Aeolian Research 8 (2013) 65–73
Contents lists available at SciVerse ScienceDirect
Aeolian Research
journal homepage: www.elsevier.com/locate/aeolia
Spatial patterns of aeolian sediment deposition in vegetation canopies:
Observations from wind tunnel experiments using colored sand
Katrin Suter-Burri a,b,⇑, Christof Gromke a,c, Katherine C. Leonard a,d, Frank Graf a
a
WSL Institute for Snow and Avalanche Research SLF, CH-7260 Davos Dorf, Switzerland
Institute for Integrative Biology IBZ, ETH Zurich, CH-8092 Zurich, Switzerland
c
Building Physics and Services, Eindhoven University of Technology, 5600 MB Eindhoven, The Netherlands
d
Cooperative Institute for Research in Environmental Sciences, University of Colorado at Boulder, 216 UCB, Boulder, CO 80309, USA
b
a r t i c l e
i n f o
Article history:
Received 28 July 2012
Revised 4 November 2012
Accepted 15 November 2012
Available online 12 December 2012
Keywords:
Wind tunnel
Vegetation
Colored sand
Erosion
a b s t r a c t
In environments affected by wind erosion, plants act as traps for aeolian sediment, which leads to a
small-scale mosaic of depositional and erosional sediment transport regimes. This wind tunnel study
used colored sand to visualize spatial patterns of sediment redistribution within grass canopies. Wind
tunnel experiments were performed with high-, medium- and low-density canopies of Lolium perenne,
corresponding to vegetation covers of 47%, 16% and 4%, respectively. In the low and medium-density canopies, the wake areas downstream of the tussocks were the primary locations of sediment deposition. In
the medium-density canopy, these wedge-shaped wake deposits overlapped with the adjacent downstream tussocks, while in the low-density canopy they did not, indicating that these vegetation densities
respectively represented wake-interference and isolated roughness ﬂow. In the high-density canopy, very
few sand grains were entrained by the wind, and were mostly deposited within the disturbed zones surrounding the tussocks. The deposited grains were evenly distributed around the tussocks in the highdensity canopy without pronounced accumulations on their upstream, downstream or lateral sides.
We interpret the high-density canopy as a skimming ﬂow aerodynamic regime. The fraction of the sand
surface which was exposed to erosion was substantially smaller than the area not covered by grasses. It
accounted for 67–78% of the non-covered surface in the low-density canopy, and for 44–77% of the surface in the medium-density canopy. This ﬁnding indicates that wind erosion models overestimate the
sediment source area if they assume the erodible area is the entire exposed surface not covered by roughness elements.
Ó 2012 Elsevier B.V. All rights reserved.
1. Introduction
Vegetation affects soil erosion by the wind at multiple spatial
and temporal scales. The above-ground parts of vegetation act as
non-erodible roughness elements that modify the near-surface
wind, while their roots help stabilize the soil. The effect of vegetation on aeolian sediment transport is attributed to three mechanisms (Wolfe and Nickling, 1993): (i) surface sheltering, (ii)
momentum extraction from the wind, and (iii) trapping of windborne sediment.
Surface sheltering is provided both by covering a portion of the
ground surface and limiting the area of bare ground available to be
eroded and by the wakes of reduced mean wind velocity downﬂow
from a plant. Morris (1955) deﬁned three ﬂow regimes depending
on the wake development and the proportion of protected to total
⇑ Corresponding author at: WSL Institute for Snow and Avalanche Research SLF,
CH-7260 Davos Dorf, Switzerland. Tel.: +41795747614.
E-mail addresses: [email protected] (K. Suter-Burri), [email protected] (C.
Gromke), [email protected] (K.C. Leonard), [email protected] (F. Graf).
1875-9637/$ - see front matter Ó 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.aeolia.2012.11.002
surface area: (i) isolated roughness ﬂow, where there is no interaction between wakes and adjacent downstream roughness elements, (ii) wake interference ﬂow, where wakes from upstream
elements intercept downstream elements, and (iii) skimming ﬂow,
where wakes completely overlap and the entire ground surface is
sheltered (Fig. 1). The ﬂow regime depends on both the size and
spacing of roughness elements and on the free-stream or mean
wind speed, and is a critical factor in determining the amount of
wind erosion a surface will experience (Morris, 1955; Lee and
Soliman, 1977; Wolfe and Nickling, 1993; Breshears et al., 2009).
Momentum extraction from the wind by vegetation reduces the
erosive force of the wind. By absorbing a part of the total shear
stress of the wind, plants decrease the shear stress acting on the
ground surface.
Trapping of sediment by vegetation occurs through two mechanisms. Windborne particles can be removed from the air stream
when they hit plant surfaces, or when the wind velocity becomes
too low in the immediate vicinity of plants. By these means, plants
cause particles to settle from the ﬂow, resulting in characteristic
66
K. Suter-Burri et al. / Aeolian Research 8 (2013) 65–73
Fig. 1. Schematic representation of the three ﬂow regimes deﬁned by Morris (1955). Figure modiﬁed after Oke (1988) and Gromke (2008).
deposition patterns within vegetation canopies. A landscape experiencing wind erosion is thus a small-scale mosaic of depositional
and erosional areas (Okin et al., 2006). These sediment redistribution processes have important implications for the development of
a landscape’s surface morphology and for soil resource availability.
