Largest wind ripples on Earth?
Abstract
Unique wind ripples attaining heights to 2.3 m, wavelengths to 43 m, and a crest maximum grain size of 19 mm occur on the Argentine Puna Plateau at ~4000 m altitude. These are the largest ripples reported on Earth, comparable only to Mars counterparts. They form in the presence of high proportions of low-density pumice clasts (0.91 g/cm3), although crests are exclusively composed of varnished, normal-density clasts (2.43 g/cm3). Mature ripple profiles are partly excavated on bedrock, so they form by a combination of deflation, winnowing of finer grains, minor wind drift of fine gravel, and lagging of clasts >4 cm. The large ripple size appears to be related to strong winds, dense saltation layers, and a long time for evolution. Ripple sizes are smaller on obstacles, as compared to flat terrain; there is a lack of correlation between clast size, wavelength, and the extreme ripple size (in spite of the thin atmosphere), all of which suggest that while small-scale gravel ripples may form according to a reptation model, their evolution into large-scale types may relate to aerodynamic instabilities originating at the saltation curtain–air interface.
GRAVELLY RIPPLES AND THE ENVIRONMENT
This report describes exceptionally large and coarse wind ripples in the southern Argentinean Puna Plateau (Fig. 1). These ripples are more than two times larger than the largest granule ripples documented elsewhere, reaching 43 m in wavelength and 2.3 m in amplitude, and they cover entire basins instead of being local, creating a landscape comparable only to Martian counterparts in dimensions and areal extent. The report describes the basic characteristics of these ripples and proposes hypotheses to explain their unusual size.
Satellite image of Puruya and Carachi Pampa ripple fields, showing positions of linear yardangs and their relations to large-scale ripples, both enlarged in insets. Lower right inset shows one of the unique high-altitude wind data sets near this latitude, depicting very fast winds. Location is indicated in large-scale map.
Gravel may be moved by wind for relatively short distances, generating a well-known bedform called a granule ripple (Sharp, 1963; Freyberger et al., 1992) or megaripple (Greeley and Iversen, 1985, p. 154), usually with grains no coarser than 4 mm. Originally described by Bagnold (1941) as bedforms (ridges), they were also described as deflation relicts, but most studies suggest that granule ripples form like common wind ripples (Ellwood et al., 1975), also called impact or ballistic ripples. Wind granule or gravel ripples on Earth are reported to reach 20 m in wavelength and >60 cm in amplitude (Bagnold, 1941, p. 155; Williams et al., 2002) and with mean grain sizes as large as 9.6 mm (Lancaster et al., 2002).
The ripples reported here show mixtures of two main clast types: dark-colored varnished volcanic clasts with ~2.43 g/cm3 density, and light-colored pumice clasts with ~0.91 g/cm3 density. They occur in two main fields. The Carachi Pampa is ~150 km2 and extends from 2950 to 4100 m elevation, and the Puruya field is ~80 km2 and extends from 3650 to 4100 m elevation. A minor ripple field of ~8 km2 near Puruya exposes the only large-scale ripples observed. In all areas, deflation has partially removed a preexisting alluvial cover, exposing an underlying tephra unit that yields large quantities of pumice. Difficult access has limited the samples available to date.
There are no wind measurements in this area, but data collected at a mine located 300 km south at a comparable altitude showed that wind gusts may exceed 400 km/h (Fig. 1). Wind-generated large-scale erosion features have been compared to Martian landforms (Wells and Zimbelman, 1989). Present winds, mainly blowing from the east-northeast, generate dust columns hundreds of meters tall. Sand and dust are both evacuated from this plateau; sand accumulates at the plateau edge, drowning older fluvial valleys, and the dust is carried to the Pampas to form loess (cf. Gaiero, 2007).
