You are on page 1of 15

Atmospheric Environment 34 (2000) 1043}1057

Wind tunnel and "eld calibration of six aeolian dust samplers


Dirk Goossens!,",*, Zvi Y. O!er#
!Laboratory for Experimental Geomorphology, Katholieke Universiteit Leuven, Redingenstraat 16 bis, B-3000 Leuven, Belgium
"Erosion and Soil and Water Conservation Group, Wageningen Agricultural University, Nieuwe Kanaal 11, NL-6709 PA Wageningen,
Netherlands
#Desert Meteorology Unit, Center for Environmental Physics, The Jacob Blaustein Institute for Desert Research, Ben-Gurion University of the
Negev, Sede Boqer Campus 84990, Israel
Received 1 February 1999; accepted 13 August 1999

Abstract

The e$ciency of six aeolian dust samplers was tested via wind tunnel experiments and "eld measurements. In the wind
tunnel, four samplers designed to measure the horizontal dust #ux and one sampler designed to measure the vertical dust
#ux (in the downward direction, i.e., deposition) were calibrated against an isokinetic reference sampler. The horizontal
dust #ux samplers were: the big spring number eight sampler (BSNE), the modi"ed Wilson and Cooke sampler (MWAC),
the suspended sediment trap (SUSTRA), and the wedge dust #ux gauge (WDFG). Vertical deposition #ux was measured
using a marble dust collector (MDCO). A modi"ed Sartorius SM 16711 dust sampler with adjustable #ow rate
(SARTORIUS) was used as isokinetic reference sampler. In the "eld experiments, the WDFG was replaced by a Sierra
ultra high volume dust sampler (SIERRA). Wind tunnel calibrations were carried out at "ve wind velocities ranging from
1 to 5 m s~1. Field calibrations were conducted during seven periods of two weeks each. The most e$cient samplers are
the MWAC and the SIERRA, followed by the BSNE and the SUSTRA. The WDFG is more e!ective than the BSNE at
velocities below 3 m s~1, but its e$ciency drops quickly at higher wind speeds. The most recommendable sampler for
"eld measurements is the BSNE, because its e$ciency varies only very slightly with wind speed. In the absence of
horizontal #ux samplers, the MDCO collector can be used as an alternative to assess horizontal dust #ux and airborne
dust concentration provided the appropriate calibrations are made. ( 2000 Elsevier Science Ltd. All rights reserved.

Keywords: Wind; Aeolian dust; Calibration; Dust #ux gauge; Dust deposit gauge

1. Introduction necessary to assess the intensity of aeolian processes in


a given environment. Horizontal and vertical sediment
The measurement of sediment #ux by wind remains #ux are important input parameters in geomorphic mod-
one of the most problematic procedures in aeolian els. Airborne dust concentration, for example, is usually
geomorphology. Reliable and direct measurements of the calculated from horizontal #ux data. Also, the rate of
sediment #ux are not only required for the con"rmation deposition of particles on the surface is directly a function
and calibration of theoretically derived #ux equations of both the vertical and the horizontal sediment #ux.
(Nickling and McKenna Neuman, 1997) but are also The most important characteristic of a sediment trap is
its e$ciency. As an obstacle in the #ow, a sampler always
a!ects that #ow and, consequently, also the trajectory of
submerged particles. Most collectors have e$ciencies
* Corresponding author. Present address: Laboratory for below unity, typifying the frictional losses in the #uid and
Experimental Geomorphology, Katholieke Universiteit Leuven, in the particle's motion. E$ciencies larger than unity are
Redingenstraat 16 bis, B-3000 Leuven, Belgium. also possible, however, especially when the air#ow near
E-mail address: dirk.goossens@geo.kuleuven.ac.be (D. Goos- the inlet is accelerated due to venturi e!ects caused by the
sens) collector's design. A sediment collector need not be 100%

