Intermittent saltation drives Mars-like sand transport on Titan

,


Introduction
The Cassini-Huygens mission has revealed that Titan's low latitude surface presents a variety of landforms [Lorenz et al., 2006, Lopes et al., 2019, MacKenzie et al., 2021], including gigantic linear dunes similar in shape to those of the Namib desert [Radebaugh et al., 2008[Radebaugh et al., , 2010]].Analyses of Cassini spectral data, combined with atmospheric and radiative transfer modeling, have further revealed that Titan presents an active dust cycle [Charnay et al., 2015, Rodriguez et al., 2018].This observational evidence suggests that, much like on Earth, Titan dunes actively evolve by an aeolian, or wind-driven, transport process known 1 as saltation: after being lifted and accelerated by the wind, surface grains hop along the granular bed, rebounding and splashing other grains into the airflow [Kok et al., 2012, Pähtz et al., 2020].Titan's sand grains are not made of silicates as on Earth but mainly of solid organics precipitated from the atmosphere [Lorenz, 2014].Even though their physical properties are not precisely known, previous studies have suggested that these organic grains could be less dense and more cohesive than quartz sand [Imanaka et al., 2012, Hörst and Tolbert, 2013, He et al., 2017, Méndez-Harper et al., 2017, Yu et al., 2017, 2020a,b].This, in combination with Titan's denser atmosphere and reduced gravity, leads to fundamental differences in dune formation on Earth and Titan.One of the main, yet poorly understood differences is that Titan's dunes appear to be shaped by surface winds opposite in direction to the prevailing atmospheric circulation [Tokano, 2008, McDonald et al., 2016, Ewing et al., 2015].A commonly accepted explanation is that the threshold wind speed required to initiate saltation, the so-called fluid threshold, lies above the speed of the prevailing easterly winds [Burr et al., 2015] but below the speed of stronger westerlies generated by equatorial methane storms at equinox [Charnay et al., 2015].This is based on the consensus that particle lifting on Titan is done primarily by aerodynamic forces and that transport cannot be sustained below the fluid threshold [Kok et al., 2012].Recent studies, however, have suggested that saltation of cohesive grains, such as those on Titan, can be sustained through rebound and granular splash at much lower wind speeds than those required to initiate grain motion [Comola et al., 2019a, Pähtz et al., 2021].The role of the fluid threshold in Titan's dust cycle and landscape evolution may therefore be less relevant than previously thought.
Here, we aim to shed light onto aeolian transport processes on Titan through a combination of laboratory experiments, theory, and numerical modeling.For this purpose, we propose novel parameterizations for the aerodynamic entrainment and granular splash processes that account for the effect of cohesive forces among surface grains.We specify the key physical parameters of these parameterizations, namely grain density, elasticity, and cohesion, based on recent experimental investigations [Yu et al., 2017] and test their performance against the results of a discrete element model.We then account for the proposed entrainment parameterizations in the saltation model COMSALT [Kok and Renno, 2009] to investigate how sediment mass flux scales with friction velocity on Titan.We finally include the mass flux parameterization in the general circulation model TAM [Lora et al., 2015[Lora et al., , 2019] ] to quantify yearly sediment transport rates and drift directions on Titan.We find that Titan's prevailing circulation drives intermittent sediment transport of the order what is found in mobile Martian dune fields.

Fluid threshold on Titan
A correct estimation of the fluid threshold is essential to understand the conditions that allow for aerodynamic entrainment of grains on Titan.To estimate the fluid threshold, we use the well-known parameterization [Shao and Lu, 2000] where A N = 0.0123 is an empirical dimensionless parameter, g ≈ 1.35 m s −2 is the gravitational constant, d is the particle diameter, ρ f ≈ 5.2 kg m −3 is the air density, ρ p ≈ 950 ± 450 kg m −3 is the particle density (uncertainty estimations throughout the paper refer to standard errors), and γ is a cohesion coefficient.
Cohesion is related to the intrinsic stickiness of the material (the surface energy), the particle shape and roughness, the stiffness of the contacting grains, and the moisture conditions [Israelachvili, 1986].It is usually assumed that γ ∝ β = F φ /d, that is, the ratio between the cohesive force F φ and the particle size d.It is important to note that β represents the average cohesive force between grains, whereas γ represents the cohesive force acting on the grains that are more readily lifted by the wind.Because of this discrepancy, the proportionality constant between γ and β is generally unknown.We therefore estimate γ for Titan grains by assuming that the ratio of γ and β is equal on Earth and Titan, that is, Measurements of fluid threshold for quartz sand suggest that γ E ≈ 0.33 ± 0.17 mN m −1 [Shao and Lu, 2000].
Furthermore, laboratory measurements of the cohesive forces for quartz sand and Titan-analog grains, known as tholins, suggest that β E ≈ 1.2 mN m −1 [Corn, 1961] and β T ≈ 27 ± 20 mN m −1 [Yu et al., 2017].Based on equation ( 2) and accounting for error propagation, we estimate that γ T ≈ 7.3 ± 6.7 mN m −1 .We use this range of γ T in equation ( 1) to estimate the variation in fluid threshold with particle size (Figure 1).The results indicate that the minimum shear velocity required to lift a grain on Titan (red curve in Figure 1) is u * ,f t ≈ 0.12 m s −1 , which is approximately three times larger than expected if cohesive forces among organic grains on Titan were equal to those among sand particles on Earth (green curve in Figure 1).Furthermore, this minimum value corresponds to a particle size d ≈ 2 mm, meaning that the particles that are easiest to lift on Titan are roughly one order of magnitude larger than sand particles that are easiest to lift on Earth (blue curve in Figure 1).Critically, we find that previous measurements carried out in a wind tunnel with environmental conditions similar to those on Titan (black circles in Figure 1) may have significantly underestimated the fluid threshold and the size of the more mobile grains on Titan due to the low cohesion of the sediments used for the experiments [Burr et al., 2015, Yu et al., 2017].
3 Impact threshold on Titan Our analyses have so far suggested that the fluid threshold on Titan may be significantly higher than previously thought due to the high cohesion of surface grains.We now investigate the effect of cohesion on the minimum wind speed required to sustain saltation through the granular splash process, the so-called impact threshold u * ,it [Pähtz and Durán, 2018].Granular splash is a complex and highly stochastic process controlled by interparticle collisions and cohesive bonds among neighboring grains.To predict the mean velocity of the splashed grains v s , we extend an expression for loose granular materials [Kok and Renno, 2009] with an additional term that accounts for the effect of cohesion (see supporting information section S1 for the analytical derivation) In equation (3), φ is the elastic energy released upon the breaking of cohesive bonds, and δ ≈ 0.3 is the fraction of elastic energy dissipated.The elastic energy φ is a function of the cohesive force F φ and the effective bond elastic modulus E, which we estimate from experiments on tholin particles [Yu et al., 2017] (see supporting information section S1 and Table S2).Further, µ ≈ 0.15 is the average fraction of impacting momentum spent on splashing surface particles [Kok and Renno, 2009].The proportionality coefficients a ≈ 0.03 and b ≈ 1.2, which scale the contributions of collisional and cohesive forces to the ejection velocity, are assigned based on literature values [Kok and Renno, 2009] and by fitting data from discrete element simulations of splash process over cohesive surfaces (see supporting information section S1).
To estimate the mean number of splashed grains N s , we adopt a splash model derived from the energy and momentum conservation equations [Comola and Lehning, 2017].This model was shown to be in good agreement with a variety of experimental results, including granular splash data of cohesive snow and ice grains.For a granular bed of uniform spherical grains, the average number of splashed grains predicted by the energy conservation equation equals where ǫ r is the fraction of impact energy retained by the rebounding grain, P r is the probability of rebound [Anderson andHaff, 1991, Andreotti, 2004], and ǫ f is the fraction of energy dissipated to the bed.Furthermore, the average number of splashed grains predicted by the horizontal momentum conservation equation equals where µ r is the fraction of momentum retained by the rebounding grain, µ f is the fraction of momentum lost to the bed, α i is the vertical impact angle, cosα s is the cosine of the vertical splash angle, and cosβ s the cosine of the horizontal splash angle.The values of all parameters in equations ( 4) and ( 5) are assigned based on experimental measurements [Willetts and Rice, 1986, 1989, Rice et al., 1995, 1996, Nalpanis et al., 1993, Ammi et al., 2009] (see Table S1 in the supporting information).Following previous approaches [Kok andRenno, 2009, Comola andLehning, 2017], we take the number of splashed grains as to represent the transition from a momentum-limited to an energy-limited splash process.We discuss the generalizations of equations ( 3)-( 5) for mixed-sized granular beds in the supporting information (section S1).
We test the predictions of equations ( 3)-( 5) against the results of a discrete element model that was previously used to investigate the role of cohesion in the granular splash process [Comola et al., 2019a] (see the supporting information section S2 for details on the model equations).Equations ( 3)-( 5) closely reproduce the variation in velocity and number of splashed grains with cohesion predicted by the discrete element simulations for different combinations of grain size and impact velocity (Figure 2).The results suggest that the splash process is weakly sensitive to cohesion when the energy released by cohesive bonds is small compared to the gravitational potential energy (φ/mgd 10 −1 ).Conversely, cohesion exerts a relevant control on the splash process for larger values of φ/mgd by increasing the mean splash velocity (Figures 2a and 2c) and decreasing the number of splashed grains (Figures 2b and 2d).The physical reason for these results is that stronger cohesive bonds, albeit more unyielding, release a larger amount of elastic energy upon breaking, thereby increasing the grain ejection velocity [Comola et al., 2019a].We find that cohesive forces have a small impact on the granular splash of organic grains on Titan.The cohesive energy φ, estimated from the experimental measurements [Yu et al., 2017], is in fact barely sufficient to affect the granular splash process of particles of size d = 0.25 mm (gray areas in Figures 2a and 2b).Cohesion is even less relevant for the granular splash of coarser grains of size d = 2.5 mm (gray areas in Figures 2c and 2d), which are primarily splashed through chains of interparticle collisions, similar to how sand grains on Earth are splashed (black markers in Figures 2a and 2b) [Crassous et al., 2007].
To investigate the effect of cohesion on the impact threshold u * ,it on Titan, we implement equations ( 3)-( 5) in the comprehensive saltation model COMSALT [Kok andRenno, 2009, Kok, 2010a,b] and simulate Titan saltation for a wide range of cohesive forces (see the supporting information sections S3 and S4 for the implementation details).Critically, we find that, even though cohesive forces greatly increase the fluid threshold on Titan, they only slightly affect the impact threshold of grains larger than 0.1 mm (blue lines in Figure 3), and only moderately increase the impact threshold of smaller grains.Most importantly, the minimum impact threshold u * ,it ≈ 0.03 m s −1 is a factor of four smaller than the minimum fluid threshold, suggesting that Titan saltation may be sustained at wind speeds much smaller than those required to initiate it.Furthermore, the minimum impact threshold corresponds to a particle size d ≈ 0.1 mm, which is one order of magnitude smaller than the size of particles most easily lifted by aerodynamic forces.

Size of saltating grains on Titan
Our results have thus far indicated that the minimum fluid threshold corresponds to a particle size d ≈ 2 mm, whereas the minimum impact threshold corresponds to a particle size d ≈ 0.1 mm.It follows that the size of grains in saltation may depend on the wind speed, that is, coarser near the transport initiation threshold and finer near to the transport cessation threshold.
To investigate the size range of saltating grains, we assume that Titan's surface presents mixed-sized grains in the range 0.05 − 2 mm, similar to sand grains on Earth.We further assume that, whenever the wind speed exceeds the minimum fluid threshold, all surface grain sizes are susceptible to motion according to the equal susceptibility principle [Martin and Kok, 2019].We follow a similar approach to previous studies [Greeley and Iversen, 1985, Nishimura and Hunt, 2000, Sullivan and Kok, 2017] and investigate the size distribution of grains in saltation by evaluating the ratio w s /u * , where w s is the terminal fall velocity as a function of the grain size (green curve in Figure 3).Values of w s /u * near unity indicate that gravitational and turbulent forces are of the same order of magnitude and grain transport is therefore transitional between saltation and suspension.We find that, near the threshold for transport initiation (u * ≈ u * ,f t ), w s /u * > 1 for d > 0.2 mm, whereas, near the threshold for transport cessation (u * ≈ u * ,it ), the ratio w s /u * > 1 for d > 0.05 mm.
These results suggest that the size of saltating grains at the onset of transport lies in the range d ≈ 0.2 − 2 mm, as smaller grains become suspended in turbulent eddies.Conversely, close to the cessation of transport, the size of saltating grains lies in the lower range d ≈ 0.05 − 0.1 mm, because the wind speed is not sufficient to sustain saltation of larger grains through rebound and splash.

Mass flux scaling on Titan
Our analyses indicate that initiation and cessation of saltation on Titan occur at very different wind speeds, yielding a ratio between the impact and fluid thresholds u * ,it /u * ,f t ≈ 0.25 much smaller than previously thought [Kok et al., 2012].This suggests that saltation on Titan can be sustained at much lower wind speeds than those required to initiate it, similarly to the transport mechanisms on Mars [Sullivan and Kok, 2017].We find that the surface wind speeds in Titan's equatorial band (30 • S -30 • N) predicted by general circulation models [Tokano, 2010, Lebonnois et al., 2012, Lora et al., 2015, Newman et al., 2016] exceed the impact threshold 15 -30% of Titan's year and can therefore sustain sediment transport (see the supporting information section S5).To quantify the sediment transport rates driven by the prevailing circulation, we derive a saltation mass flux parameterization for Titan conditions and test its accuracy against COMSALT simulations.
Previous studies have suggested that the general expression for the steady-state saltation mass flux reads , where L is the mean hop length of saltating grains and ∆v is the mean difference in grain horizontal velocity before and after impacting the bed [Durán et al., 2011, Kok et al., 2012].In steady-state saltation, the impact velocity is bound to yield a mean replacement capacity equal to 1, that is, to generate on average one splashed grain for every impactor that fails to rebound [Ungar and Haff, 1987].It follows that ∆v is independent of u * and rather scales as ∆v ∼ u * ,it .Conversely, the hop length L is determined in part by particle speeds higher up in the saltation layer.For saltation on Earth, L is only a weak function of u * and is often assumed to scale as L ∼ u 2 * ,it /g [Durán et al., 2011, Martin andKok, 2017].However, saltation on Titan is characterized by much longer hop times than on Earth due to the higher air density, thus higher air drag, and smaller gravity.It follows that particle speeds in the upper part of the saltation layer can scale with u * without producing a strong increase in impact velocity.
Assuming similar proportions in the populations of grains in saltation and in reptation near the surface [Andreotti, 2004, Lämmel et al., 2012], the mean hop length on Titan scales as L ∼ u * ,it (u * + u * ,it )/g.
The proposed scalings for L and ∆v on Titan are confirmed by COMSALT simulations (see supporting information section S6) and yield a mass flux where A ≈ 2.3 is a dimensionless scaling coefficient and η q ∈ (0, 1) is the intermittency factor that quantifies the fraction of time that saltation is active when the unsteady wind speeds oscillate between the impact and fluid thresholds.We calculate η q using the parameterization of Comola et al. [2019b], which was validated using extensive field data from three different locations on Earth.This parameterization predicts transport intermittency based on the friction velocity and the Obukhov stability parameter, which quantify the shear-generated and buoyancy-generated turbulence driving the variability in wind speed (see supporting information section S5 for details).We find that the mass fluxes predicted with equation ( 6) are in good agreement with steady-state mass fluxes obtained with COMSALT for a variety of particle sizes and friction velocities (Figure 4a).
The mass flux scaling Q ∝ u 3 * of equation ( 6) is typical of particle flows that dissipate energy through a combination of fluid drag, particle-bed collisions, and binary collisions between airborne grains [Pähtz and Durán, 2020] and is found in another mass flux parameterization by Kawamura [1951], which has been commonly used in planetary saltation studies [e.g., White, 1979, Lee and Thomas, 1995, Charnay et al., 2015, Gebhardt et al., 2020].However, in the original parameterization by Kawamura [1951] it is assumed that fluid lifting drives continuous sediment transport and that the friction velocity at the bed, for which the threshold friction velocity is a proxy in the mass flux equation, is equal to the fluid threshold.We find that our parameterization that accounts for transport intermittency (equation ( 6)) predicts a significantly larger mass flux than the continuous transport parameterization by Kawamura [1951] when the wind speed lies between the impact and fluid thresholds, as is often the case on Titan (Figure 4b).

Aeolian activity on Titan
We assess the aeolian transport potential of Titan's general circulation by implementing the proposed mass flux parameterization (equation ( 6) in the Titan Atmospheric Model (TAM) [Lora et al., 2015[Lora et al., , 2019] ] accounting for the effect of large-scale topography [Corlies et al., 2017] (see supporting information section S7 for additional detail).We perform runtime calculations of the wind-driven saltation mass flux for five Titan years, using surface friction velocities and intermittency factors computed at the model time step of 10 minutes, horizontal resolutions of approximately 5.6 degrees, and a vertical grid with 48 levels of varying pressure thickness.We assign the instantaneous mass flux direction equal to the corresponding wind direction at the first node above the surface.
The simulated yearly mass fluxes on Titan show a significant spatial variability in magnitude and direction (red arrows in Figure 5a).We find that accounting for Titan's large-scale topography leads to drift directions that diverge significantly from Titan's prevailing easterly circulation (the drift directions in absence of topography are shown in Figure S7 of the supporting information).The effect of large-scale topography might thus partly explain the inconsistency between the direction of the prevailing winds and the eastward orientation of the linear dunes, which previous studies have thus far attributed to the occurrence of eastward-propagating methane storms [Charnay et al., 2015], long climate cycles [Ewing et al., 2015], and orbital forcing [McDonald et al., 2016].The model results indicate that Titan presents regions of significant aeolian activity, with yearly mass fluxes of the order of 10 5 kg m −1 year −1 (note that one Titan year corresponds to approximately 29.5 Earth years).Furthermore, sediment transport is active approximately 30% of the year (Figure 5b), with higher saltation activity during the summer season in the northern and southern regions (blue and red lines in Figure 5b) and with little seasonality in the equatorial region (green line in Figure 5b).

Discussion
We combined experimental results, theory, and modeling to investigate the conditions that lead to sediment transport initiation and cessation on Titan.We found that the minimum fluid threshold (u * ≈ 0.12 m s −1 ) corresponds to a particle size d ≈ 2 mm, whereas the minimum impact threshold (u * ≈ 0.03 m s −1 ) corresponds to a particle size d ≈ 0.1 mm (Figure 3).Furthermore, the impact threshold is smaller than the fluid threshold for grains smaller than 2 mm, whereas the fluid threshold is smaller than the impact threshold for larger grains.The granular splash process is thus more effective than aerodynamic forces in lifting submillimeter grains from the surface.Conversely, transport of super-millimeter grains is primarily sustained by aerodynamic entrainment, which typically occurs in dense fluid flows such as fluvial environments on Earth [Pähtz et al., 2020].It is noteworthy that the fluid threshold values predicted by equation ( 1) are representative of wind tunnel conditions, where turbulence scales are much smaller than in the atmospheric boundary layer.Because the aerodynamic entrainment is predominantly caused by turbulent fluctuation events [Pähtz et al., 2020], it is possible that the fluid threshold on Titan may be up to 50% smaller than what is predicted by equation ( 1) due to the large turbulent motions in the thick boundary layer [Pähtz et al., 2018].Despite these uncertainties, the separation between the minimum fluid and impact thresholds on Titan is likely to be significantly larger than on Earth [Martin andKok, 2018, Ho et al., 2011].Much like saltation on Mars, Titan saltation may therefore be characterized by a process of hysteresis whereby the occurrence of transport below the fluid threshold depends on the history of the wind, that is, saltation occurs only if transport was initiated (u * > u * ,f t ) more recently than it was terminated (u * < u * ,it ) [e.g., Kok, 2010a].
We investigated the size of saltating grains on Titan by evaluating the ratio between settling velocity and friction velocity, w s /u * , for a wide range of grain sizes.We found that the size range of saltating grains may depend on the wind speed, varying from 0.2 − 2 mm near the transport initiation threshold to 0.05 − 0.1 mm near the transport cessation threshold.Note that our analysis based on the equal susceptibility assumption may provide incorrect estimations of the size of saltating grains if some grain sizes are more susceptible to motion than others.For instance, Sullivan and Kok [2017] have found that 0.1 mm grains are prevalent in actively-migrating ripples on Mars even though w s /u * ,f t is much larger than one for this particle size.
Our analyses further indicated that the saltation mass flux on Titan scales with the third power of the wind friction velocity, that is, Q ∝ u 3 * (equation ( 6) and Figure 4).This suggests a higher sensitivity of the transport rate to the wind speed compared to Earth conditions, where Q ∝ u 2 * [Martin and Kok, 2017].
However, the larger separation between the fluid and impact thresholds on Titan, combined with the typically low wind speeds of the prevailing circulation, is more likely to cause intermittent transport than on Earth [Comola et al., 2019b].We implemented the proposed mass flux scaling in the Titan general circulation model TAM and estimated that the regions with more intense aeolian activity present transport rates of the order of 10 5 kg m −1 per Titan year (Figure 5a).This is similar to the transport rate observed on the most mobile dune fields on Mars [see, e.g., Bridges et al., 2012], where the atmosphere is more energetic but less dense than on Titan.Our TAM simulations indicate that transport intermittency causes saltation to be active approximately 30% of the year, with significant seasonal variations (Figure 5b).Given the poor constraints on Titan's topography and the limitations involved in solving for convective processes in current Titan GCMs, large uncertainties remain in how methane storms and fine-scale topographic features affect Titan's aeolian activity.Nevertheless, our analyses indicated that Titan's prevailing winds are capable of generating a significant "background" aeolian activity and that the effect of large-scale topography on near-surface winds is critical to explaining Titan's geomorphology and landscape evolution.
thank Tetsuya Tokano, Claire Newman, Kirby Runyon, and Benjamin Charnay for sharing their Titan GCM and RCM model outputs.The authors also wish to thank Thomas Pähtz for the insightful discussions on the uncertainty in the fluid threshold value and the effect of the viscous sublayer on the impact threshold.All data presented in this paper will be made available at the following repository doi:10.17632/97j874sph6.1.Bagnold, 1937, Chepil, 1945, Zingg, 1953, Iversen et al., 1976, Fletcher, 1976].Black circles indicate fluid threshold measurements carried out in the Titan wind tunnel [Burr et al., 2015] for sediments with weaker cohesive bonds than organic grains on Titan (silica sand, basaltic sand, glass spheres, walnut shells).[Willetts and Rice, 1985, 1986, Rice et al., 1995, Willetts and Rice, 1989, Rice et al., 1996, Gordon and McKenna-Neuman, 2011].(c) Velocity and (d) number of splashed grains from a monodisperse granular bed with particle size d = 2.5 mm.The shaded gray areas indicate the estimated range of cohesion for organic grains on Titan.6) for different combinations of particle size and wind friction velocity.Because COMSALT simulates continuous transport, we assumed η q = 1 in equation ( 6).(b) Titan mass flux scaling predicted with equation ( 6) in conditions of intermittent transport (η q < 1, dashed red line) and with the mass flux equation by Kawamura [1951] commonly used in planetary aeolian transport studies (solid blue line).We assumed that grain sizes on Titan lie within the range 0.05 − 2 mm, similar to sand on Earth, and set the impact and fluid thresholds equal to the corresponding minima in this range, that is u * ,it = 0.03 m s −1 (dashed black line) and u * ,f t = 0.12 m s −1 (dotted black line).The gray area indicates the range of saltation intermittency between the impact and fluid thresholds.We computed the intermittency factor η q assuming a neutrally stable atmosphere and a Titan boundary layer height equal to 3 km [Lorenz et al., 2010] (see supporting information for details on the calculation of η q ).[Bagnold, 1937, Chepil, 1945, Zingg, 1953, Iversen et al., 1976, Fletcher, 1976].Black circles indicate fluid threshold measurements carried out in the Titan wind tunnel [Burr et al., 2015] for sediments with weaker cohesive bonds than organic grains on Titan (silica sand, basaltic sand, glass spheres, walnut shells).[Willetts and Rice, 1985, 1986, Rice et al., 1995, Willetts and Rice, 1989, Rice et al., 1996, Gordon and McKenna-Neuman, 2011].(c) Velocity and (d) number of splashed grains from a monodisperse granular bed with particle size d = 2.5 mm.The shaded gray areas indicate the estimated range of cohesion for organic grains on Titan.6) in conditions of intermittent transport (ηq < 1, dashed red line) and with the mass flux equation by Kawamura [1951] commonly used in planetary aeolian transport studies (solid blue line).We assumed that grain sizes on Titan lie within the range 0.05 − 2 mm, similar to sand on Earth, and set the impact and fluid thresholds equal to the corresponding minima in this range, that is u,it = 0.03 m s−1 (dashed black line) and u,ft = 0.12 m s−1 (dotted black line).The gray area indicates the range of saltation intermittency between the impact and fluid thresholds.We computed the intermittency factor ηq assuming a neutrally stable atmosphere and a Titan boundary layer height equal to 3 km [Lorenz et al., 2010] (see supporting information for details on the calculation of ηq).

