Megaripples dynamics under a series of storm events
Hezi Yizhaq1*, Itzhak Katra2, Ori Isenberg1
1
Institute for Dryland Environmental Research, Jacob Blaustein Institutes for Desert
Research, Ben-Gurion University of the Negev, Sede Boqer Campus, 84990, Israel.
2
Department of Geography and Environmental Development, Ben-Gurion University
of the Negev, Beer Sheva, 84105, Israel.
*Corresponding Author
Email: yiyeh@bgu.ac.il
Phone: +972-8-6596789
Fax: +972-8-6596921
Abstract
Megaripples in Nahal Kasuy in the southern Negev desert of Israel are characterized
by a mean wavelength of about 70 cm and by a bimodal distribution of coarse
and fine particle sizes, the latter of which is necessary for megaripple
formation. We show that storms can affect megaripples in different ways
depended on their sizes. Larger enough megaripples which their crests
protrude above the saltation layer can be flatten as the wind dislodges the
cover of the coarse particles at the crest. Medium megaripples can be broken
into smaller segments or change their orientation but the characteristic
bimodal distribution of particles at the crests remains. In contrast, smaller
megaripples can grow under the action of storms. These different behaviors
are further depended on the wind velocity and on the size of the coarsest
particles at the crests. The effects of storms are non spatially uniform but
locally depended on the specific characteristics and morphometry of the
ripples. Thus, megaripple height, which was believed to grow indefinitely
(Bagnold, 1941) is self-limiting, and our results also explain the positive
correlation between maximum grain size at the crest and ripple wavelength.
INTRODUCTION
Aeolian ripples, which form regular patterns on sandy beaches and desert floors,
indicate the instability of flat sand surfaces under the wind-induced transport of sand
grains. Two different kinds of sand ripple are observed in nature—normal ripples and
megaripples (Bagnold 1941; Sharp 1963). Their main features are summarized in
Table 1.
Megaripples have been described in many places, among them the Kelso Dunes
and Coachella Valley sands in Southern California (Sharp, 1963), in the Libyan desert
(Bagnold, 1941; El-Baz, 1986), the northern Sinai (Tsoar, 1990), Swakopmund,
Namibia, (Fryberger et al., 1992), northeastern Iceland (Mountney and Russell, 2004),
the coast of northeastern Brazil (Yizhaq, 2008) and on the Great Sand Dunes National
Park and Preserve in south-central Colorado (Zimbelman et al., 2011). Ginat
megaripples were documented in Carachi Pampa, Argentina, at a height of 4000 m
above mean sea level (Milana, 2009). Composed of volcanic pebbles, these
megaripples were formed by the action of extremely strong winds (probably the
strongest winds known on Earth, ~400 km/h). Megaripple wavelengths were up to 43
m and their heights were about 2.3 m (Milana, 2009) with a crest maximum grain size
of 19 mm.
Table 1: Main features of normal aeolian ripples and megaripples
The physical mechanism responsible for the formation of sand ripples is the action
of the wind on loose sand. When wind strength exceeds some threshold, grains
displaced by the direct action of the wind are lifted into the air. However, even strong
winds cannot keep sand grains suspended indefinitely (they are too heavy), and
therefore, they eventually fall to the ground. During their flight, sand grains reach
velocities that are approximately equal to that of the wind. Upon impact with the
ground surface, the grains impart energy and momentum into the sand and eject other
grains. For sufficiently high wind velocities, the bombardment by sand grains
accelerated by the wind generates a cascade process, creating an entire population of
saltating grains “hopping” on the sand surface. When the saltating, high-energy grains
collide with the bed, their impacts eject smaller, lower energy grains, termed reptons
(Andreotti et al., 2004). The windward slope, characterized by small bumps, is
subjected to more impacts than the lee slope. As such, the flux of reptons is higher
uphill than downhill, which causes the bumps to increase in size. Size analyses of
grains from different parts of the megaripples and from normal ripples showed that a
bimodal mixture of grain sizes is needed for megaripple formation and that the coarse
particles are more abundant at the crest (Isenberg et al., 2011; Yizhaq et al., 2009).
In a recent study (Isenberg et al., 2011), we used a photogrammetric technique to
show that megaripples start out as normal ripples and grow due to a rapid coarsening
process. Their evolution is a function of wind power and of the variability in wind
direction. The final wavelength is not simply correlated to the mean saltation length
(Elwood et al., 1975), but rather, it develops through the interactions between ripples
of different sizes (Isenberg et al., 2011; Yizhaq et al., 2009). Larger wavelengths
probably reflect longer development times and stronger winds, characteristics
common to bedforms in different environments, such as ripples and dunes in rivers,
oceans, and deserts (Werner, 1995).
The megaripple system exhibits self-organization, such that spatio-temporally
ordered structures emerge spontaneously (Anderson, 1990; Kocurek and Ewing, 2005;
Werner, 1995). During megaripple evolution, the fraction of coarse particles at the
crest increases, leading to the development of an armored layer that protects the
megaripple from wind erosion and that enables its continued growth. Strong winds
above the fluid threshold of the coarse particles, however, can erode the armored layer
and destroy the megaripples, as we recently observed (Isenberg et al., 2011, Yizhaq et
al., 2012). Therefore, a correlation exists between the large grains at the crest and
megaripple wavelength and height: the coarser the grains at the crest, the larger the
wavelength. Despite the knowledge gained about ripple dynamics, we still possess
only a rudimentary understanding of megaripple formation and destruction processes,
which dictates the need for more quantitative research.