1352-2310/00/$ - see front matter ( 2000 Elsevier Science Ltd. All rights reserved.
PII: S 1 3 5 2 - 2 3 1 0 ( 9 9 ) 0 0 3 7 6 - 3
1044 D. Goossens, Z.Y. Ower / Atmospheric Environment 34 (2000) 1043}1057

e$cient to be useful (Drew and Lippmann, 1978). Numerous aeolian sediment traps have been described
Low e$ciencies may be an advantage, for example in the literature. The great majority were designed for the
when high-intensity events with a long duration are collection of sand, i.e., particles '63 lm, which are
sampled. To obtain correct #ux data, however, it is usually transported in saltation. A recent overview of the
necessary that the collector's e$ciency is always known, aeolian sand trap literature was published by Nickling
in any condition, and that the appropriate corrections and McKenna Neuman (1997). In the present publication
are made. we deal with aeolian dust traps, designed to trap particles
The e$ciency of a collector depends on di!erent fac- that are merely transported in suspension.
tors. The "rst (and most important) of these is collector Generally speaking there are two groups of dust
design, i.e., the size and shape of the collector when samplers: those measuring horizontal dust #uxes (the
submerged in the #ow. Both parameters determine the sampling ori"ce is in a vertical position) and those
degree of obstruction (in aerodynamic terms: the drag) measuring vertical dust #uxes (the sampling ori"ce is in
of a collector, and, consequently, the rate of perturbation a horizontal position). Hybrid types, measuring the hori-
of the particles' trajectories. The second (and equally zontal #ux by sucking in the air through a horizontal
important) factor is the speed of the wind relative to that ori"ce, also exist.
of the collector. Particle #ow near and in a collector Vertical dust #ux samplers have been described by
strongly depends on wind speed, a!ecting collector Clements et al. (1963), KoK hler and Fleck (1963),
e$ciency. Clough (1975), Ganor (1975), Ralph and Barrett (1976),
Particle size, the third factor, is important since it SkaK rby (1977), Pattenden (cited in Steen, 1979), Good-
determines the degree of inertia of the grains. Very small man et al. (1979), BuK cher (1986), Hall and Waters (1986),
particles closely follow the streamlines of the air (very Hall and Upton (1988), Goossens and O!er (1990),
important near the inlet of a collector!) but will also Orange et al. (1990), Pye (1992), O!er et al. (1992), Goos-
easily leave the collector if no appropriate trap device has sens and O!er (1993) and Littmann (1997). Vertical sam-
been installed. Large particles are characterised by plers are usually passive, i.e., the air is not actively sucked
a higher inertia and are more easily trapped by the into the trap.
collector. Horizontal dust #ux samplers have been described by
The fourth factor is the sediment trap in the collector Steen (1977), Wilson and Cooke (1980), White (1982),
itself. The type of trap is very important in the case of "ne Fryrear (1986), Ralph and Hall (1989), Stout and Fryrear
aeolian dust as the particles penetrate easily through (1989), Janssen and Tetzla! (1991), Leys and Raupach
even very small ori"ces in the trap. (1991), Nickling and Gillies (1993), Shao et al. (1993) and
The last factor, "nally, is time. Traps become more Hall et al. (1994). In these instruments the sediment "rst
and more saturated as sampling continues, and this may enters the trap via an ori"ce and is then trapped by
have positive as well as negative e!ects. Active samplers a "lter or in a settling chamber. Several of the samplers
sucking in the air (and the particles) at a constant rate were originally constructed to trap sand or soil particles,
become more e$cient as the "lters inside the instrument but they are also suitable for dust measurements. These
become polluted with sediment, but if there is no auto- are all passive devices, except for the samplers used
matic adjustment of the #ow (compensating for "lter by Nickling and Gillies (1993) and Leys and Raupach
blockage) collector e$ciency will rapidly drop. Analog- (1991), which use pumping to maintain air#ow through
ous ambivalent e!ects may occur in static (non-sucking) the trap. A detailed overview of commercially available
samplers. active traps (high and low air volume samplers) is given
Ideally, a collector should always be isokinetic. by Furtado (1978). Although this publication is more
This means that the instantaneous wind speed through than 20 yr old, many instruments described are still avail-
the sampling ori"ce is equal to the ambient wind speed able.
along the #ow streamline just upwind of the ori"ce Studies investigating the absolute e$ciency of horizon-
(Nickling and McKenna Neuman, 1997). In such condi- tal dust #ux samplers are scanty. Moreover, many calib-
tions, sampling does not distort the #ow at the sampling rations published in the literature are unreliable because
ori"ce and there is no distortion of the particles' trajecto- they are based on incorrect experimental procedures. The
ries. However, sampling isokinetically does not guaran- only way to measure the absolute e$ciency of a horizon-
tee an e$ciency of 100%. There may be problems with tal #ux sampler is to compare the #ux measured by the
particles rebounding on the suction tube close to the sampler to the #ux measured by an isokinetic sampler
sampling ori"ce, and, especially in the case of dust, there installed at the same location and in identical conditions.
is the risk of spontaneous deposition of particles in the In this paper, several horizontal single-point dust #ux
suction tube itself (Belyaev and Levin, 1972). A careful samplers are calibrated against an isokinetic sampler. To
construction of the sampler is therefore necessary, and in assess the relevancy of wind tunnel calibrations to "eld
such cases, e$ciencies approaching 100% may be ob- situations, calibrations were carried out both in the wind
tained. tunnel and in "eld experiments.
D. Goossens, Z.Y. Ower / Atmospheric Environment 34 (2000) 1043}1057 1045