Wind erosion also depletes nutrients from plant interspaces and
accumulates them within and around plants, a process known as
the resource island phenomenon (Bielders et al., 2002; Li et al.,
2007).
Several methods have been used to study small-scale spatial
patterns of aeolian sediment deposition. Udo and Takewaka
(2007) used a laser displacement sensor to measure wind-induced
bed elevation changes in ﬂexible artiﬁcial plant canopies in a wind
tunnel setting. They found that canopies of low height, high density and high ﬂexibility promoted bed accumulation and reduced
sand transport downstream of the canopy. In several ﬁeld studies,
small-scale erosion and deposition processes have been monitored
by marking the changing level of the soil surface on steel rods driven into the ground (Wiggs et al., 1995; Hesse and Simpson, 2006;
Li et al., 2008). Other ﬁeld studies have measured windborne sediment mass ﬂuxes in heterogeneous plant canopies, indirectly providing information about deposition and erosion processes at the
bed (Gillette and Pitchford, 2004; Gillette et al., 2006). Another approach to investigate aeolian sediment transport is the use of ﬂuorescent tracer sand, but it has only been used on unvegetated
surfaces to date (Berg, 1983; Willetts and Rice, 1988; Cabrera
and Alonso, 2009).
More detailed information on deposition processes and wake
development comes from wind tunnel experiments with model
canopies of solid roughness elements (Sutton and McKenna
Neuman, 2008a,b; Gillies et al., 2006). Various methods have been
used to visualize bed-level ﬂow and erosion patterns around solid
roughness elements in wind tunnel experiments. Flow visualization techniques include ink and oil methods (Langston and Boyle,
1982; Bullard et al., 2000; Sutton and McKenna Neuman, 2008b)
and a method based on the ammonia–manganese chloride reaction
(Donat and Ruck, 1999). Sutton and McKenna Neuman (2008a)
examined sediment transport initiation downwind of solid roughness elements by placing strips of sand across the wind tunnel ﬂoor
and observing where they had eroded after exposing them to a
wind event.
However, results from experiments with solid roughness elements do not necessarily apply to live plant canopies (Walter
et al., 2012). Since live plants are porous and ﬂexible, they affect
the wind ﬂow and sediment ﬂux differently than solid, rigid objects. Plants allow the wind and sediment particles to pass through
them, and they have larger drag coefﬁcients than solid objects, i.e.
they can produce greater overall ﬂow resistance because they have
a larger surface subject to skin friction (Gillies et al., 2002; Gromke
and Ruck, 2008).
The study presented here differs from previous wind tunnel research in two respects. First, we used live vegetation instead of
artiﬁcial roughness elements. Second, we used colored sand to
visualize spatial patterns of sediment deposition within those vegetation canopies. The objective of this study was to identify and locate areas experiencing net deposition in three canopy densities of
Perennial Ryegrass (Lolium perenne). Following a visual determination of net deposition patterns in different ﬂow regimes, we estimate the fractional areas of sheltered and erodible ground
surface within these grass canopies.
K. Suter-Burri et al. / Aeolian Research 8 (2013) 65–73
67
allowed ﬁrm rooting of the plants and provided a high waterretention capacity. To create a readily erodible surface, a 1 cm thick
layer of quartz sand was spread over the planted soil. The grain size
distribution of this sand is presented in Table 2. The third and ﬁfth
tray downstream were ﬁlled with red and blue resin-coated sand
respectively, which was produced from the uncolored sand (Carlo
Bernasconi AG, Switzerland, Fig. 2). Effects of the resin-coating
on the erodibility of the sand grains, e.g. possible impacts on its
grain size distribution or inter-particle electrostatic forces (Rasmussen et al., 2009), were not considered in this study.
The frontal area index k was determined based on scaled frontal
view images of individual plants. It was calculated as k = na/A, with
n being the number of tussocks with a mean frontal area a occupying the ground area A (Table 1). At the time of the wind tunnel
experiments, the tussocks had a mean height h of 10 cm (sd = 1,
n = 20) and a mean frontal area a of 64 cm2 (sd = 18, n = 20). The
time between experiments was at least 24 h, which allowed the
plants to return to an upright position. During the 4 weeks of
experimental work, the plants did not grow observably and they
showed no visual evidence of sandblast injury. Vegetation cover
Cv was deﬁned as the proportion of the ground surface occupied
by a vertical projection of the plants’ aerial parts (Greig-Smith,
1964 in Collins and Becker, 2001). It was determined based on
scaled nadir images of whole plant trays (Fig. 3). Images were digitized and analyzed with the software ImageJ (Version 1.39u,
http://rsb.info.nih.gov/ij/).
2. Material and methods
2.1. Wind tunnel facility
The experiments were performed in the boundary layer wind
tunnel of the WSL Institute for Snow and Avalanche Research SLF
in Davos, Switzerland, located at 1650 m a.s.l. (Clifton et al.,
2006). The wind tunnel operates in suction mode, with a fan downstream of the test section drawing air from outside through a honeycomb. It has a nominal cross-section of 1 m 1 m and a total
length of 15 m. For the experiments described here, that length included a 2 m contraction section (4:1), a 5 m inlet duct with a
smooth wooden ﬂoor, and an 8 m test section with an adjustable
ﬂoor.