The goal of the following work is to study the spatiotemporal dynamics of
megaripples under a series of storms during February and March 2009. Recently,
(Isenberg et al., 2011) we showed that these storms flattened one of the plots in the
study site (plot D, in Figure 2) covered with large megaripples. There are two
mechanisms that can mobilize the coarse particles on the megaripple crests and bring
them into saltation: direct lifting by fluid drag and indirect lifting due to the impacts
of saltating particles (‘splash’). For flat sand beds, theory and numerical models show
that the absorption of wind shear stress by saltating particles reduces the wind stress at
the surface to a value below the fluid threshold (e.g., Ungar and Haff, 1987; Werner,
1990; Anderson and Haff, 1991; Andreotti, 2004; Kok and Renno, 2009). Particles on
the surface are thus sheltered from the wind, and the wind stress at the surface
actually decreases as u* increases above the fluid threshold. For a flat sand bed,
coarse particles will thus not be lifted by fluid drag even for u* > u*t. However, this
situation may differ at the megaripple crest, which protrudes into the saltation layer,
thereby reducing the sheltering effect at the soil surface due to wind stress absorption
by the saltating particles above. Field measurements made by Greeley et al. (1996)
and Namikas (2003) and compiled in Fig. 5 from Kok and Renno (2008) indicate that
the mean saltation height for ~250-µm sand grains is ~3-4 cm. Since the height of the
megaripples at Nahal Kasuy was about 5 cm, their crests likely protruded from the
saltation layer, and as such, they would have experienced substantially higher wind
shear stress than that felt in the troughs. As a consequence, an increase in u* would
thus produce an increase in the surface wind stress at the megaripple crest to the point
that coarse particles could be lifted. For a given megaripple height, there thus exists a
threshold shear velocity below which megaripples grow but above which the
megaripples are flattened due to fluid entrainment of the coarse grains. Moreover, this
mechanism implies a feedback between the height of the megaripple and the wind
speed at which it is flattened: the higher the megaripple, the more it protrudes into the
saltation layer, the higher the shear stress at the crest surface at a given u* , and thus,
the lower the critical u* at which coarse particles can be lifted by the fluid. This effect
seems to explain why the high megaripples disappeared during the storms while the
smaller megaripples often did not as we show in this work. Moreover, in other parts of
the study site, incipient megaripples were formed by the same storms.
Our main goal here, is to show that storms impact locally depends on ripples
morphology, such that it differs from one specific location to the other in the same
field. The same series of storms can destroy large megaripples but build small ones.
MATERIALS AND METHODS
Field experiment
Our field experiment was carried out in the southern Negev at Nahal (wadi)
Kasuy sand dunes (Figure 2), which cover an area of 15 km2 (Ginat, 1991). Seif,
falling and climbing dunes developed in the area.
Sand in this area is comprised 60% calcite and 35% quartz. It drifts into Nahal
Kasuy from the Uvda Valley on southwestern storm winds and piles up in the wadi
bed (Fig. 2). The sand particles were transported from the valley margins and from the
extensive Pliocen Conglomerate, west of Uvda Valley. The particles were deposited
about 3 km westward in the Hiyyon stream in an area that contains 70 m of fine
alluvial sediments. Western winds transported the sand into valleys of Nahal Kasuy
and Nahal Yitro, and the northern winds accumulate it into dunes.
The annual precipitation here is about 37 mm concentrated in the rainy season
(November to April), and shrubs of Haloxylon persicum cover the wadi bed sparsely.
The megaripple field is located in the middle of the wadi, where coarse grains
abound. The mean megaripple wavelength is about 70 cm, with a mean height of
about 7 cm ( RI 10 ; ripple index defined as the ratio between the wavelength to
height). Smaller ripples superimposed on the megaripples reflect the most recent wind
direction. Because the Kasuy megaripples are small compared to those in other parts
of the world, they are expected to be more sensitive to the storms that form and
modify them and that can even destroy them (Isenberg et al., 2011).
To study mega-ripple evolution we flattened four plots (plots A, B, C and F) and
hand mixed the grains to achieve a uniform distribution of coarse and fine grains. The
plot sizes and the dates of flattening are given in Table 2 (see also the map in Figure
2). The fourth (D) and the fifth (E) plots were not flattened but were marked to track
changes of the large and medium size mega-ripples. In this work we show results
from plots B, D (see Isenberg et al., 2011), F (see Yizhaq et al, 2012) and area N.
Table 2: Plot descriptions at Nahal Kasuy field study.
