenomenon.
How can this flux be specular in origin, though, if the pixels themselves are not at the specular point on Titan’s surface? Calculation of the surface specular point assumes that the surface conforms to a local equipotential surface. If the surface is instead tilted, or if portions or facets of the surface are so tilted, then either it or some of its facets can achieve a specular geometry away from the nominal specular point.
We thus suggest that the bright, specular pixels within Punga Mare in Figure 2 could represent a rough, wet surface at those particular locations.
We develop a numerical model of planetary specular reflections to evaluate whether the brightened areas in Figure 2 could plausibly result from specular reflection from a rough, wet surface. We use the SPICE package  combined with a downhill simplex numerical minimization algorithm  to calculate the precise orientation for a specular facet at any given point on Titan’s surface given a specific observation geometry. For each latitude/longitude point, we vary both the angular deviation from zenith (θ) and azimuthal orientation (φ) for which the angular distance between the specular vector and the Sun direction vector is zero.
To then calculate the expected brightness in I/F at each point under the roughened liquid specular scenario, we calculate the fraction of randomly oriented facets that would achieve specularity with some portion of the extended solar disk. This calculation requires an assumption for the distribution of the orientation of facets within the pixel. For this work (as in ) we assume a two dimensional Gaussian distribution with varying widths σ. Although this assumption about the distribution may not provide the highest fidelity, given our modest 4 pixels we elect to leave a more realistic distribution including wind directionality to future work as data warrant.
and Γ(θ,φ) is a function equal to 1.0 when the facet at (θ,φ) corresponds to a specular direction inside the solar disk, and equal to 0.0 outside it.
For purposes of calculating the fraction of specular facets numerically, we break the solar disk into a 30-sided polygon with the vertices around the Sun’s limb. We then calculate the area of that 30-sided polygon in θ- φ space and multiply the area by the Gaussian distribution value corresponding to the center of the Sun. This technique is much faster than an explicit two-dimensional numerical integral. Furthermore, given that the angular diameter of the Sun as seen from Titan (0.05°) is much smaller than the typical width of the facet deviation distribution that we explore (σ=1−10°) the approximation is highly accurate as well.
Finally, we normalize the total integral of the Gaussian facet deviation distribution to an assumed overall I/F value. Doing so obviates the necessity of making assumptions regarding the liquid’s index of refraction and therefore of its precise composition. The final pixel value then is equal to the total solar specular flux parameter times the fraction of specular facets — this ensures that in the case where σ=0°, the specular I/F of the pixel at the specular point would be I/Fmax. Note that because the original and model pixel I/F values are normalized, and are not true measured fluxes, summing I/F over the affected pixels does not yield I/Fmax.
To fit these data, we drove the specular brightness model with a Levenberg-Marquardt χ2 minimization fit  to fit for both the maximum I/F and the facet distribution width σ. The best-fit values that result from that fit are I/Fmax=106±16 and σ=6°±1°. The best-fit line (based on an assumed approximately radial profile path away from the specular point) is shown in black in Figure 11.
It is not clear why the best-fit value for I/Fmax is discrepant from the Barnes et al.  value (I/F=32.4) by a factor of 3. The main specular pixel is saturated, so the error could be a result of the Barnes et al. 2013 reconstruction of the saturated values. It could also be that the true surface facet distribution does not match the 2-D Gaussian that we have assumed or that the liquid surface of Kivu Lacus where the prime specular reflection was measured is partially covered with some kind of opaque film (pond scum). Differences in the real index of refraction caused by composition differences between Kivu Lacus and Punga Mare cannot account for the offset, as at this phase angle, the first-surface reflectance of pure methane and that of pure ethane differ by only ∼20% . We note, however, that this new value of I/Fmax is closer to the theoretical value as derived using the Soderblom et al.  relations of ∼70 for methane and ∼90 for ethane than that from the Kivu Lacus observation .
Our fit shows that the flux coming from the three isolated areas of Punga Mare studied is consistent with a rough liquid surface having characteristic slopes of 6°. Such surfaces could come about as a result of wetted mudflats, for instance, similar to the wet sidewalk in Figure 7B. A bright, dry playa surface could generate a roughened specular signal as well (see  Figure ten). Finally, the rough patches might also result from a purely liquid sea surface with wave activity.
The unusually high brightness of all of the areas is sufficiently high so as to only be explicable with a liquid surface: dry, or even merely moist ground will not do. Dry surfaces would also not create the bright-lake effect of reflected sky brightness when seen at high emission angle as shown in Figure 5. Furthermore, the low albedo of Punga Mare could not produce a dry specular reflection of the type seen at Etosha Pan. The brightness constraint then leaves two options for the rough surfaces in Punga Mare: wet mudflats or waves.