Fig. 1. Photography and construction scheme of the BSNE sampler.

2. Description of the samplers "ne particles out of the top of the screen. A rubber
retainer closes any small holes in the back or front of the
Five horizontal dust #ux samplers and one vertical assembled sampler. A wind vane at the rear assures the
dust #ux sampler were selected from the literature and sampler to be turned to the wind.
compared to an isokinetic sampler. A description of the More details about the sampler can be found in the
samplers is provided below. original description by Fryrear (1986).

Sampler 1: BSNE Sampler 2: SUSTRA


The big spring number eight (or BSNE) sampler was The suspended sediment trap (or SUSTRA) was de-
developed by D.W. Fryrear in 1986. Although it was veloped by W. Janssen and G. Tetzla! in 1991. Similar to
originally designed to collect airborne dust, it is now also the BSNE, it is now used for the collection of di!erent
frequently used to collect soil and sand. A picture and types of sediment (dust, sand and soil). A picture and
a technical scheme are shown in Fig. 1. The sampler is a technical scheme are shown in Fig. 2. The dust-laden
constructed of 28-gauge galvanised metal, galvanised 18- air enters the instrument via a horizontal metal tube 5 cm
mesh screen, and stainless-steel 60-mesh screen. Dust- in diameter and rebounds onto a metal plate inside
laden air passes through a vertical 2 cm]5 cm sampler a central vertical pipe. Particles settle onto a plastic dish
opening. Once inside the sampler, air speed is reduced placed on top of an electronic balance underneath the
and the dust settles out in a collection pan. Air discharges pipe, except the "nest ones, which may leave the sampler
through a 60-mesh screen. An 18-mesh screen reduces the via a small outlet on top of the pipe (together with the air
movement of the deposited material, preventing break- discharging from the instrument). Similar to the BSNE,
down of the collected sediment and potential loss of very a wind vane turns the instrument into the wind at all
1046 D. Goossens, Z.Y. Ower / Atmospheric Environment 34 (2000) 1043}1057

Fig. 2. Photography and construction scheme of the SUSTRA


sampler.

Fig. 3. Photography and construction scheme of the MWAC


sampler.
times. To minimise air#ow disturbance near the surface,
the balance is placed in a metal box dug into the ground.
In the original version the horizontal inlet tube is at and an outlet tube have been added (Fig. 3). The bottle is
a height of about 23 cm above the surface, but this value installed vertically, with the inlet oriented to the wind.
can be adjusted while digging the balance box into the Sediment entering the bottle will be deposited due to the
ground. More details on the instrument are described in pressure drop created by the di!erence in diameter be-
the original paper by Janssen and Tetzla! (1991). tween the bottle and the inlet and outlet tubes. The clean
air then discharges from the bottle via the outlet.
Sampler 3: MWAC The original concept was later slightly modi"ed by
The modi"ed wilson and cooke (or MWAC) sampler is Kuntze et al. (1990), who attached the bottle in a horizon-
based on an original design developed by S.J. Wilson and tal (not vertical) position to a mast provided with a wind
R.U. Cooke in 1980. The sampler consists of a plastic vane. Attaching several bottles at di!erent levels to the
bottle, "guring as settling chamber, to which an inlet tube mast, vertical #ux pro"les can be measured (Sterk, 1993).
D. Goossens, Z.Y. Ower / Atmospheric Environment 34 (2000) 1043}1057 1047