For each experiment, eight wooden trays (98 99 cm) were installed in the test section of the wind tunnel, planted with a prescribed number of grass tussocks per unit ground surface (Fig. 2).
The test section ﬂoor was lowered so that the surface matched
the upwind inlet section, and the height of the wind tunnel roof
was adjusted for each canopy density to provide a zero streamwise
pressure gradient.
2.2. Vegetation
Wind tunnel experiments were performed with three canopy
densities of Perennial Ryegrass (Lolium perenne cv. Alligator, Alpine
Garden Center Schutz Filisur, Switzerland), namely 5.25, 24.5 and
91 tussocks per square meter (Table 1). In the following discussion
they are referred to as the low-, medium- and high-density canopies. The arrangement of the plants followed the same pattern of
staggered rows in all experiments (Fig. 2).
Lolium perenne was grown in a commercial horticulture nursery
substrate (Topferde 140, Ricoter, Switzerland) in plant cultivation
trays with integrated conical cells (20 ml). After seedlings reached
a height of approximately 6 cm, they were removed from the cultivation trays with their entire root balls and planted in the eight
wooden trays used for the wind tunnel experiments. The trays
were ﬁlled with a 3:1 mixture of crushed limestone sand (grain
sizes 0–3 mm, pH 8.4) and soil (grain sizes 0–2 mm, pH 7.5), which
2.3. Wind tunnel experiments
The wind tunnel experiments discussed here were conducted as
a component of a larger study of sediment mass ﬂuxes in vegetation canopies described in more detail by Burri et al. (2011). The
ﬁrst experiment had a vegetation density of 91 plants per square
meter. Subsequently, plants were removed from the trays to conduct experiments with area-averaged densities of 24.5 and 5.25
plants per square meter. After each experiment, the 1 cm quartz
sand layer was renewed to provide consistent surface conditions,
i.e. the eroded sand was replaced by new sand and the sand bed
was ﬂattened.
wind tunnel side wall
c
wind
a
b
c
wind tunnel side wall
8m
Fig. 2. Wind tunnel test section with eight aligned plant trays (98 99 cm), the third and ﬁfth tray downstream being ﬁlled with red and blue quartz sand, respectively. Lowdensity canopy (top left), medium-density canopy (top middle), high-density canopy (top right), and schematic representation (bottom). a = upstream part of the plant tray,
b = downstream part of the plant tray, c = lateral parts of the plant tray close to the side walls of the wind tunnel. (For interpretation of the references to color in this ﬁgure
legend, the reader is referred to the web version of this article.)
68
K. Suter-Burri et al. / Aeolian Research 8 (2013) 65–73
Table 1
Characteristics of the three canopy densities and experimental conditions. The air
temperature [°C], relative air humidity [%] and matric potential of the sand [MPa]
were determined as described in Burri et al. (2011).
Planting density [plants/m2]
Vegetation cover Cv [%]
Frontal area index k
Inter-plant spacing d [cm]
Mean canopy height h [cm]
Mean free stream velocity [m/s]
Duration of erosive event [s]
Air temperature [°C]
Relative air humidity [%]
Matric potential of the sand [MPa]
Lowdensity
Mediumdensity
Highdensity
5.25
4
0.03
56
10
15.4
174
7.3
81.3
10.8
24.5
16
0.16
28
10
15.7
610
7.4
68.3
21.4
91
47
0.58
14
10
15.6
7224
9.9
54.5
25.1
Table 2
Grain size distribution of the quartz sand used in this study.
Grain size class [mm]
Percentage [%]
>0.80
0.63–0.80
0.40–0.63
<0.40
4
54
40
2
Each canopy was subjected to an erosive wind event consisting
of three phases. The wind tunnel motor was programmed to accelerate within 120 s to a free stream wind velocity of 15.5 m/s, this
velocity was maintained for a set time and then the motor decelerated to 0 m/s within 100 s. Depending on the intensity of erosion in
a given experiment, the duration of constant wind velocity was
varied to yield a sufﬁcient amount of erosion for recognizing patterns in the colored sand while not depleting the loose quartz sand
on the beds.
2.4. Analysis of deposition patterns
The analysis of deposition patterns described here is restricted
to the trays containing red and blue sand, subsequently referred
to as the red and blue sections (Fig. 2). Deposition in these trays
was more obvious than in the six white trays because all the sand
invading from upstream was of a different color than the sand with
which the trays were originally ﬁlled. In contrast, the areas of sand
deposition in the trays with white sand were less pronounced because the deposits contained both colored and white grains. There
were 2 m of white sand upstream of the red section (Fig. 2). The
source area upstream of the blue section included 3 m of white
sand and 1 m of red sand (Fig. 2). Thus, all white sand grains that
were found in the red section and all white and red sand grains
in the blue section after the wind tunnel experiments represented
deposits originating from upstream. To describe the locations of
sediment deposits, we divided the sand bed into the following
zones, according to their position related to a given tussock
(Fig. 3): (a) the bleed ﬂow zone, referring to the space within reach
of the moving grass blades, projected on the ground, (b) the lateral
zones, (c) the upstream area, and (d) the wake area. The elongated
bare soil areas oriented in the direction of the wind are referred to
as ‘streets’ (Gillette and Pitchford, 2004).