If the liquid overlies a solid surface, then it must conform to that surface as a thin layer of liquid. This would be like the bottom part of the specular reflection shown on the sidewalk in Figure 7B. At the central specular point in Figure 7B the water has already drained downhill somewhat (toward the bottom-left in this image), leaving that part of the concrete wet but not presently covered in a layer of water. If the expanses of Punga Mare indicated by the arrows in Figure 2 represent mudflats, they must be almost entirely presently covered in sea liquid (probably a methane/ethane/nitrogen solution ,) because the specular reflection is so bright.
However, in order for that liquid surface to have the measured roughness characteristics the liquid must drape over a solid surface. Such a liquid covering over solid can occur, as evidenced by the liquid-covered sidewalk in Figure 7B. Mudflats on Earth can have very low slopes, but even then it would be difficult to achieve an appropriately thick layer of liquid over an expanse tens of kilometers long. In addition, we see the rough liquid only at discrete locations within Punga Mare. Mudflats might be expected to occur preferentially at the sea’s margin, as at Ontario Lacus  (though in detail their distribution would depend on the sea’s bathymetry). The best imaging of Punga Mare to date occurred on T94, as shown at the top left of Figure 4. Unlike the T38 Ontario Lacus data , the T94 data do not have the spatial resolution or signal-to-noise ratio needed to discern mudflats. Wetted floating ice  would be a similar solution, but would need to be similarly liquid-covered and extensive.
The other possibility is that the bright specular patches represent liquid expanses roughened by wave activity. Wind-induced waves should be possible on Titan –,, even though searches until now had shown lakes and seas to be perfectly flat ,. The 6° typical slopes within the specular patches is not too dissimilar from typical slopes on Earth’s oceans (4° ), particularly given that Hayes et al.  predict that Titan waves should be 7 times higher and 2 times steeper than Earth waves produced with the same wind speeds. Indeed, the angular width of sea-surface specular reflections has been used on Earth as a proxy for windspeed for 60 years .
Although not fully replicated by empirical studies, an experiment at 1 bar and Earth’s surface gravity produced 12 mm amplitude waves with winds of 5 m/s . While the wind speed associated with our 6° waves would vary with composition and viscosity, the resulting wavefield is nearly independent of composition. Hence viscosity-related systematic errors afflict the wind speed determination, the significant wave heights are mostly free from compositional systematic errors. Our calculation is in broad agreement with Lorenz et al.  who did a similar calculation for RADAR detectivity of waves.
Dissipation occurs more rapidly for short-wavelength capillary waves (waves with surface tension as the restoring force); it is the longer wavelength gravity waves (waves with gravity as their restoring force) that more easily propagate over long distances. Because the first waves that would result from winds just above the threshold velocity would be at the short-wavelength end of where gravity waves are possible, if those waves were incited in patches they would only propagate outside those patches with very low amplitudes.
Thus we posit that our observation may represent waves on Punga Mare that were incited by patchy winds at or just exceeding the wave-generation threshold. If correct, these would represent the first waves on open liquid detected on a body other than the Earth.
Although Global Circulation Models (GCMs) have been used to study the prospects for waves at Ligeia Mare in detail , no specific study has done the same for Punga Mare. In general, however, predictions show that in Titan’s arctic the winds should begin to pick up as northern summer approaches . We leave an investigation of whether, and under what conditions, GCMs can replicate this scenario on Punga Mare at the time in Titan’s season of T85 to future work.
If real, the VIMS T85 Punga waves solve the prior paradox of waves’ absence in Titan lakes and seas . Indeed Titan’s maria are liquid and do not have the viscosity of molasses (good for potential lake lander missions ). Instead, as suggested by Lorenz et al. , wind conditions may not have been favorable for the production of waves until recently.
We note that the patchiness of the putative waves that we see means that it would be possible for observations at a single point or even along a single chord (like  and ) to miss them. Previously unexplained variations in specular brightness on Kraken Mare on T59  could be due to differing degrees of roughness across the face of the sea.
Had Cassini RADAR been observing Punga Mare on T85, could it have seen the putative waves that we describe here? Using the usual Synthetic Aperture RADAR (SAR) mode with high incidence angles (∼30°) it would not detect these waves — in that Bragg regime their signal might be -30 or -40 db , well below the single-pixel noise floor of ∼−20 db. At lower incidence angles, however, around 10° or less, the quasi-specular signal should make 6° waves evident even in SAR imaging. Cassini’s RADAR has only observed Titan at such low incidence angles once, and in that one observation possible wave activity was observed in one location on Ligeia Mare . Processing Cassini’s RADAR data in real aperture mode can beat down the noise and potentially reveal wave signals, but such signals would be convolved with the return from the sea floor .