Wind measurements
Wind speed and direction measurements were taken at 10-min intervals at a height
of 3.3 m using an anemometer recorder placed at the eastern edge of the megaripple
field. To complement our data, we compared it with the wind data (averaged hourly)
from a nearby meteorological station at Uvda airport (30N, 34.883E; 3 km west to
Nahal Kasuy). There was good agreement between the wind measurements from the
two locations (Yizhaq et al., 2009). The wind speed was used to calculate the drift
potential (DP) and the resultant drift potential (RDP) (Fryberger, 1979). Theoretical
and empirical studies show that the potential sand volume transported by the wind
through a 1-m-wide cross section per unit time is proportional to DP (Bullard, 1997;
Fryberger, 1979). DP is calculated from
DP  u 2 (u  ut ) ,
(1)
where u is the wind speed (in knots; 1 knot = 0.514 m/s) measured at a height
of 10 m and averaged over time, and ut is the minimal threshold velocity (= 12 knots)
necessary for sand transport for a typical sand grain (with an average diameter of
0.25 mm) (Fryberger, 1979). RDP is the vector summation of DP from different
directions and over the n measurements. Mathematically, it can be written as:
RDP  RDPx 2  RDPy
2
(2)
n
n
i 1
i 1
where RDPx   DPi cos i , RDPy   DPi sin i , and  i is related to the wind
direction. The direction of RDP is referred to as the resultant drift direction RDD,
which is defined as RDD  arctan(RDD y / RDD x ) . RDD expresses the net trend of sand
drift, namely, the direction in which sand would drift under the influence of winds
blowing from various directions. The ratio of RDP to DP (RDP/DP) is an index of the
directional variability of the wind (e.g., RDP/DP=1 stands for unidirectional wind,
and RDP/DP=0 characterizes multidirectional winds that vectorially cancel each other
out). DP is the potential sand drift, but the actual sand drift potential further depends
still on the mean grain diameter, the degree of surface roughness, the amount of
vegetation cover, and sand moisture.
In order to use our wind speed measurements to calculate DP we extrapolated it to
the height of 10 assuming a logarithmic profile and calibrated surface roughness (see
Isenberg et al. 2011). Validation tests of the method vs. direct measurements
confirmed that it was very good (see Yizhaq et al., 2009; Isenberg et al., 2011).
Grain size analysis
Samples of sand, all of which were collected using the same method, were
retrieved from the field with a tin can (diameter 84 mm, height 35 mm) by pressing
the can into the cross-section of the ripple under study. The samples were scooped out
of the can with a flat scraper. Here, we concentrated on samples taken from the crests
as the crest GSD is a good indication of the ripples development (see Yizhaq et al.,
2012).
Average sample weight was 310 g (with values ranging from 282 to 336 g). Grain
size analysis was performed by ANALYSETTE 22 MicroTec Plus laser diffraction,
which measures particles in the range of 0.08 to 2000 μm. The preparation of each
soil sample included sample splitting for replicate samples by a micro-splitter device
and the removal of distinct organic matter. For the analysis, three replicates (4 to 5 g)
of each sand sample were dispersed in a Na-hexametaphosphate solution (at 0.5%)
and by sonication (38 kHz). Due to the negligible number of clay-sized particles in the
sandy samples, GSD data were calculated using the Fraunhofer diffraction model with
an error < 5.0%.
Photogrammetry
We produced digital photographs using red-green-blue (RGB) images from a
digital Nikon D80 camera with a Sigma 10-20 mm lens. Processing took place with
Erdas Imagine ver. 9.1 and its Leica Photogrammetry Software (LPS) extension. The
small focal-length of the lens (10 mm), which corresponds to a 94.5° field of view,
reduced the number of photographs needed to cover the plots. To avoid interfering
with plot dynamics, the imaging and the ground control point (GCP) markings had to
be made from outside each plot. The camera was mounted on a special rail (5 m long
and secured at each end to a tripod), along which it could be moved using two cords
attached to the camera for this purpose. We used a remote control cable to operate the
camera (see Figure 3). Image analyses were carried out with the LPS Project
Manager. To reduce the need for a large number of GCPs for each plot, the LPS
Project Manager uses the self-calibrating bundle block adjustment method. With this
approach, the internal geometry of each image and the relationship between
overlapping images is determined with a small number of GCPs.
The only manual process needed to implement this approach is geometric
rectification, after which the program automatically extracts all the data needed for the
“Automatic Terrain Extraction” feature embedded in the LPS. The main
photogrammetry output we used in this study was the DEM that provided the two
most important parameters in ripple measurement, i.e. wavelength and ripple height.
Measuring these parameters together with continuous wind measurements enabled us
to track temporal topographic changes.
DEM quality depended on many factors. We obtained the best results when
images where taken in the late afternoon when contrast was maximized. By selecting
the camera’s Auto Mode option, aperture and shutter speed were chosen
automatically; no significant deviations in color or hue were noticed among the
pictures (for more details about the method see Isenberg et al., 2011).
RESULTS
Wind speed measurements
Two strong storms were recorded during our study (Figure 4). In each of them the
wind speed reached 15 m/s (at a height of 3.3 m) and their prevailing direction was
south westerly typical to the winters storms . Table 3 summarizes the wind statistics
of the period 20/2-31/3/09. The windiest time interval was between 20/2-8/3/12 where
DP was 12.99.
Table 3: Wind data from Nahal Kasuy for the period 20/2-31/3/09. DP is the potential
drift potential; RDP is resultant drift potential; RDD is RDP direction; RDP/DP is the
wind directionality and t is the time (in percent) that the wind is above the fluid
threshold for sand transport (taken as 6 m/s).
Grain-size analysis
During these storms an intense saltation occurred in the plots as shown in Figure
5. Figure 6 shows the GSD of the saltation traps (near plot C). The medians (D50) w
during the first two periods were 217.4 and 188.4  m respectively whereas during
the calm period (24/3-31/3) the median was 177.7  m, which is still above the
median of typical normal ripples in Nahal Kasuy which is 158  m (Yizhaq et al.,
2009). The coarse mode during the period with the storm of 27/2 was 338.5  m.
These coarse grains have large momentum and kinetic energy which upon impact
with the surface can dislodge the grains at the crest into reptation and even to
saltation. Figure 7 shows the GSD of plots B, D F and N and the full statistics of the
GSD is in Table 4.
Table 4. Grain size distribution of samples taken from megaripple crests in the
different plots. F1and N megaripples after the first storm (23.03.2009). F2, B, and D =
megaripples after the second storm (31.03.2009).
The bimodality distribution has two aspects: one is by grain size segregation (the
difference between coarse and fine grain diameter;
The second is frequency segregation which can be described by the difference in the
frequencies between the two modes. In order to better describe these two aspects we
define here new parameters: 1 ,  2 which describe the normalized grain size
segregation and the normalized frequency segregation respectivell. In addition we
define  the resultant segregation vector and  the direction of  in the following
way:
1 
2 
Dc  D f
Dc
fc  f f
fc
, 0  1  1
,  1  1  1
(3)
  12   22 , 0    2
  arctan( 1 /  2 ) 0    1800
where subscript 'c' stands for coarse and subscript 'f' is for fine. Thus, each bimodal
distribution can be represented by a point in the plane spanned by the coordinates
(  2 , 1 ) as shown in Figure 8. Note, that for 2  0 the bimodal distribution is
inverted since the maximum frequency associated with fine mode, for 2  0 the
bimodal distribution is typical to well developed megaripples (see Yizhaq et al. 2012).
Each distribution can be represented in polar coordinates where  defines the distance
from the origin and  is the angle with positive axis (  2 ). The larger is the value of
 the larger the segregation. The different values of these parameters are defined in
Table 5. Note that that the size segregation ( 1 ) for plot D and N is the same, but
they differ in the frequency segregation (  2 ) which is negative for plot N and very
small but positive for plot D.
Within this representation the effect of the storm on the GSD can be graphically
represented by a vector which connects the two points 1 and 2(before and after the
storm event) in the plane. This vector has a size and a direction defined by:
  ( 1t 2  1t1 )2  (  2t 2   2t1 ) 2 , 0    2
  arctan(( 1t 2  1t1 ) / (  2t 2   2t1 )), 0    3600
(4)
The angle  denotes the direction of the change for 90    270 the change is toward
inverted bimodal distribution and for 0    90 and 270    360 the change is
toward bimodal distribution i.e. megaripples construction. For example, for the vector
 shown in Figure 8 which represents the effect of the storm of March 23rd on the
ripples in plot F,   0.21 and   238.00 which means the change was towards an
inverted bimodal distribution. This new method of representation of the GSD results
for bimodal distribution can graphically shows the time evolution of the segregation at
the megaripple crest which gives valuable information on the megaripple
morphodynamics.
Table 5.  analysis of the samples from Table 4.
Ripples morphodynamic in the plots
The storms had a different impact on the ripples in the different plots. In plot B which
was characterized by megaripples (Figure 9) with an average wavelength of 70 cm
and average height of 4 cm, the skeleton of the megaripples remained although the
megaripples were broken into small segments and part of them to normal ripples.
Figure 9 shows snapshots of the plot during the period between 20/2-31/32009. This
behavior is supported by the bimodal distribution (see Figure 7) of grains at the crest
which is typical to megaripples.
In plot F on March 23rd the area was cover by small megaripples (  ~ 40 cm ), but
on March 31st these ripples were covered by secondary normal ripples perpendicular
to their main axis. However, the small megaripples can still be observed in Figure 10.
Plot D exhibit the most dramatic change (Figure 11) as we already showed in Isenberg
et al. (2011). Addressing megaripple dynamics, Bagnold (1941) predicted that
megaripples will disappear and regular ripples will be formed when the coarse grains
in the crest begin to saltate. This is probably what happened in Plot D. According to
the rough estimation of the threshold velocity applicable in the absence of saltation of
fine particles, wind speeds above 15 m/s can drive the coarse grains into saltationThe
first storm of 27th February broke the crests into smaller segments and the second
storm of 23rd March continued that trend, further breaking the megaripples down into
normal ripples. Another important factor that contributed to the progressive
disappearance of the megaripples is the fact that the two storms came from the same
direction (west), making their effects cumulative. In contrast, we observed that when
the wind’s direction changes between storm events, the result will be a complex
pattern of megaripples (each with a broken crest-line) with smaller ripples extending
in different directions between them.
The assumption that megaripples were broken down due to coarse grain saltation is
supported by grain-size analyses of samples taken from Plot D (from ripple crests)
before and after the storms (Figure 13). The typical bimodal distribution of grain-sizes
in megaripple crests was replaced after the storm by a unimodal distribution, like that
which characterizes normal ripples.
In the area which we denoted as N, incipient megaripples (Figure 14) were developed
on 8th March with an averaged wavelength of 30 cm and 1 cm height. These ripples
can be considered between megaripples and normal ripples as can also supported
from the GSD from the crest which shows bimodal distribution but with a fine mode
maximum. Before the storms the area was covered by regular normal ripples and the
larger ripples probably developed by the abundant of coarse particles which the storm
brought into the area. Table 6 summarizes the different response of the plots to the
storms between 20.2.2009 and 31.3.2009.
Table 6: The different response of the plots to storms between 20.2.2009 and
31.3.2009.
DISCUSSION
The megaripples in Nahal Kasuy are relatively small compared to those in other
locations in the world (Yizhaq et al., 2011), which implies that they are more sensitive
to storms and may even disappear altogether when the wind is too strong, a
phenomenon also observed by Isenberg et al. (2011). In the current work we showed
that the effects of the storms varied between the plots, form complete destruction to
initial building of megaripples.
Ripple destruction are the result of two different processes. The first, which is more
common, is due to storms that blow perpendicular to the prevailing winds. The crests
break into shorter segments, thus changing ripple orientation as the overall ripple
pattern becomes more disordered and the mean wavelength decreases. The second,
more dramatic process involves a series of strong storms blowing from almost the
same direction. Such conditions can destroy the megaripple armored layer comprising
the coarse grains, which can lead to the building of small ripples instead.
Two mechanisms can mobilize the coarse particles on megaripple crests and
bring them into saltation: direct lifting by fluid drag and indirect lifting due to the
impacts of saltating particles (i.e., splash). For flat sand beds, theoretical and
numerical models show that the absorption of wind shear stress by saltating particles
reduces the wind stress at the surface to a value below the fluid threshold (e.g.,
Anderson and Haff, 1991; Andreotti, 2004; Kok and Renno, 2009; Ungar and Haff,
1987; Werner, 1990). Particles on the surface are thus sheltered from the wind, and
the wind stress at the surface actually decreases as u* increases above the fluid
threshold. Thus, for a flat sand bed, coarse particles will not be lifted by fluid drag
even for u* > u*t, where u*t is the fluid threshold for the coarse particles. However,
this situation may differ at the megaripple crest, which extends into the saltation layer,
thereby reducing the sheltering effect at the soil surface conferred by the absorption of
wind stress by the particles saltating above the crest. Field measurements made by
Greeley et al. (1996) and Namikas (2003) indicate that the mean saltation height for
~250-µm sand grains is ~3-4 cm. Since the heights of the Nahal Kasuy megaripples at
plot D were about 5 cm, their crests likely extended above the saltation layer, and as
such, they would have experienced substantially higher wind shear stress at their
crests than in the troughs. As a consequence, an increase in u* would thus produce a
sufficient increase in the surface wind stress at the megaripple crest such that coarse
particles could be lifted.
For a given megaripple height, there thus exists a threshold shear velocity, below
which megaripples grow but above which they are flattened due to fluid entrainment
of the coarse grains. Moreover, this mechanism implies a feedback between the height
of the megaripple and the wind speed at which it is flattened: the higher the
megaripple, the more it protrudes into the saltation layer, the higher the shear stress at
the crest surface at a given u* , and therefore, the lower the critical u* at which coarse
particles can be lifted by the fluid. This effect seems to explain why the higher
megaripples disappeared during storms while the smaller megaripples often remained
although their pattern changed like in plots B and F. Moreover, this effect implies
that megaripple height, which was believed to grow indefinitely (Bagnold, 1941), is
self-limiting, and it also explains the positive correlation between maximum grain size
at the crest and megaripple wavelength (Stone and Summers, 1972).
In addition to the direct fluid lifting mechanism, particles on the ripple crest could
also be mobilized by splashing, which dominates particle lifting during steady-state
saltation over flat sand beds (Anderson and Haff, 1991; Kok and Renno, 2009; Ungar
and Haff, 1987; Werner, 1990). As wind speed increases, the mean speed of
impacting, saltating particles stays approximately constant (Andreotti, 2004;
Creyssels et al., 2009; Kok, 2010a, 2010b; Kok and Renno, 2009), but the probability
distribution of impact speeds widens. This outcome is the result of the increase in the
difference in typical fluid speeds between the bottom of the saltation layer (where
fluid speeds decrease with u* ) and the top of the saltation layer (where fluid speeds
increase strongly with u* ), and it is evident both in measurements (see, for example,
Fig. 4 in Bagnold, 1938) and in numerical models (see, for example, Figure 3 in
Ungar and Haff, 1987, and Fig. 12 in Kok and Renno, 2009). The increase in wind
speeds near the top of the saltation layer causes the population of fast-moving
particles to grow, such that the chance of an unusually high saltator impact speed
(e.g., > 5 m/s) increases drastically with u* . Consequently, the fraction of saltating
particles impacting megaripple crests and splashing coarse particles into saltation
increases. For u*  0.5 m/s, which is representative of the winds during the storms in
Nahal Kasuy (15 m/s at 3.3m height), only ~0.1% of the mass flux is due to the coarse
fraction (Isenberg et al., 2011). This may explain why more than one storm was
needed to remove the coarse grains from the ripple crests and to flatten the
megaripples at plot D.
Coarse grain removal and ripple flattening are more pronounced for small
megaripples like the ones at Nahal Kasuy. Larger megaripples, which typically have
larger wavelengths and heights, are covered with coarser grains, and therefore,
especially strong winds are needed to change their morphologies. Because very strong
storm events are rare, however, the megaripple can continue to grow slowly until such
an extreme storm occurs.
These suggested mechanisms can also explain the formation of the incipient
megaripples at N, because they are relatively small and their crests are below the
saltation layer, the coarse grains cannot be entrained into saltation but rather reptate
and accumulate at the ripple crest. Thus, in this case the regular mechanism of
megaripples development is acting.
CONCLUSION
We show that the effect of storms on the evolution of megaripples depends on the
current development stage of the megaripples at the field. Higher and mature
megaripples have larger chances to be broken or flatten in storms. Smaller
megaripples will be broken to smaller segments but their main pattern will remain. In
contrast, storms can start the building process of incipient megaripples. We show that
these different responses to storms are also reflected by the GSD of grains taken form
crests.
Whether megaripples can continue to grow indefinitely over time as suggested by
Bagnold (1941) is still an open question that needs further study. Based on our
research in Nahal Kasuy, we conclude that small megaripples will reach equilibrium
between the coarsest particles at the ripple crest and the site’s characteristic winds,
which cannot drive the coarsest particles into saltation. Megaripple equilibrium will
be disrupted, however, in the event of an unusually strong storm, which can flatten the
megaripples and start the building process anew as we observed in the field. More
long-term studies of megaripple evolution and morphodynamics in other locations are
needed to confirm our conclusions for larger megaripples.
ACKNOWLEDGMNET
This work was supported by the Israel Science Foundation (grant N531/06). We thank
Roy Talbi for providing the wind data from Nahal Kasuy.
TABLE CAPTIONS
Table 1: Main features of normal aeolian ripples and megaripples.
Table 2: Plot descriptions at Nahal Kasuy.
Table 3: Wind data from Nahal Kasuy for the period 20/2-31/3/09. DP is the potential
drift potential; RDP is resultant drift potential; RDD is RDP direction; RDP/DP is the
wind directionality and is the time (in percent) that the wind is above the fluid
threshold for sand transport (taken as 6 m/s).
Table 4: Grain size analysis of samples taken from megaripples in different plots and
from sand traps located in the meggariple field.
Table 5:  analysis of the samples from Table 4.
Table 6: The different response of the plots to storms between 20.2.2009 and
31.3.2009.
FIGURE CAPTIONS
Figure 1
(a) The research area (indicated by a black square) is located in the southern Negev,
46 km north of the Gulf of Eilat. (b) An aerial photo of Nahal Kasuy. The megaripples are situated in the middle of the wadi (indicated by the white arrow; their
location is 29° 59' 14'' N; 34° 59' 25' E, 430 m above mean sea level). The white
triangular indicates the area of the new formed megaripples (denoted by N).
Figure 2
Schematic map of the plots in Nahal Kasuy. Plots A, B, C and F were artificially
flattened, while plots D and E were used to track megaripple spatial dynamics.
Figure 3
Field methods used in the research. (a) The rail: the camera was mounted on a special
rail (5 m long) that was fixed on two tripods at its edges. The camera could be moved
along the rail from one side to the other by two cords that were attached to the
camera. We used a remote control cable to take the pictures. The plot seen in the
picture is A with the GCPs used for geometrical rectification to derive DEM (b) The
pictures of the same area were taken from at least 2 different angles, in order to create
overlapping of around 60% of the area. The panel show the overlapping between the
pictures for plot B. (c) Marking holes on the surface of plot D by using a giant
“comb." The comb (5 m long) pinches the surface and leaves marks on the sand at
intervals of 15 cm that are used as GCPs. (d) A closer look at the holes made by the
comb.
Figure 4
Recordings (speed and direction) of the main storms that occurred during the course
of the study. The measurements were taken at 10-min intervals and at a height of
3.3 m. Note that the maximum wind speed was 15 m/s.
Figure 5
The field study under the storm March 23, 2009. (a) The saltation cloud near plot B;
(b) Plot F on March 23, 2009. The distances between the iron sticks are 0.5 m.
Figure 6
Grain size distributions of sand from saltation traps during the unusually windy
Marchof 2009. Ts1and Ts2 = sand traps sampled during the first and the second
storms, respectively. T = sand trap sampled under weak winds. Note that during the
storms coarser grains move in saltation.
Figure 7
Grain size analysis during the study of samples taken from the megaripple crests.
Table 4 gives the statistical parameters of the GSD.