Image sequences were taken through the perspex roof of the
wind tunnel to evaluate the temporal development of deposition
patterns during the experiments. In the low- and medium-density
canopies, sediment ﬂux was high and the deposited sediments
were clearly visible. In the high-density canopy case, net sediment
ﬂux was low and it was difﬁcult to detect the deposited sand grains
in the images. In light of this, two different strategies were applied
to analyze the deposition patterns in the different canopy densities. In the low- and medium-density cases a set of three consecutive images were selected to represent an initial, intermediate and
ﬁnal stage of each experiment (Figs. 4 and 5). The time intervals
between the images were not recorded and put in relation to the
duration of the experiment. Therefore, the image sequences only
allow qualitative analysis of the temporal development of deposition patterns. The deposition patterns were intensiﬁed by increasing contrast, saturation and brightness of these images with Adobe
Photoshop CS4. In the high-density canopy, the analysis of deposition patterns in these image sequences was supplemented with
close-up images that were taken after the experiments (Fig. 6).
The individual deposited grains were automatically selected based
on their color, marked in black and enlarged by using Adobe Photoshop CS4. The automatic selection was then checked manually to
ensure that the marked areas corresponded to deposited grains.
In addition to describing the location of sand deposition relative
to the plants and their ﬂow characteristics, we estimated the fractions of the ground surface that were sheltered and potentially exposed to wind erosion in each experiment. We deﬁned the
sheltered portion (SP) as the fraction of the total surface that was
Fig. 3. Schematic ﬂow areas around a tussock: (a) bleed ﬂow zone, (b) lateral zones, (c) upstream area, (d) wake area. The elongated bare soil areas that are oriented in the
direction of the wind are referred to as ‘streets’ (dark shaded areas).
K. Suter-Burri et al. / Aeolian Research 8 (2013) 65–73
69
Fig. 4. Image sequences of sediment deposition patterns in the low-density canopy (inter-plant spacing = 56 cm). In the red section (A1–A3), white sand was invading from
upstream. The dark spots at the top of image A3 are areas where erosion exposed the substrate underneath the red sand. In the blue section (B1–B3), red and white sand was
invading from upstream. Wind direction: from left to right. (For interpretation of the references to color in this ﬁgure legend, the reader is referred to the web version of this
article.)
either covered by plants or experienced net deposition during the
experiment. The remaining ground surface, which was potentially
exposed to erosion, was deﬁned as the erodible portion EPtot. EPtot
is deﬁned here relative to the total surface area, while EPnc, is deﬁned in relation to the non-covered surface, i.e. the total surface
area minus the area covered by grasses. The determination of SP
and EPtot was made by visually discriminating between the colors
of deposited sand and the sand with which the trays were ﬁlled
originally. These discriminations were made while adhering to
the following criteria and assumptions. Areas completely covered
by upstream colored sand were attributed to SP, as they clearly
experienced net deposition. Regions containing ripple patterns of
alternating colors were attributed to EPtot, even if a substantial portion was covered by upstream colored sand. Sand in these ripples
was assumed not to represent net deposition, as ripples migrate
and are thus only temporary deposition sites (Fig. 7). Areas completely covered by sand with which the trays were ﬁlled originally
were also attributed to EPtot, because these areas did not experi-
Fig. 5. Image sequences of sediment deposition patterns in the medium-density canopy (inter-plant spacing = 28 cm). In the red section (A1–A3), white sand was invading
from upstream. In the blue section (B1–B3), red and white sand was invading from upstream. Wind direction: from left to right. (For interpretation of the references to color
in this ﬁgure legend, the reader is referred to the web version of this article.)
70
K. Suter-Burri et al. / Aeolian Research 8 (2013) 65–73
Fig. 6. Sand grains deposited in the red and blue section of the high-density canopy (inter-plant spacing = 14 cm). The grains were automatically selected based on their color,
marked in black and enlarged by using Adobe Photoshop CS4. Wind direction: from left to right. (For interpretation of the references to color in this ﬁgure legend, the reader is
referred to the web version of this article.)
Fig. 7. Visually determined proportion of sheltered surface SP in the red and blue section of the low-density canopy (top, inter-plant spacing = 56 cm) and the mediumdensity canopy (bottom, inter-plant spacing = 28 cm). Wind direction: from left to right. (For interpretation of the references to color in this ﬁgure legend, the reader is
referred to the web version of this article.)
ence net deposition. However, areas with unchanged color did not
provide conclusive evidence for or against the following sediment
transport scenarios: (i) neither erosion nor deposition, (ii) deposited grains originated in the same tray, (iii) more erosion than
deposition (net erosion). The sheltered ground portions SP were
manually selected and marked in the images representing the ﬁnal
stages of the experiments using Adobe Photoshop CS4 (Fig. 7).
3. Results
3.1. High-density canopy
In the high-density canopy, the bleed ﬂow zones of neighboring
tussocks overlapped, covering almost the whole surface. Because
net sediment ﬂux was low, only isolated deposited sand grains
K. Suter-Burri et al. / Aeolian Research 8 (2013) 65–73
were found (Fig. 6). Such grains were distributed throughout the
canopy, with particularly high concentrations in the bleed ﬂow
zones and lower concentrations in the plant interspaces. Within
the bleed ﬂow zones, the deposited grains were evenly distributed
around the tussocks without pronounced accumulations on the upstream, downstream or lateral sides of the tussocks.
3.2. Medium-density canopy
In the medium-density canopy, deposition occurred primarily
in the bleed ﬂow zones and wake areas (Figs. 3 and 5). In some
parts of the canopy, deposition was also observed in the lateral
zones. In general, the sand deposits in the wake areas were
wedge-shaped in plan view and overlapped with the adjacent
downstream tussocks.
In the initial stage of the experiment (Fig. 5A1 and B1), deposition in the wake areas of the plants occurred primarily at the upstream ends of the red and blue sections, with less intense
deposition occurring in the downstream areas of those trays (Figs. 2
and 5). In the intermediate stage (Fig. 5A2 and B2) deposition became more intense over the whole sections and the deposition
areas in the wakes of the tussocks expanded laterally. During this
intermediate stage of the experiment, rudimentary ripples evolved
in the streets between the tussocks. In the ﬁnal stage of the experiment (Fig. 5A3 and B3) the deposition areas increased in width,
and expanded into the open streets between the tussocks in the
central areas of the trays. Along the outside edges of the plant
trays, i.e. between the wind tunnel side walls and the second plant
rows from the edges, there was less deposition than in the center of
the trays during all three experiment stages (Figs. 2 and 5).
The deposition areas were generally larger in the blue section
than in the red section (e.g. Fig. 5A3 and B3). The sheltered ground
surface SP equaled 33% in the red section and 62% in the blue section (Fig. 7 and Table 3). The erodible portion EPnc accounted for
77% of the surface not covered by grasses in the red section and
44% in the blue section (Table 3).
3.3. Low-density canopy
In the low-density canopy, discrete deposition patterns could be
observed in all four aerodynamic zones around the tussocks: (a)
the bleed ﬂow zone, (b) the lateral zones, (c) the upstream area,
and (d) the wake area (Figs. 3 and 4). In contrast to the mediumdensity canopy, the wake deposits did not overlap with the adjacent downstream tussocks.
In the initial stages of the experiments (Fig. 4A1 and B1) deposition was found in all four zones mentioned above. The deposits in
the wake areas were wedge-shaped in plan view tapering to indeterminate downwind ends during the initial stage. In the intermediate and ﬁnal stages of the experiments (Fig. 4A2–A3 and B2–B3)
the wake deposits expanded laterally and a clearly discernible ripple pattern evolved in the streets between the tussocks. The
Table 3
Portions of sheltered surface SP, and erodible surface in relation to the total surface
EPtot and in relation to the surface which was not covered by grasses EPnc.
Low-density canopy
Medium-density canopy
Red section Blue section Red section Blue section
Grass-covered area [%]
Not-covered portion [%]
Sheltered portion SP [%]
Erodible portion EPtot [%]
Erodible portion EPnc [%]
3
97
24
76
78
5
95
36
64
67
13
87
33
67
77
13
87
62
38
44
71
downstream ends of the deposits in the wakes of the tussocks became more pronounced due to the development of ripples approximately 10 cm upwind of the adjacent downstream tussocks.
Analogous to the medium-density canopy case, there was more
deposition in the blue than in the red tray, and more deposition in
the center than in the areas close to the side walls of the wind tunnel in all three stages (Figs. 2 and 4). The sheltered ground surface
SP amounted to 24% of the red section and 36% of the blue section
(Fig. 7 and Table 3). The erodible area EPnc totaled 78% of the noncovered red section and 67% of the blue section.
4. Discussion
The spatial patterns of colored sand found in this study demonstrate that multiple processes control aeolian sediment entrainment and deposition on a vegetated surface. One important
process was sediment deposition in the sheltered wake areas of
the tussocks. In the low- and medium-density canopies, such wake
deposits were the primary regions of sediment deposition (Figs. 4
and 5). The size of these wake deposits in relation to the plant
spacing indicates that the two canopy densities represented an isolated roughness ﬂow and a wake-interference ﬂow regime, respectively. In the low-density canopy, the wake deposits did not
intercept the adjacent downwind tussocks, while in the mediumdensity canopy they overlapped with the downstream vegetation.
In the high-density canopy, the lack of wake deposits and the even
distribution of deposited sand grains within the bleed ﬂow zones
(Fig. 6) suggests a skimming ﬂow regime in which the whole surface was sheltered, with no individual wake areas.
In the immediate vicinity of the grass tussocks, i.e. within reach
of the moving grass blades, windborne particles were deposited
when they hit plant surfaces, lost momentum and fell. This process
caused the sediment accumulations in the bleed ﬂow zones around
the tussocks. It was probably also responsible for deposits in the
areas upstream of the tussocks in the low-density canopy, visible
in the last pictures of the image sequences (Fig. 4A3 and B3). Some
of the windborne particles in the sand cloud approaching the tussocks may have rebounded from the grass blades and come to settle in front of the tussocks (Bagnold, 1941). In the high- and
medium-density canopies the wake areas overlapped with the
downstream tussocks so deposits upstream of a given tussock
could not be identiﬁed as up-wind deposits or deposits in the wake
of the upstream tussock (Figs. 5 and 6).
Walter et al. (2012) measured the bed level shear stress distribution around tussocks of L. perenne and solid cylinders. Compared
to a situation without a roughness element, they found increased
shear stress to the sides of both the grass tussock and the cylinder,
with a higher peak shear stress adjacent to the cylinder than the
grass tussock. This reﬂects the fact that ﬂow acceleration around
a solid obstacle is higher than around porous obstacles (Bowker
et al., 2008). Hence, more sediment entrainment is to be expected
around a solid, rigid obstacle than around a porous, ﬂexible obstacle. Leenders et al. (2007) observed in their ﬁeld study that sediment accumulated upstream and laterally of a single shrub,
whereas sediment scouring occurred around a single tree trunk.
In the present study we found sediment deposits immediately to
the sides of the tussocks, indicating that the shear stress acting
on the ground was below the threshold for sediment entrainment.
This is most evident in Fig. 4A3, where these lateral deposits are
clearly larger than the bleed ﬂow zones.
In the low-density canopy, the tussocks in the lateral parts of
the plant trays close to the side walls of the wind tunnel produced
asymmetric deposition without lateral deposition on their outer
sides (Figs. 2 and 4). This phenomenon is probably caused by enhanced turbulences close to the side walls of the wind tunnel
72
K. Suter-Burri et al. / Aeolian Research 8 (2013) 65–73
due to ﬂow recirculation and corner vortices. Further evidence for
enhanced erosion near the side walls of the wind tunnel was found
in the medium-density canopy, where erosion was particularly intense in the streets adjacent to the side walls, i.e. between the outermost and second outermost plant rows (Fig. 5). The inﬂuence of
the tunnel side-walls died out within two plant-spacings from the
edge, and was not observed in the central portion of the tunnel on
which the results presented here are based.
The areas of net deposition were generally smaller in the red
section than in the blue section. Since the red section was closer
to the upwind beginning of the test section, it had a smaller sediment fetch than the blue section. Aeolian sediment ﬂux tends to increase up to saturation with the distance over which the erosion
occurs (the ‘‘fetch effect’’, Gillette et al., 1996). Additionally, the
characteristic ﬂow regimes of the three canopy densities required
some distance to fully develop because the entering air encountered a change in surface roughness when it passed from the
smooth inlet duct of the wind tunnel to the planted test section.
Therefore, the near ground wind velocities were probably highest
and most efﬁcient in entraining sediment at the beginning of the
test section, leading to higher erosion rates near the beginning of
the test section than towards the downwind end. The changing
erosive capacity of the ﬂow along the test section offered an opportunity to study the development of deposition patterns under different erosion intensities. In the course of an erosive event, the
sand deposit in the wake of an obstacle gradually increases in areal
cover and height until its slopes stand at the limiting angle of repose (Bagnold, 1941). Once this angle is reached, any additional
sand slips down the slopes and leaves the sheltered wake area,
so it can be entrained by the stronger wind outside the wake. In
the blue section where the erosive capacity of the wind ﬂow was
lower than in the red section this process enabled the deposition
areas to spread into the open streets between the tussocks
(Figs. 4B3 and 5B3). In the red section, in contrast, there was only
minor deposition in the streets (Figs. 4A3 and 5A3). Accordingly,
the erodible surface areas EPtot were higher in the red section than
in the blue section (Table 3).
In spite of these differences between the red and blue sections,
EPtot was in any case substantially smaller than the ground surface
that was not covered by grasses (Table 3). Many wind erosion
models for vegetated surfaces, however, approximate the erodible
area as the total surface area not covered by non-erodible elements
(e.g. Marticorena and Bergametti, 1995). The present study supports the ﬁndings of Chappell et al. (2010) who showed that this
approximation can lead to substantial over-estimates of aeolian
sediment ﬂuxes. Chappell et al. (2010) developed a model based
on angular reﬂectance and estimated the erodible portion for different arrays of solid hemispheres. For an array with a vegetation
cover Cv of 5% and a frontal area index k of 0.04, they estimated EPnc
to be 83% of the non-covered surface. The visually determined EPnc
in our low-density canopy, which had a similar Cv and k (Table 1),
was 78% of the non-covered surface in the red section and 67% in
the blue section (Table 3), demonstrating that the erodible surface
fraction in canopies of ﬂexible, porous grass tussocks is even smaller than in comparable arrays of solid roughness elements. For future studies it would be interesting to investigate in more detail
the performance of models that approximate the erodible area as
the uncovered area, e.g. by validating them with laboratory wind
tunnel measurements.
The present wind tunnel study was distinct from previous
research in that colored sand was used to examine erosion
and deposition processes in live plant canopies. Colored sand
proved to be useful for this purpose, as it produced high-resolution bed-level data for the whole ground surface under investigation. This is of particular importance when dealing with live
plants, as they are non-uniform complex structures that interact
with the wind in more variable ways than solid artiﬁcial roughness elements.
The application of colored sand is a relatively easy and low-cost
method to study aeolian sediment transport in vegetation canopies, and it may therefore be useful not only for wind tunnel studies, but also for larger scale ﬁeld studies. The sensitivity of the
method, i.e. the detectability of colored sand deposits depends on
several factors, such as the scale of the images, the applied image
analysis procedure and the characteristics of the underlying surface. Previous ﬁeld studies have used ﬂuorescent tracer sand on
unvegetated surfaces and transferred deposits via sticky tapes to
the laboratory for analysis under ultraviolet light (Berg, 1983;
Cabrera and Alonso, 2009). The use of colored resin-coated sand,
in contrast, may be suitable to collect on-site data, by taking
high-resolution close-up images directly in the ﬁeld.
5. Conclusions
The present study describes wind tunnel experiments with live
grass tussocks. The spatial patterns of redistributed colored sand
within a high-, medium- and low-density canopy indicated three
distinct aerodynamic ﬂow regimes, i.e. a skimming ﬂow, a wakeinterference ﬂow and an isolated roughness ﬂow. Previous research on the factors determining which air ﬂow regime a canopy
will experience has been based on experiments with solid roughness elements (Morris, 1955; Lee and Soliman, 1977; Wolfe and
Nickling, 1993).
In the low and medium-density canopies, sediment deposits
were primarily found in the wake areas downstream of the
tussocks. While they overlapped with the adjacent downstream
tussocks in the medium-density canopy, they did not in the lowdensity canopy. In the high-density canopy, very few sand grains
were entrained by the wind, and the deposited grains were evenly
distributed within the disturbed zones surrounding the grass
tussocks.
We found that the fraction of the sand surface which effectively
experienced erosion (EPnc) was substantially smaller than the area
which was not covered by grasses in three vegetation canopy densities associated with different ﬂow regimes. The three experiments conducted in this study do not allow us to estimate the
proportionality factor between EPnc and vegetation density. Further studies are needed to establish this relationship under various
conditions. While the values of EPnc and the boundary conditions
for each ﬂow regime in this experimental study may not be exactly
replicated by ﬁeld studies under natural wind conditions, our general ﬁndings and trends should be representative.
Acknowledgments
This study and the main author were ﬁnancially supported by
the Velux Foundation. Additional funding was provided by the
Swiss National Science Foundation and the Vontobel-Foundation.
Sensirion AG kindly provided relative humidity sensors and technical support. The Alpine Garden Center Schutz Filisur shared their
knowledge with us and provided access to their plant nursery facilities. We wish to thank Dr. Costantino Manes for his help in planning and designing this study and Dr. Martin Schneebeli for his
assistance with the software ImageJ. We would also like to thank
Silvio Burger, Hans Herranhof, Andreas Moser, Christian Simeon
and Andreas Tröger for their hard work in the wind tunnel and
for manufacturing several experimental devices. We wish to
acknowledge in particular the constructive and very helpful suggestions of Prof. Ottmar Holdenrieder, Prof. Michael Lehning and
Benjamin Walter that led to several important reﬁnements of the
manuscript.
K. Suter-Burri et al. / Aeolian Research 8 (2013) 65–73
References
Bagnold, R.A., 1941. The Physics of Blown Sand and Desert Dunes. Methuen, London.
Berg, N., 1983. Field evaluation of some sand transport models. Earth Surface
Processes and Landforms 8, 101–114.
Bielders, C.L., Rajot, J.L., Amadou, M., 2002. Transport of soil and nutrients by wind
in bush fallow land and traditionally managed cultivated ﬁelds in the Sahel.
Geoderma 109, 19–39.
Bowker, G.E., Gillette, D.A., Bergametti, G., Marticorena, B., Heist, D.K., 2008. Finescale simulations of aeolian sediment dispersion in a small area in the northern
Chihuahuan Desert. Journal of Geophysical Research 113, F02S11. http://
dx.doi.org/10.1029/2007JF000748.
Breshears, D.D., Whicker, J.J., Zou, C.B., Field, J.P., Allen, C.D., 2009. A conceptual
framework for dryland Aeolian sediment transport along the grassland–forest
continuum: effect of woody plant canopy cover and disturbance.
Geomorphology 105, 28–38.
Bullard, J., Wiggs, G., Nash, D., 2000. Experimental study of wind directional
variability in the vicinity of a model valley. Geomorphology 35, 127–143.
Burri, K., Gromke, C., Lehning, M., Graf, F., 2011. Aeolian sediment transport over
vegetation canopies: a wind tunnel study with live plants. Aeolian Research 3
(2), 205–213.
Cabrera, L.L., Alonso, I., 2009. Correlation of aeolian sediment transport measured by
sand traps and ﬂuorescent tracers. Journal of Marine Systems 80, 235–242.
Chappell, A., Van Pelt, S., Zobeck, T., Dong, Z., 2010. Estimating aerodynamic
resistance of rough surfaces using angular reﬂectance. Remote Sensing of
Environment 114, 1462–1470.
Clifton, A., Rüedi, J.D., Lehning, M., 2006. Snow saltation threshold measurements in
a drifting-snow wind tunnel. Journal of Glaciology 179, 585–596.
Collins, W.B., Becker, E.F., 2001. Estimation of horizontal cover. Journal of Range
Management 54, 67–70.
Donat, J., Ruck, B., 1999. Simulated ground deposition of ﬁne airborne particles in an
array of idealized tree crowns. Boundary-Layer Meteorology 93, 469–492.
Gillette, D.A., Herbert, G., Stockton, P.H., Owen, P.R., 1996. Causes of the fetch effect
in wind erosion. Earth Surface Processes and Landforms 21, 641–659.