VIMS T85 observations of Titan’s north pole show specular flux coming from areas within Punga Mare away from the specular point in Kivu Lacus. We develop a numerical model to simulate the appearance of a broad specular reflection off a rough surface on a spherical planet (previous work by  assumed very small roughness dispersion and is therefore not appropriate for moderately rough surfaces observed globally). The spectra, locations, and intensities are consistent with a surface covered in liquid and rough at wavelengths much longer than 5 μm with a typical angle of 6°±1°. The inferred surface wind speeds of ∼0.7 m/s are consistent with GCM predictions of increasing wind activity as northern summer approaches (e.g., ).
The rough patches could represent either wet mudflats or the development of waves on the sea surface. Because such mudflats would need a thin layer of liquid draped over rough mud to be consistent over areas tens of kilometers across, we prefer the waves interpretation. Future observations could definitively differentiate between the two ideas: if the regions are consistently rough as a function of time, then they are likely mudflats, whereas if the rough areas are different on a future flyby then that observation would be more consistent with waves.
The patchy nature of the putatively wavy seas implies locally variable winds near the threshold for wave generation. While future specular observations of the quality seen on T85 will be rare, there will be a few. In particular, a specular observation is planned for T101 on 2014 May 17. Combining the T85 observation with those future measurements should allow us to piece together a time-resolved picture of the frequency and intensity of high-wind events across Titan’s north polar seas. The Cassini RADAR instrument also has prospects for detection of wave activity in the future through low-incidence synthetic aperture radar observations, bistatic experiments (T101, T102, and T106) and an altimetry pass over Kraken Mare (T104). Observations from Cassini or other imaging missions such as JET (Journey to Enceladus and Titan)  or an airplane  or balloon could monitor wave activity in the future by planning observations of the specular point at high phase at close enough range for the roughness effect to be seen.
The authors acknowledge support from the NASA/ESA Cassini Project. JWB acknowledges support from NASA Cassini Data Analysis and Participating Scientists (CDAPS) grant NNX12AC28G. AGH acknowledges CDAPS grant NNX13AG03G. JMS acknowledges CDAPS grant NNX12AC25G.
- 1.Stofan ER, Elachi C, Lunine JI, Lorenz RD, Stiles B, Mitchell KL, Ostro S, Soderblom L, Wood C, Zebker H, Wall S, Janssen M, Kirk R, Lopes R, Paganelli F, Radebaugh J, Wye L, Anderson Y, Allison M, Boehmer R, Callahan P, Encrenaz P, Flamini E, Francescetti G, Gim Y, Hamilton G, Hensley S, Johnson WTK, Kelleher K, Muhleman D, et al.: The lakes of Titan. Nature 2007, 445:61–64.ADSCrossRefGoogle Scholar
- 4.Sotin C, Lawrence KJ, Reinhardt B, Barnes JW, Brown RH, Hayes AG, Le Mouélic S, Rodriguez S, Soderblom JM, Soderblom LA, Baines KH, Buratti BJ, Clark RN, Jaumann R, Nicholson PD, Stephan K: Observations of Titan’s northern lakes at 5 μm: implications for the organic cycle and geology. Icarus 2012, 221:768–786.ADSCrossRefGoogle Scholar
- 7.Cornet T, Bourgeois O, Le Mouélic S, Rodriguez S, Lopez Gonzalez T, Sotin C, Tobie G, Fleurant C, Barnes JW, Brown RH, Baines KH, Buratti BJ, Clark RN, Nicholson PD: Geomorphological significance of Ontario Lacus on Titan: integrated interpretation of Cassini VIMS, ISS and RADAR data and comparison with the Etosha Pan (Namibia). Icarus 2012, 218:788–806.ADSCrossRefGoogle Scholar