Figure 8
 analysis of the samples from plots B, D, F and N. Each bimodal distribution is
represented by a point in the plane spanned by the axis 1 (size segregation) and  2
(frequency segregation). The arrow indicated the effect of wind on the GSD between
23rd -31st of March 2009 on the megaripples in plot F (see Table 5 for more details).
Figure 9
Plot B before on November 11, 2008, two months before the storms. (a) The DEM of
the plot extracted from plot photo (b). White color describes the higher elevation
places and the dark colors represent the lower places. The inset shows a cross section
between the points 1 and 2 (along the wind direction). The average ripple height is
around 4 cm.
Figure 10
Snapshots of plot D during the period between 20th February and 31st March
2009. The large megaripples in this plot were destroyed by the strong storms, and
new, small ripples took their place (see also Isenberg et al, 2011).
Figure 11
Snapshots of plot F during the period between 20th February and 31st March
2009. The normal ripples reflect the action of winds perpendicular to the prevailing
winds which created the megaripples.
Figure 12
Megaripple morphodynamics in plot B during 20th February (a) 8th March (b) and
31st March 2009 (c). The colors correspond to the megaripple crests on the successive
visits.
Figure 13
Grain size analyses of crest samples (plot D) taken before the storms (05.06.08), after
the first storm (23rd March 2009) and after the second storm (31st March 2009). The
typical bimodal distribution of megaripples changed to a unimodal distribution of
normal ripples. The grain size analysis for these samples have been done by means of
standard sieves suspended on a shaker. The aperture of each sieve was greate by 1/4 
then the one above (    log 2 d , where d is the grain diameter in mm). This method
is with a lower resolution compared to the other results we show which were obtained
by the by ANALYSETTE 22 MicroTec Plus laser diffraction.
Figure 14
Incipient megaripples at N on 8th March 2009. The white arrow shows the prevailing
wind direction. Note that the segregation is also observed in the darker color of the
coarse grains which concentrate at the ripple crest. The overall pattern of these new
formed ripples is with high order and with almost straight crests which are more
typical to normal ripples (see Yizahq et al., 2012).
Table 1
Normal Ripples
Megaripples
Up to 30 cm
30–43 m
>15
<15
Time scale
Minutes
Days and years
Sorting
Unimodal distribution of grain
Bimodal distribution of grain
sizes (typically 0.100–0.300 mm
sizes, with coarse grains
in diameter)
0.7−4 mm in diameter
Saltation and reptation (creep) of
Saltation and reptation of fine
fine grains
grains and creep of coarse grains.
Wavelength
( )
Ripple index
(RI)
Basic Processes
Table 2
Plot
Begin treatment date and characteristics
Size
A
Flattened in January 2008 and marked with iron
55 m
rods that indicated Ground Control Points.
B
Flattened in January 2007
5.55.5 m
C
Flattened in November 2006
44 m
D
March 2008, large megaripples
55 m
E
June 2008, medium megaripples
53 m
F
Flattened in October 2006
44 m
N
Normal ripples
20 x 10 m (located
300 m south to plot
C)
Table 3
Period
DP
RDP
RDD
RDP/DP
t[%]
21/2-8/3-2009
12.99
10.85
251
0.83
18.2
9/3-23/3-2009
5.16
4.0
257
0.78
6.3
24/3-31/3-2009
1.83
1.74
252
0.95
1.0
Table 4
Plot
F1
F2
B
D
N
Ts1
Ts2
T
Statistical parameters (µm)
Mean
Sorting
Mode 1
Mode 2
D10
D50
D90
363.9
252.0
613.5
168.7
96.6
279.6
729.5
248.2
139.0
373.5
168.7
93.7
221.9
446.1
371.8
248.9
613.5
206.0
86.4
301.0
704.8
251.7
148.2
412.5
168.7
93.2
214.5
470.1
261.1
160.6
455.5
186.5
100.5
211.5
510.2
234.7
128.3
338.5
227.0
96.6
217.4
408.5
214.8
123.5
308.7
168.7
84.33
188.4
395.0
180.5
85.6
227.0
---80.5
177.7
292.8
Fraction weight (%)
Clay (< 2 μm )
Silt (2-50 μm )
Fine sand (50-250μm )
Medium sand (250-500 μm )
Coarse sand (500-2000μm )
1.0
2.9
44.2
22.5
29.4
1.2
3.5
51.2
28.4
15.6
1.3
4.2
41.5
17.6
35.4
1.2
3.2
55.7
35.0
4.8
1.2
3.2
58.1
29.2
8.4
1.5
3.7
56.8
36.8
1.2
1.5
3.9
63.5
30.1
1.1
2.2
4.1
76.5
17.1
0.0
Note: F1and N = megaripples after the first storm (23.03.2009). F2, B, and D =
megaripples after the second storm (31.03.2009). Ts1and Ts2 = sand traps sampled
during the first and the second storms, respectively. T = sand trap sampled under weak
winds.
Table 5
F1
F2
B
D
N
Mode 1 (  m)
613.5
373.5
613.5
412.5
455.5
Frequency (%)
6.24
6.73
9.32
5.78
5.04
Mode 2 (  m)
168.7
168.7
206
168.7
186.5
Frequency (%)
4.43
5.53
4.01
5.74
6.83
1 (size
0.725
0.548
0.664
0.591
0.591
0.290
0.178
0.570
0.007
-0.355

0.781
0.577
0.875
0.591
0.689
 (deg.)
68.0
72.0
49.0
89.0
121.0
segregation )
 2 (frequency
segregation)
Table 6
Plot
Before 20.2.2009
On 31.3.2009
B
Developed megaripples. Typical
The area covered with
bimodal distribution.
normal ripples but the
megaripples still remained
in the plot. Bimodal
distribution at the crest.
D
Large megaripples with normal ripples
Destruction of megaripples
between them. Bimodal distribution of
to normal ripples. Inverted
grains at the crest with coarse mode of
bimodal distribution.
780  m.
F
Disordered small megaripples.
Disordered small
megaripples with normal
ripples perpendicular to
the megaripples.
N
Regular normal ripples. Unimodal
Incipient megaripples with
distribution,
inverted bimodal
distribution.
REFERENCES
Anderson, R.S., 1990. Eolian ripples as examples of self-organization in
geomorphological systems. Earth-Science Reviews. 29, 77-96.
Anderson, R.S., Haff, P.K.,1991. Wind modification and bed response during
saltation of sand in air. Acta Mechanica Supplement 1. 21-51.
Andreotti, B., 2004. A two-species model of aeolian sand transport. Journal of Fluid
Mechanics. 510, 47-70.
Bagnold R.A., 1941. The Physics of Blown Sand and Desert Dunes. Methuen,
London.
Bagnold, R.A., 1938. The measurement of sand storms, Proceedings of the Royal
Society of London Series. A-Mathematical and Physical Sciences. 167(A929), 02820291.
Bullard, J. E., 1997. A note on the use of Fryberger method for evaluating potential
sand transport by wind. Journal of Sedimentary Research. 67, 499.
Creyssels, M., Dupont, P., Ould el Moctar, A., Valance, A., Cantat, I., Jenkins, J.T.,
Pasini, J.M., Rasmussen, K.R., 2009. Saltating particles in a turbulent boundary layer:
experiment and theory. Journal of Fluid Mechanics. 625, 47-74.
Fryberger, S.G., 1979, Dune forms and wind regime. In McKee, E.D. (Ed). A Study
of Global Sand Seas. U.S. Geological Survey Profile Pap. 1052, Washington, pp. 137169.
Ginat, H., 1991. The Geology and Geomorphology of the Yotvata Region. Thesis
submitted for the degree of Master of Science. The Hebrew University of Jerusalem
(In Hebrew).
Goldreich, Y., 1998. The Climate of Israel: Observations, Research and Applications.
Bar-Ilan Press, Ramat Gan.
Greeley, R., Blumberg, D.G., Williams, S.H., 1996. Field measurements of the flux
and speed of wind-blown sand. Sedimentology. 43, 41-52.
Isenberg, O., Yizhaq, H., Tsoar, H., Wenkart, R., Karnieli, A., Kok, J. and Katra, I.
2011. Megaripple flattening due to strong winds. Geomorphology, 131, 69-84.
Kocurek, G., Ewing, R.C., 2005. Aeolian dune field self-organization – implications
for the simple versus complex dune-field patterns. Geomorphology. 72, 94-105.
Kok, J.F. Renno, N.O., 2009. A comprehensive numerical model of steady state
saltation (COMSALT). Journal of Geophysical Research. 114, D17204,
doi:10.1029/2009 JDO11702.
Kok, J.F., 2010a. Difference in the wind speeds required for initiation versus
continuation of sand transport on Mars: Implications for dunes and sand storms.
Physical Review Letters. 104, 074502. doi: 10.1103/PhysRevLett.104.074502.
Kok, J.F., 2010b. An improved parameterization of wind-blown sand flux on Mars
that includes the effect of hysteresis. Geophysical Research Letters. 37, L12202.
Manwell, J.F., McGowan, A.L., Rogers, A.L., 2009. Wind Energy Explained. Wiley,
Chippenham, Wiltshire, U.K.
Namikas, S.L., 2003. Field measurement and numerical modelling of aeolian mass
flux distributions on a sandy beach. Sedimentology. 50, 303-326.
Sharp, R.P., 1963. Wind ripples. Journal of Geology. 71, 617-636.
Stone, R.O., Summers, H.J., 1972. Study of Subaqueous and Subareial Sand Ripples.
US Office of Naval Research, Final Report. USC Geology 72-1, Arlington, Virginia,
274 pp.
Ungar, J.E., Haff P.K., 1987. Steady-state saltation in air. Sedimentology. 34, 289299.
Werner, B.T., 1990. A steady-state model of wind-blown sand transport. Journal of
Geology. 98, 1-17.
Werner, B.T., 1995. Eolian dunes. Computer simulations and attractor interpretation.
Geology. 23, 1107-1110.
Yizhaq, H., Isenberg, O., Wenkart, R., Tsoar, H., Karnieli, A., 2009. Morphology and
dynamics of aeolian megaripples in Nahal Kasuy, southern Israel. Israel Journal of
Earth Sciences. 57, 145-161.