DIFFERENT SCALES OF GRAVEL RIPPLES
Small-Scale Gravel Ripples
Small-scale gravel ripples are present over flat gravelly surfaces and are superimposed on larger gravel ripples and troughs and are sometimes oblique to the direction of crest lines. They form quickly; flattened surfaces became rippled after a single windy season (Fig. 2A). Mean grain size ranges from 6 to 13 mm. The amplitude is a few grains thick (2–5 cm) and wavelengths are 30–40 cm for finer-grained ripples, reaching 50–90 cm in coarser-grained ones, showing correlation between wavelength and grain size. They may pass laterally into medium-scale ripples (Fig. 2A), which are formed by a progressive deepening of troughs, merging of crests, and progressive segregation of the coarsest pebbles toward crests. Evolution into larger-scale bedforms seems to occur by winnowing of fine-grained fractions of a preexisting sediment blanket, and transport of the coarse lag fraction up to larger ripple crests. Ripples evolving from thinner sediment blankets end as isolated ridges of well-sorted gravel separated by clean troughs over bedrock, like those ripples shown by Williams et al. (2002). The widespread occurrence of these small-scale gravel ripples suggests that they might be a necessary precursor for the formation of larger-scale types and responsible for wind-drifted gravel transport within larger ripples.
Gray arrows indicate wind direction. A: Small-scale gravel ripples, passing laterally into medium-scale varieties. Car track (22 cm wide) for scale. B: Medium-scale gravel ripple reaching height of almost 1.8 m. C: Field of linguoid megaripples passing laterally into straight crested over obstacle; note ripple wavelength increase as wind passes obstacle crest. D: Medium-scale ripple interior showing alternating pumice- and gravel-rich beds. Pencil is 11 cm long. E: Large-scale ripple, showing small-scale ripples over bedrock, and how light and dark clasts are segregated in the granular part. Bike is 2.1 m long. F: Large-scale ripple showing foresets and boundary between bedrock and granular covering material (black arrow). Shovel (~ 30 cm long) overlies bedrock.
Medium-Scale Gravel Ripples
Medium-scale gravel ripples are the dominant bedform at both studied fields. Ripple trains have ~15 m wavelengths and 60–80 cm amplitudes on average, with maxima of 18–20 m wavelength and 1.2 m amplitude for trains, and 1.8 m (Fig. 2B) for single bedforms near obstacles. Upwind slopes are 10°–15° and are capped by well-sorted dark varnished volcanic pebbles with a predominant (80%) mode at 6–10 mm. Lee faces dip 22°–25° near the crest, decreasing rapidly to 10°–15°, showing more light-density pumice gravel and sand as in troughs (Fig. 2E). Gravel ripples change in dimensions and grain size over hills. Ripples at hillcrests are the coarsest (mean grain size ~13 mm) and the smallest (5–10 cm amplitude and 2–3 m wavelength) (Fig. 3). Downwind from the hillcrest, ripple values shift progressively to normal values (6–8 mm mean grain size, 60–80 cm amplitude, and 10–16 m wavelength). Thus, an inverse correlation between grain size or wind speed and ripple size occurs over obstacles, contrary to what has been reported elsewhere (cf. Seppälä and Lindé, 1978).
Change of ripple crest design and wavelength in association to 30–50-m-high obstacles in Puruya field. A: Obstacle portrayed onsite. B: Obstacle portrayed by aerial photograph. C: Sketch of thickness variation of transport layer subjected to possible gravity waves showing also increasing size of vortices and compared to wavelength change (D) as observed over another obstacle. Ripple index is wavelength/height ratio.
Ripple foresets dip at ~20° maximum, suggesting they are not avalanche slipfaces, and show pebble-rich and sand-rich foreset laminae alternating in 3–7-cm-thick beds (Fig. 2D). Crest plan shapes are variable: linguoid and sinuous shapes characterize deflated corridors, while straight-crested varieties are common on flat areas and over obstacles (Fig. 3). Isolated ripples occur where sediment availability is low, and they show either a V form with lee faces opening at ~140°, or a crescent shape with horns opening 140°–160° downwind, seemingly dependent on local wind patterns. Ripple troughs expose a consolidated tuff or a silt-rich deposit called “chusca,” although in a few places unconsolidated sediments similar to the ripple material occur. Clasts as large as 200 mm in diameter are sparsely distributed on trough surfaces. Medium-scale ripples advanced as much as 20 cm over mining trails abandoned 4 yr ago in Puruya, and the same ripple types have advanced several hundred meters over the Carachi Pampa salt pan, possibly throughout the Holocene, which has been dryer. If the alternating laminae sets are produced by seasonal wind variations, maximum migration rate would be 5–4 cm/yr.
Large-Scale Gravel Ripples
Large-scale gravel ripples only occur at one small field of ~8 km2, northeast of the Puruya Basin, which is flanked upwind and downwind by linear yardangs with a mean spacing of 180 m and 20–40 m relief. Ripple wavelength sampled across the entire field is 24.2 m (n = 150), with very straight and low anastomosing crests. Maximum wavelength measured in situ was 44 m, with an amplitude of 2.3 m. As bedrock is exposed in troughs, the granular parts compose only ~15% of the field, almost entirely collected at the ripple crests. This is comparable to the largest ripples previously reported by Williams et al. (2002). Crests are entirely composed of normal-density clasts (2.42 g/cm3). Two samples showed mean clast sizes of 19 mm and 16.7 mm (Fig. 3). Light-density clasts occur on the lee side, at variable distances from the crest (Fig. 2E). Isolated clasts at troughs are clustered in small-scale gravel ripples.
These ripples differ from medium-scale ones in the following ways. (1) Troughs are eroded into the tuff: the bedform curved profile starts from the crest to the lee-side granular part of the ripple, continues into the bedrock, and ends without topographic breaks in the stoss-side clasts of the next crest (Fig. 2F). (2) The lowest foreset directly overlies the bedrock, a boundary cropping out on the stoss face, parallel to and near the crest (Fig. 2F). (3) Lateral ripple terminations are located on the lee sides of smooth bedrock ridges, suggesting that sheltering promotes bedform initial location. Further evolution probably shapes the complete waveform of these ripples, capturing fugitive grains at crests and eroding troughs into the bedrock, with minimal crest migration. (4) Sparse vegetation at crests and signs of caliche suggest these bedforms are more stable than medium-scale ripples (Fig. 2E). (5) The higher crest linearity and lesser crest junction areal density suggest a more mature bedform network, according to Werner and Kocurek (1999).
INTERPRETATION AND DISCUSSION
Most gravel ripples seem to develop initially as small-scale ripples on flat, pebbly surfaces. They further evolve by winnowing of fine grains, causing a deepening of troughs and an overall lowering of the surface due to deflation, forcing coarse material to migrate up to the dominant crest. True bedform migration occurs in small and medium types until they reach a large, stable crest. Large clasts that cannot be moved are left in the troughs; pebbles from 4 to 40 mm are trapped on some dominant gravel ripple crests, and finer grains are trapped at the lee face or winnowed. The end product of this selective deflation process is a large ripple overlying bedrock, with all fine grains (<2–4 mm) winnowed. If crest material is limited in availability, then isolated ripples form; otherwise, well-organized ripple fields form (linear crests) with few triple junctions. Exceptions to this selective deflation occur at places such as behind obstacles, where traditional ripple accumulation occurs.
The correlation between ripple shape and position with respect to obstacles suggests that their shape relates to wind structure (Fig. 3). Linguoid shapes and sinuous crests are common around obstacles, where turbulence increases, while straight crests occur where wind direction is coherent. This, together with other features described above, suggests that gravel ripples are formed by a combination of processes. The most likely mechanisms are suggested below.
Ballistic Hypothesis
This hypothesis may apply to the widespread small-scale gravel ripples that form rapidly. The correlation between wavelength and grain size suggests that they are connected to a saltation or reptation path length (Anderson, 1987). Medium- and large-scale gravel ripples seem not related to a ballistic mechanism, due to the fact the coarsest ripples are over hillcrests where the wind is fastest. This should result in longer grain path lengths due to both grain-size and wind-speed increase, but instead they show lower amplitude and wavelength (Fig. 3). They can also be as much as three times larger (wavelength, amplitude) without any change in grain size and with a likely similar age, seemingly controlled by underlying topography. Thus, the ballistic hypothesis is not consistent with medium-scale and large-scale ripples.
Gravity Wave Hypothesis
Medium- and large-scale ripples might be formed in response to Helmholtz instabilities as a result of shear along the interface
at the top of the saltation layer (cf. Brugmans, 1983). The abundance of low-density grains and chusca dust may help to generate a denser saltation layer that behaves as a boundary
layer, thinning and speeding up over obstacles (Fig. 3), where the velocity difference at the sheared fluid boundary decreases together with related hydrodynamic instabilities,
causing wavelength to decrease. The Helmholtz gravity-wave formula explains the progressive growth into large-scale ripples,
as these hydrodynamic instabilities grow at a rate of eμt (t is time) with μ given by
(1) where λ is the wavelength and U and U′ are the flow speeds of the two fluids (Drazin and Reid, 1981). These perturbations become stable at a critical wavelength λc:
(2) where ρ′ and ρ are the densities of the lower and upper fluids, respectively. This model thus explains the observed abnormal
ripple size only found in the presence of low-density clasts; the largest ripples at wind-protected places; growth until stabilization
of ripple profiles, including excavation of bedrock; and the lack of correspondence between clast size and ripple size. However,
in this model, the expected larger density contrast at the fluid boundary would act to decrease wavelength.
Roll Vortex Hypothesis
Roll vortices can explain the transverse wavelengths of features aligned parallel with the wind direction (i.e., cloud streets, dusty air streaks), and it has been suggested that transverse dunes between linear dunes evidence the presence of roll vortices. The similarity of progressively greater size of roll vortices as an obstacle is passed (cf. Greeley and Iversen, 1985) and the change of ripple dimensions observed over Puna's obstacles (Fig. 3) suggest this relation. Roll vortices, unlike gravity waves, scale with the flow depth, and their association to the largest ripples at Puna is also suggested by the regular linear yardangs upwind and downwind of that ripple field (Fig. 1). Using a relationship between the transverse wavelength of linear features and flow thickness (cf. Cooke et al., 1993), the yardangs suggest a 40–60-m-thick flow (the transport layer), fitting the order of magnitude of ripple wavelength, making this hypothesis plausible.
Extraplanetary Considerations
Mars surveys showed that large wind ripples are more frequent and are significantly larger than on Earth (Williams et al., 2002; Wilson et al., 2003; Wilson and Zimbelman, 2004), reaching wavelengths as long as 60 m. The largest terrestrial ripples previously reported reach 20 m in wavelength; the ripples reported here of wavelengths as long as 43 m help as potential analogs to Mars. Terrestrial wind ripples tend to share the same ripple index (RI: wavelength/height ratio) of ~15 (Fig. 4). Estimations of RI on Mars are 6.7 for a mean ripple spacing of ~38 m and an amplitude of ~5.7 m (Williams and Zimbelman, 2003). Thus, while the Mars and Puna large ripples overlap in size and share other similarities, like the tendency to grow larger under the sheltering of obstacles and a lack of slip faces, the groups are well differentiated by the RI.
Diagrams showing how the medium- and large-scale Puna ripples (grouped by dashed lines) relate to other known megaripples (MR) including those of Mars. AFB—Air Force Base.
Like the Puna ripples, the Martian ripples were considered too large to be aerodynamic (impact) ripples (Wilson et al., 2003). Thus the key for a unified megaripple genetic theory lies in the RI that synthesizes its wave-like profile, achieved through long periods of time, as shown by the perfect match between the bedrock excavated segments and the granular segments of the large-scale ripple profiles. The lower gravity of Mars would favor a higher critical wavelength λc; however, it does not explain the higher amplitude, as this is not addressed in the gravity wave formula. An improved understanding of large ripple development on Mars may thus be achieved by considering the turbulent structure of the saltation layer over large, stabilized ripples, such as the Puna ones.
CONCLUSIONS
Wind gravel ripples surveyed suggest that they are generated by a combination of processes. Initial ripple geometry might be related to reptation path length, but further evolution seems to be related to wind flow characteristics. The distribution of ripple sizes around obstacles suggests that their wavelength can be regulated either by Helmholtz-like aerodynamic instabilities due to shear at the saltation curtain and the free air interface, or by stable transverse roll vortices within the transport layer, ruling out the ballistic hypothesis. Both the gravity wave and the vortex hypotheses can explain some of the characteristics observed, and both require the development of a thick and concentrated sediment transport layer. This is consistent with the fact that these very large ripples are associated with an abundance of light-density grains. A better understanding of the structure of the wind layer where high sediment concentration occurs during strong wind events capable of moving fine gravel is needed in order to understand large ripples on both Earth and Mars. Thus in situ measurements are required in order to solve the megaripple problem.
Acknowledgments
This research was helped by G. Delendatti (Exeter), H. Sánchez, F. Segovia (Universidad Nacional de Tucumán), and Rod Stevens (Göteborg University). Geology reviewers were extremely helpful. Consejo Nacional de Investigaciones Científicas y Técnicas funded this field work.
Footnotes
-
- Received 29 July 2008.
- Revision received 9 December 2008.
- Accepted 11 December 2008.
- © 2009 Geological Society of America