- 8.Cornet T, Bourgeois O, Le Moué S, Rodriguez S, Sotin C, Barnes JW, Brown RH, Baines KH, Buratti BJ, Clark RN, Nicholson PD: Edge detection applied to Cassini images reveals nomeasurable displacement of Ontario Lacus’margin between 2005 and 2010. J Geophys Res (Planets) 2012, 117:E07005.ADSCrossRefGoogle Scholar
- 12.Hayes AG, Lorenz RD, Donelan MA, Manga M, Lunine JI, Schneider T, Lamb MP, Mitchell JM, Fischer WW, Graves SD, Tolman HL, Aharonson O, Encrenaz PJ, Ventura B, Casarano D, Notarnicola C: Wind driven capillary-gravity waves on Titan’s lakes: hard to detect or non-existent? Icarus 2013, 225:403–412.ADSCrossRefGoogle Scholar
- 15.Barnes JW, Soderblom JM, Brown RH, Soderblom LA, Stefan K, Jaumann R, Le Moué S, Rodriguez S, Sotin C, Buratti BJ, Baines KH, Clark RN, Nicholson PD: Wave constraints for Titan’s Jingpo Lacus and Kraken Mare from VIMS specular reflection lightcurves. Icarus 2011, 211:722–731. [http://www.sciencedirect.com/science/article/B6WGF-5161PFV-4/2/7a95516e9191ed9db6df21c7bf5b97b4]ADSCrossRefGoogle Scholar
- 16.Wall S, Hayes A, Bristow C, Lorenz R, Stofan E, Lunine J, Le Gall A, Janssen M, Lopes R, Wye L, Soderblom L, Paillou P, Aharonson O, Zebker H, Farr T, Mitri G, Kirk R, Mitchell K, Notarnicola C, Casarano D, Ventura B: Active shoreline of Ontario Lacus, Titan: a morphological study of the lake and its surroundings. Geophys Res Lett 2010, 37:L5202.Google Scholar
- 19.Brown RH, Baines KH, Bellucci G, Bibring JP, Buratti BJ, Capaccioni F, Cerroni P, Clark RN, Coradini A, Cruikshank DP, Drossart P, Formisano V, Jaumann R, Langevin Y, Matson DL, McCord TB, Mennella V, Miller E, Nelson RM, Nicholson PD, Sicardy B, Sotin C: The Cassini Visual And InfraredMapping Spectrometer (Vims) investigation. Space Sci Rev 2004, 115:111–168.ADSCrossRefGoogle Scholar
- 27.Press WH, Teukolsky SA, Vetterling WT, Flannery BP: Numerical Recipes: The Art of Scientific Computing. Cambridge: University Press; 2007.Google Scholar
- 29.Hofgartner JD, Lunine JI: Does ice float in Titan’s lakes and seas? Icarus 2013. (0). [http://www.sciencedirect.com/science/article/pii/S0019103512004824]Google Scholar
- 30.Walsh EJ, Hagan DE, Rogers DP, Weller RA, Fairall CW, Friehe CA, Burns SP, Khelif D, Vandemark DC, Swift RN, Scott JF: Observations of sea surface mean square slope under light wind during the Tropical Ocean-Global Atmosphere Coupled Ocean-Atmosphere Response Experiment. J Geophys Res C Oceans 1998, 103(C6):12603–12612. [http://dx.doi.org/10.1029/98JC00780]ADSCrossRefGoogle Scholar
- 34.Hofgartner JD, Hayes AG, Lunine JI, Zebker H, Stiles BW, Sotin C, Barnes JW, Brown RH, Encrenaz P, Kirk RD, Le Gall A, Lopes RM, Lorenz RD, Malaska M, Mitchell KL, Paillou P, Radebauch J, Turtle E, Wall S, Wood C, The Cassini RADAR Team: The discovery of transient features in a Titan sea. Nat Geosci 2014, 7:493–496. doi: 10.1038/ngeo2190.ADSCrossRefGoogle Scholar
- 37.Sotin C, Altwegg K, Brown RH, Hand K, Soderblom J, JET Team: JET: a Journey to Enceladus and Titan. In AAS/Division for Planetary Sciences Meeting Abstracts #42, Volume 42 of AAS/Division for Planetary Sciences Meeting Abstracts; 2010:#49.31. http://adsabs.harvard.edu/abs/2010DPS....42.4931S.Google Scholar
- 38.Barnes JW, Lemke L, Foch R, McKay CP, Beyer RA, Radebaugh J, Atkinson DH, Lorenz RD, Le Mouélic S, Rodriguez S, Gundlach J, Giannini F, Bain S, Flasar FM, Hurford T, Anderson CM, Merrison J, Ádámkovics M, Kattenhorn SA, Mitchell J, Burr DM, Colaprete A, Schaller E, Friedson AJ, Edgett KS, Coradini A, Adriani A, Sayanagi KM, Malaska MJ, Morabito D, et al.: AVIATR—Aerial Vehicle for In-situ and Airborne Titan Reconnaissance. A Titan airplanemission concept. Exp Astron 2012, 33:55–127.ADSCrossRefGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited.