<< . .

. 4
( : 51)

. . >>

tion, we will focus ¬rst on the transport of dust in most of the sediments it carries to settle out and
arid environments (i.e. silt and clay transport), us- be deposited on the playa. This dust is then sus-
ing Franklin Lake Playa in Amargosa Valley, Cali- ceptible to windblown transport when the playa
fornia, as a type example. Second, we will explore becomes dessicated. Some of this windblown dust
the transport of sand across an alluvial fan using is deposited immediately downwind (e.g. on the
Whiskey Petes™ alluvial fan as an example. Finally, alluvial fan terraces of Eagle Mountain alluvial
we will explore the formation mechanisms of eo- fan), while some is transported thousands of kilo-
lian dunes. meters away to the ocean or to a river that ulti-
mately drains into the ocean. As such, the dust
1.2.1 The dust cycle in arid environments cycle is not a closed system on the continents: at-
The dust cycle in arid environments refers to the mospheric transport results in a net transport of
transport of dust as it moves from sources (e.g. dust from continents to the ocean. Rock weather-
playas and dry channel beds) to sinks (e.g. alluvial ing continually generates new dust for this cycle.
fan terraces) and ultimately back to sources, pri- The amount of dust emitted from playas
marily by the action of ¬‚oods. Franklin Lake Playa varies greatly in space and time. The hydrologic
and the Amargosa River drainage basin (Figure state of a playa plays a signi¬cant role in con-
1.15) provide a classic example of the dust cycle in trolling this spatial and temporal variation in
action. Most of the dust that is deposited on the dust-emitting potential (Fig. 1.16) (Reynolds et al.,

2007). Generally speaking, wet playas (those with of the surface elements (e.g. sand ripples, plants,
shallow groundwater tables and hence signi¬- rocks, etc.). Particles can be entrained by the wind
cant surface moisture) can be expected to emit when the shear velocity is greater than a thresh-
less dust than dry playas under similar wind con- old value, u—td , given by
ditions, because in wet playas the surface mois-
ρs ’ ρa
ture creates a cohesive force between grains that u—td = A gD (1.15)
is not present in dry playas. Franklin Lake Playa,
however, is one of the most active playas in the where A is a constant of proportionality (equal
western US despite its very shallow water table to 0.1 for air), ρs is the density of sediment, ρa
(less than 3 m below the surface in most por- is the density of air, g is the acceleration due to
tions of the playa). The shallow water table be- gravity, and D is grain size diameter.
neath Franklin Lake Playa causes a vapor dis- If the soil is wet, Eq. (1.15) must be modi¬ed
charge that disrupts the formation of surface to include the effects of cohesion. Chepil (1956)
crusts that would otherwise serve to protect the developed an empirical equation to represent the
surface from erosion. The result is a soft puffy effects of moisture on the threshold shear veloc-
surface on Franklin Lake Playa that promotes ity:
dust emission (Czarnecki, 1997). In general, dust
production on playas is relatively low for very wet 2
u—t = +
u2 (1.16)
or very dry playas, and relatively high for playas —td
ρa θ1.5
in an intermediate state.
where θ is the volumetric soil moisture and θ1.5
Temporal variations in dust emissions from
is the soil moisture at a pressure of ’1.5 MPa (i.e.
a single playa are also quite complex. In many
playas, dust emissions are observed to be nega- the wilting point).
tively correlated with antecedent precipitation. Quantifying the dust cycle is important for
In such cases, ¬‚ood events cause a pulse of geomorphic, pedologic, and hydrologic reasons.
recharge that raises the water table, wetting the Dust deposition, for example, controls the perme-
surface and increasing the threshold wind ve- ability of alluvial fan terraces and their rates of
locity. In other playas, however, dust emissions soil development. Windblown dust is also a very
are positively correlated with antecedent rainfall signi¬cant human health hazard. Windblown
(Reheis, 2006). In these playas, dust emissions are particulate matter includes both natural (e.g.
limited by the supply of ¬ne-grained material to mineral dust from playas) and anthropogenic
the playa. In such playas, signi¬cant dust storms (e.g. smoke from ¬res), but natural sources rep-
are produced only when antecedent ¬‚oods bring resent a signi¬cant portion of the total. Studies
suf¬cient sediment to the playa to be mobilized have shown a correlation between daily mortality
when the playa becomes dessicated. and high levels of particulate matter (PM) in the
Particles are entrained by the wind when the atmosphere (Samet et al., 2000). This means that,
Bernoulli lift created by wind ¬‚owing over the in addition to causing long-term respiratory ail-
top of the particle exceeds the weight of the par- ments, high dust concentrations also cause sud-
ticle. The Bernoulli force, in turn, is directly re- den death in some individuals. Figure 1.17 is a
lated to the shear velocity exerted by the ¬‚ow, photograph of a dust storm that hit Las Vegas,
given by (Bagnold, 1941) Nevada on April 15, 2002. Although such dust
κum storms have long been a hazard, rapid popula-
u— = (1.14)
tion growth in the western US may be contribut-
ln z0
ing to more dust storms through groundwater
where κ is the von Karman constant (equal to withdrawal and land disturbance.
0.4), um is the wind velocity measured at a height Atmospheric transport of particulate matter
zm above the ground, and z0 is the aerodynamic can be modeled as a combination of downwind
roughness length of the surface. The roughness advection, turbulent diffusion, and gravitational
length z0 , in turn, is related to the size and shape settling. The steady-state equation describing

whether or not the particles are falling under
gravity. Silt, for example, has a negligible settling
velocity but a ¬nite depositional velocity because
silt particles are deposited as they fall or lodge
between the crags of a rough surface (e.g. clasts
in a desert pavement).
Dust deposition plays an important role in
creating desert pavement, one of the most enig-
matic landforms of the arid environment. Desert
pavement is a type of surface landform formed
on alluvial fan terraces in low-elevation parts of
the western US and other arid regions around
the world. Desert pavement is characterized by
a monolayer of stones at the surface, each stone
Fig 1.17 Photograph of a dust storm in Las Vegas, Nevada,
tightly sutured to the next like pieces in a jig-
on April 15, 2002, when particulate matter (PM)
concentrations reached above 1400 μg/m3 . saw puzzle. Digging into the pavement reveals
that, in most places, the desert pavement sits
atop a silt-rich layer comprised predominantly
these processes is given by
of wind-blown material (Figure 1.19) (McFadden
‚ 2c ‚ 2c ‚ 2c ‚c ‚x et al., 1987; Wells et al., 1995). While desert pave-
+ 2+ 2 ’u +q =0
‚ x2 ‚y ‚z ‚x ‚z ment was originally thought to represent a lag of
(1.17) coarse material left behind as ¬ner material was
eroded out, the observation of a layer of ¬ne ma-
where K is the turbulent diffusivity, c is the
terial underneath the stony monolayer is most
particle concentration, x is the downwind dis-
consistent with a model for pavement formation
tance, y is the crosswind distance, z is the vertical
in which windblown silt and clay is deposited
distance from the ground, u is the mean wind
in the interstitial spaces between surface stones.
velocity, and q is the settling velocity (model
Once an initial layer of silt and clay is deposited
geometry shown in Figure 1.18a). Solutions to
on and immediately underneath the surface, the
Eq. (1.17) are known as Gaussian plumes. This ver-
reorganization of stony clasts can occur during
sion of the advection-diffusion-settling equation
wetting and drying events, freeze--thaw cycles,
assumes that K and u are uniform. More com-
and bioturbation (all of which require, or are en-
plex models are also available in which K and
hanced by, ¬ne-grained sediments). The reorga-
u vary with height to better represent transport
nization of clasts, in turn, helps protect the un-
processes close to the ground (e.g. Huang, 1999).
derlying ¬ne-grained layer from erosion. In this
Deposition from a Gaussian plume is modeled
way, the desert pavement and underlying eolian
by treating the ground as a sink for particles.
deposit coevolve. This coevolution is thought to
Deposition in this model is characterized by a
take several thousand years to initiate, but may
deposition velocity, p, de¬ned as the fraction of
occur faster in areas of higher dust deposition
the particle concentration just above the ground
rates (Pelletier et al., 2008a).
that undergoes deposition per unit time. In this
The texture of the alluvial-fan parent mate-
model framework, deposition at the ground sur-
rial plays an important role in determining the
face is equal to the downward ¬‚ux due to turbu-
speci¬c mode of pavement formation. In gravel-
lent diffusion and particle settling. This balance
rich parent material (i.e. those found near the
provides a ¬‚ux boundary condition at z = 0:
proximal part of an alluvial fan close to the
+ qc(x, y, 0) = pc(x, y, 0) mountain front), clast motion on the surface
K (1.18)
is unlikely to be signi¬cant until the underly-

Physically, the deposition velocity represents the ing eolian layer is suf¬ciently thick to cause
trapping ability of the surface, independent of expansion/contraction of the near-surface layer


z depositional

wind x

2 km
0.1 0.2 0.4 1.0

wind direction

numerical model results
u = 5 m/s, K = 5 m2/s, p = 0.05 m/s

wind direction
through wetting--drying and freeze--thaw cycles.
Fig 1.18 (a) Schematic diagram of model geometry.
In cases of sand-dominated parent material (e.g.
Depositional topography shown in this example is an inclined
the distal portions of many alluvial fans), the ini-
plane located downwind of source (but model can accept any
tial surface has relatively few clasts with which
downwind topography). (b) Grayscale maps of 3D model
to form a pavement. In these cases, pavement
results illustrating the role of variable downwind topography.
In each case, model parameters are u = 5 m/s, K = 5 m2 /s, clasts must ¬rst be pushed to the surface by
and p = 0.05 m/s. Downwind topography is, from left to right, freeze--thaw or wetting--drying cycles, or possibly
a ¬‚at plane, an inclined plane, and a triangular ridge. Modi¬ed made available through progressive fracturing of
from Pelletier and Cook (2005).
larger clasts (McFadden et al., 2005). As the surface
gains clast-material coverage, lateral migration
serves to interlock and suture the clasts as in the


eolian deposits

sand poor

parent material

Fig 1.19 Layered stratigraphy of typical alluvial fan terrace
(Eagle Mountain piedmont), illustrating stony monolayer
comprising the pavement, underlain by silt-rich eolian
deposits sitting atop parent alluvium.

gravel-dominated parent material sand-dominated parent material
bar bar


clast migration
Fig 1.21 (a) Virtual oblique aerial photograph of the margin
of Roach Playa and the adjacent Whiskey Pete™s alluvial fan
along the southern Nevada/California border. (b) Schematic
diagram showing how signi¬cant along-strike relief prevents
continuous saltation in the proximal portion of the fan,
resulting in local storage and accumulation of sand. On the
distal portion of the fan, conversely, continuous saltation
Fig 1.20 Schematic illustration of two modes of enables an eolian corridor to form.
parent-material formation associated with gravel-dominated
and sand-dominated parent material.
Playa is distinctly different from that of silt-
dominated playas such as Franklin Lake Playa. In
gravel-dominated case. These two distinct modes particular, the topographic relief of the alluvial
of pavement formation are illustrated schemati- fan plays a signi¬cant role in controlling the spa-
cally in Figure 1.20. tial pattern of deposition.
The distal portion of the Whiskey Pete™s allu-
1.2.2 Sand-dominated eolian systems vial fan appears lighter in color than the prox-
Many playas are dominated by sand rather than imal portion of the fan. In the ¬eld, this dif-
silt or clay. Roach Playa, located along the south- ference in surface re¬‚ectivity corresponds to an
ern Nevada--California border, is one example of abrupt increase in the sand content of soils be-
a sand-dominated playa. Sand from Roach Playa low a threshold elevation on the fan. Above this
is transported off the playa and onto the nearby elevation, sand sourced from the playa and from
Whiskey Pete™s alluvial fan (Figure 1.21). Because channels draining the mountain accumulates lo-
transport takes place by saltation rather than cally and is not transported across the fan. Be-
large-scale atmospheric mixing, the spatial pat- low this elevation, both coarse and ¬ne eolian
tern of sand deposition downwind from Roach sand from nearby playa and channel sources is

readily transported across the fan. These distal-
fan eolian corridors are controlled by a thresh-
old fan-surface relief (Cook and Pelletier, 2007).
(mm) ripples dunes
When along-strike relief falls below a threshold mega
value, an eolian transportational surface devel- dunes
ops. When the along-strike relief rises above the
threshold value, sand is trapped locally in low 102
10’1 101 103
10’2 104
l (m)
spots and a continuous surface of transporta-
tion is prevented from developing. Along-strike
Fig 1.22 Relationship between coarsest fraction (20th
relief promotes the storage of windblown sand in
percentile) of sand grains and bedform wavelength, », based
two ways. First, topographic obstacles create aero-
on measurements of ripples, dunes, and megadunes in the
dynamic recirculation zones on their lee sides.
Sahara Desert (after Wilson, 1972).
These recirculation zones are characterized by
very low bed shear stresses. Sand that is deposited
behind those obstacles may be stored inde¬nitely.
Second, the presence of topographic obstacles in-
creases the force necessary to move the sand. It
is not only suf¬cient for sand particles to be en-
trained from the surface, they must be picked up
with suf¬cient force to be transported over large
steps downwind. The transport of sand over a
complex surface, therefore, involves topographic
controls on both the shear stress exerted on the
particle and on the value of threshold shear
(b) top view side view
stress necessary to move the particle past down- VP
wind topographic obstacles.
Eolian bedforms can be classi¬ed into three
basic types depending on their spatial scale and wind
position within the bedform hierarchy: ripples,
dunes, and megadunes (Wilson, 1972). Ripples
are the smallest of the three bedforms, and are
typically spaced by 0.1--1 m and have heights of
1--10 cm with higher, more widely spaced rip-
ples forming in areas with stronger winds and/or
coarser sand. Dunes are the next level in the Fig 1.23 (a) Barchan dunes of the Salton Sea area,
California. (b) Schematic diagram illustrating how a barchan
bedform hierarchy and form only when ripples
gets its shape. The migration velocity of a dune is proportional
are present. Dunes commonly have spacings of
to the ratio of the perimeter to the area of the cross section
10--100 m and heights of 1--10 m. Megadunes form
perimeter. Therefore, the central core of the dune will move
when dunes are present, and may attain spac-
slower than the sides, causing arms to form over time (c).
ings of several kilometers and heights of sev-
eral hundred meters. All three bedform types ex-
hibit a correlation between spacing and grain trols the air ¬‚ow above it, and the pattern of ero-
size in the Sahara Desert (Wilson, 1972) (Figure sion and deposition and subsequent modi¬cation
1.22). This correlation is not observed everywhere, of the bedform size and shape through time.
however, partly because geographic variations in Eolian dunes come in a variety of types de-
wind speed also control bedform spacing. The pending primarily on sand supply and wind-
precise mechanisms responsible for bedform gen- direction variability. Barchan dunes (Figure 1.23)
esis are still a subject of active research. All three provide a nice example of both of these controls.
bedform types, however, likely involve a positive Barchan dunes have a central core ¬‚anked by two
feedback between the bedform shape, which con- arms oriented downwind. Barchan dunes form

in a steady wind direction when the sand sup- of deposition also depends on whether the sand
ply is relatively low (leaving some areas of bare had been transported into the lee side of an incip-
ground exposed). Dune formation under such ient dune. In the model, lee sides are mapped by
conditions requires that sand coalesce and ac- de¬ning a shadow zone (i.e. all areas that were lo-
cated in shadow given a sun angle of 15—¦ oriented
cumulate in patches separated by bare ground.
How does this ˜˜coalescence™™ happen? The reason parallel to the wind direction) with a high proba-
why sand tends to collect in patches rather than bility of deposition. Finally, sand deposited on the
spread out uniformly on the valley ¬‚oor has to bed rolls down the direction of steepest descent if
do with the higher coef¬cient of restitution of deposition causes the surface to be steeper than
soft sandy surfaces relative to hard bare ground. the angle of repose.
Sand in saltation will tend to bounce off of bare Werner™s model is capable of reproducing the
ground and land more softly on patches of sand. four principal dune types (transverse, barchan,
As a result, sand saltating across a valley ¬‚oor has star, and longitudinal) by varying the sand supply
a higher probability of deposition in sand-rich ar- and wind direction variability (e.g. Figure 1.24b).
eas than sand-free areas. This success is remarkable considering that his
Next consider how different portions of an in- model does not include any details on the micro-
cipient sand dune will migrate downwind. Fig- scopic physics of grain-to-grain interactions that
ure 1.23b illustrates an idealized incipient dune had long been assumed to be essential for under-
with a semi-circular cross-sectional shape. Dunes standing the formation of eolian bedforms. Al-
migrate by moving sand along their surfaces, and though Werner™s model does not successfully re-
the rate of sand movement is proportional to the produce all aspects of dunes (e.g. topographic pro-
perimeter, P , of the dune cross section. In order ¬les), his model includes the essential feedbacks
to move the dune a given distance downwind, of dune formation necessary for understanding
however, all of the sand in the cross sectional how different types of dunes are formed. First,
area, A, has to be moved. The rate of dune mi- Werner™s model illustrates how sand blown from
gration, therefore, is proportional to the ratio of a river bed tends to coalesce into discrete mounds
the perimeter to area. In the center of an in- rather than spreading out. Because the probabil-
cipient dune, the ratio of perimeter to area is ity of sand deposition is higher on an area al-
lower than it is on the sides of the dune. As ready covered by sand, a steady supply of sand
such, the sides will migrate faster, causing the particles saltating across an area will be prefer-
development of ˜˜arms™™ oriented downwind (Fig- entially deposited on areas already covered by
ure 1.23c). The wind direction must be steady for sand. In this way, discrete patches of sand will
barchan dunes to form because a slight change form and increase their heights relative to the
in wind direction is suf¬cient for dunes of this surrounding topography. Second, as the heights
type to become unstable. A shift in wind direc- of incipient bedforms increase, the length of
tion can cause sand from one arm of the barchan their shadow zones (which mimics the recircula-
to be transferred to the other arm, lengthening tion zone downwind of the dune crest) will also
one arm at the expense of the other over time. lengthen. This lengthening allows dunes to grow
Eventually, a longitudinal dune is formed with a higher since their heights are limited only by the
crestline oriented parallel to the wind. angle of repose and the lengths of their shadow
Werner (1995) developed a numerical model zones. Werner™s model is further discussed in
for the formation of eolian dunes using sim- Chapter 7.
ple geometric rules. Discrete units of sand in
Werner™s model are picked up from a sand bed
at random and transported a characteristic dis-
1.3 A tour of the glacial system
tance downwind. Sand units are deposited back
down on the bed with a probability that is rela-
tively low on bedrock surfaces compared to sandy Today, ice sheets and glaciers are restricted to
surfaces (re¬‚ecting the higher coef¬cient of resti- a relatively small portion of the Earth™s surface.

<< . .

. 4
( : 51)

. . >>