Supplementary Files
This is a list of supplementary les associated with this preprint.Click to download.suppinfo.pdf

Figure 1 :
Figure 1: Variation in the fluid threshold with particle size on Earth and Titan.The red curve refers to organic grains on Titan, the blue one refers to quartz grains on Earth, and the green one is the hypothetical fluid threshold of quartz grains on Titan.The shaded areas indicate standard errors, obtained by propagating the uncertainties in the cohesion coefficient γ and particle density ρ p .Black diamond markers indicate wind tunnel measurements of the fluid threshold in Earth conditions[Bagnold, 1937, Chepil, 1945, Zingg, 1953, Iversen et al., 1976, Fletcher, 1976].Black circles indicate fluid threshold measurements carried out in the Titan wind tunnel[Burr et al., 2015] for sediments with weaker cohesive bonds than organic grains on Titan (silica sand, basaltic sand, glass spheres, walnut shells).

Figure 2 :
Figure 2: Variation in velocity and number of splashed grains with cohesion.Colored lines refer to analytical results of equations (3)-(5), colored markers refer to discrete element simulations, and black markers refer to experimental results.Marker and line colors indicate different values of impact velocity v i , with red color indicating v i = 1 m s −1 , green color v i = 3 m s −1 , and blue color v i = 5 m s −1 .(a) Velocity and (b) number of splashed grains from a monodisperse granular bed with particle size d = 0.25 mm.The number of splashed grains is normalized by the Froude number of the impacting grain Fr i = v i / √ gd.Black markers indicate experimental results for sand particles of similar size ( d ≈ 0.3 mm) [Willetts and Rice,  1985, 1986, Rice et al., 1995, Willetts and Rice, 1989, Rice et al., 1996, Gordon and McKenna-Neuman, 2011].(c) Velocity and (d) number of splashed grains from a monodisperse granular bed with particle size d = 2.5 mm.The shaded gray areas indicate the estimated range of cohesion for organic grains on Titan.

Figure 3 :
Figure 3: Variation in fluid threshold, impact threshold, and settling velocity of grains on Titan.The fluid threshold u * ,f t (red curve, also shown in Figure1) is estimated with equation (1).The impact threshold u * ,it is estimated with the saltation model COMSALT for three different values of the cohesion coefficient β (blue curves), which span the whole uncertainty range of cohesive forces among organic grains on Titan.The settling velocity w s (green curve) is calculated by balancing the gravitational, drag, and buoyancy forces acting on spherical grains in still air.The shaded areas indicate one standard error from the mean.

Figure 4 :
Figure 4: Saltation mass flux scaling for Titan conditions.(a) Comparison of mass fluxes predicted by COMSALT and estimated with equation (6) for different combinations of particle size and wind friction velocity.Because COMSALT simulates continuous transport, we assumed η q = 1 in equation (6).(b) Titan mass flux scaling predicted with equation (6) in conditions of intermittent transport (η q < 1, dashed red line) and with the mass flux equation byKawamura [1951] commonly used in planetary aeolian transport studies (solid blue line).We assumed that grain sizes on Titan lie within the range 0.05 − 2 mm, similar to sand on Earth, and set the impact and fluid thresholds equal to the corresponding minima in this range, that is u * ,it = 0.03 m s −1 (dashed black line) and u * ,f t = 0.12 m s −1 (dotted black line).The gray area indicates the range of saltation intermittency between the impact and fluid thresholds.We computed the intermittency factor η q assuming a neutrally stable atmosphere and a Titan boundary layer height equal to 3 km[Lorenz et al., 2010] (see supporting information for details on the calculation of η q ).

Figure 5 :
Figure 5: Sediment transport rates and intermittency on Titan.(a) Cumulated mass fluxes (yellow and red arrows) predicted by Titan general circulation model TAM using equation (6) for one Titan year (approximately 29.5 Earth years).Dashed black lines and background blue colors indicate surface elevation at the model resolution.(b) Intermittency factor annual variability in the equatorial region (green line), southern region (red line), and northern region (blue line).Higher values of η q indicate more intense saltation activity.The shaded areas indicate one standard error from the mean.

Figures
Figures