Gillette, D.A., Herrick, J.E., Herbert, G.A., 2006. Wind characteristics of mesquite
streets in the northern Chihuahuan Desert, New Mexico, USA. Environmental
Fluid Mechanics 6, 241–275.
Gillette, D.A., Pitchford, A.M., 2004. Sand ﬂux in northern Chihuahuan desert, New
Mexico, USA, and the inﬂuence of mesquite-dominated landscapes. Journal of
Geophysical Research 109, F04003. http://dx.doi.org/10.1029/2003JF000031.
Gillies, J.A., Nickling, W.G., King, J., 2006. Aeolian sediment transport through large
patches of roughness in the atmospheric inertial sublayer. Journal of
Geophysical Research 111, F02006. http://dx.doi.org/10.1029/2005JF000434.
Gillies, J.A., Nickling, W.G., King, J., 2002. Drag coefﬁcient and plant form response to
wind speed in three plant species: Burning Bush (Euonymus alatus), Colorado
Blue Spruce (Picea pungens glauca), and Fountain Grass (Pennisetum setaceum).
Journal of Geophysical Research 107, 4760. http://dx.doi.org/10.1029/
2001JD001259.
Gromke, C., 2008. Einﬂuss von Bäumen auf die Durchlüftung von innerstädtischen
Strassenschluchten. Ph.D. Thesis, University of Karlsruhe, KIT Scientiﬁc
Publishing, p. 142. ISBN: 978-3-86644-339-6.
73
Gromke, C., Ruck, B., 2008. Aerodynamic modelling of trees for small-scale wind
tunnel studies. Forestry 81 (3), 243–258.
Hesse, P.P., Simpson, R.L., 2006. Variable vegetation cover and episodic sand
movement on longitudinal desert and sand dunes. Geomorphology 81, 276–
291.
Langston, L., Boyle, M., 1982. A new surface-streamline ﬂow visualization
technique. Journal of Fluid Mechanics 125, 53–57.
Lee, B.E., Soliman, B.F., 1977. Investigation of forces on 3 dimensional bluff bodies in
rough wall turbulent boundary-layers. Journal of Fluids Engineering –
Transactions of the ASME 99, 503–510.
Leenders, J.K., Boxel, J.H., Sterk, G., 2007. The effect of single vegetation elements on
wind speed and sediment transport in the Sahelian zone of Burkina Faso. Earth
Surface Processes and Landforms 32, 1454–1474.
Li, J., Okin, G.S., Alvarez, L., Epstein, H., 2007. Quantitative effects of vegetation cover
on wind erosion and soil nutrient loss in a desert grassland of southern New
Mexico, USA. Biogeochemistry 85, 317–332.
Li, J., Okin, G.S., Alvarez, L., Epstein, H., 2008. Sediment deposition and soil nutrient
heterogeneity in two desert grassland ecosystems, southern New Mexico. Plant
and Soil. http://dx.doi.org/10.1007/s11104-008-9850-7.
Morris, H.M., 1955. Flow in rough conduits. Transactions of the ASAE 120, 373–398.
Marticorena, B., Bergametti, G., 1995. Modeling the atmospheric dust cycle: 1.
Design of a soil-derived dust emission scheme. Journal of Geophysical Research
100 (D8), 16415–16430.
Oke, T.R., 1988. Street design and urban canopy layer climate. Energy and Building
11, 103–113.
Okin, G.S., Gillette, D.A., Herrick, J.E., 2006. Multi-scale controls on and
consequences of aeolian processes in landscape change in arid and semi-arid
environments. Journal of Arid Environments 65, 253–275.
Rasmussen, K.R., Kok, J.F., Merrison, J.P., 2009. Enhancement in wind-driven sand
transport by electric ﬁelds. Planetary and Space Science 57, 804–808.
Sutton, S.L.F., McKenna Neuman, C., 2008a. Sediment entrainment to the lee of
roughness elements: effects of vertical structures. Journal of Geophysical
Research 113, F02S09. http://dx.doi.org/10.1029/2007JF000783.
Sutton, S.L.F., McKenna Neuman, C., 2008b. Variation in bed level shear stress on
surfaces sheltered by nonerodible roughness elements. Journal of Geophysical
Research 113, F03016. http://dx.doi.org/10.1029/2007JF000967.
Udo, K., Takewaka, S., 2007. Experimental study of blown sand in a vegetated area.
Journal of Coastal Research 23 (5), 1175–1182.
Walter, B., Gromke, C.B., Leonard, K.C., Manes, C., Lehning, M., 2012. Spatiotemporal surface shear-stress variability in live plant canopies and cube arrays.
Boundary-Layer Meteorology 143 (2), 337–356.
Wiggs, G.F.S., Thomas, D.S.G., Bullard, J.E., Livingstone, I., 1995. Dune mobility and
vegetation cover in the south-west Kalahari Desert. Earth Surface Processes and
Landforms 20, 515–529.
Willetts, B.B., Rice, M.A., 1988. Particle dislodgement from a ﬂat sand bed by wind.
Earth Surface Processes and Landforms 13, 717–728.
Wolfe, S.A., Nickling, W.G., 1993. The protective role of sparse vegetation in wind
erosion. Progress in Physical Geography 17 (1), 50–68.