Fig. 4. Photography and construction scheme of the WDFG sampler.

The dimensions of the bottle used in our experiments The box contains a particle trap made from 10 pores in~1
are indicated in Fig. 3. The inlet and outlet tubes were open-celled foam, which is normally sprayed with
made of glass 1.25 mm thick, with an inner diameter of a thin sticky coating to retain any impacting particle. The
7.5 mm and an outer diameter of 10.0 mm. foam we received for the calibration tests was not treated
with the coating, however, and the results presented in
Sampler 4: WDFG this paper refer to clean-foam conditions. The layer of
The wedge dust #ux gauge (or WDFG) was developed foam was 3 cm deep and was set with its rear face 2 cm
by Hall et al. (1993). A picture and a technical scheme are from the back face of the box. More details about the
shown in Fig. 4. The sampler we calibrated was a half- WDFG, including an aerodynamic discussion of its de-
scale model of the original sampler, and the dimensions sign, can be found in the original description by Hall
shown in Fig. 4 refer to that half-scale version. The et al. (1994).
WDFG is commercially available in both dimensions.
The WDFG consists of a simple, parallel sided box, Sampler 5: SIERRA
wedge shaped in elevation and with extended sides to- The SIERRA sampler (Model Series Number: 680)
wards the rear holding a ba%e plate. The #at, horizontal is a commercially available ultra high volume dust
bottom of the box is 18 cm long and 10 cm wide. The top aspirator with cascade impactors, manufactured by
slopes upwards at an angle of 24.53. Sediment-laden air Sierra/Misco, Inc. A picture and a scheme are shown in
enters the instrument via a 1.9]10.0 cm rectangular slot. Fig. 5. Dust-laden air is sucked into the apparatus via
1048 D. Goossens, Z.Y. Ower / Atmospheric Environment 34 (2000) 1043}1057

Fig. 5. Photography and construction scheme of the SIERRA sampler.

a 38.5]39.0 cm horizontal inlet, over which a protecting rate. It has been registered as the Sartorius Membran"l-
roof has been installed. The air then passes through ter SM 16711. In its original version, the dust-laden air is
a 22.8]17.8 cm glass micro"bre "lter trapping even the directly sucked into a "lter holder containing a "lter with
smallest particles, and the clean air then discharges from a diameter of 4 cm. The "lter holder is connected to the
the instrument. Flow discharge through the "lter is ad- sampler via a #exible plastic tube, allowing the holder to
justable and remains constant during the whole measur- be installed at any desired location well away from the
ing period. The sampler is thus an active sampler, though sampler. Flow discharge through the "lter is adjustable
not isokinetic. between 200 and 1800 l h~1, and the actual #ow rate can
The "lter type used in the experiments was Whatman be read at any time from the instrument. A #ow regulator
glass micro"bre "lter, cat. No. 1827866. Although the button allows the operator to adjust the air#ow, for
sucking of the dust-laden air near the roof of the instru- example to compensate for clogging of the "lter during
ment occurs in a vertical direction, the instrument is used the measurement.
by many researchers to determine the horizontal dust To measure at wind velocities between 1 and 5 m s~1,
#ux by multiplying the airborne dust concentration a funnel-shaped extension cap was attached to the "lter
(measured by the instrument) with the horizontal wind holder during the experiments (see Fig. 6). Using this cap,
speed at inlet level. In this study, horizontal dust #uxes the diameter of the inlet of the suction slot was reduced
for the SIERRA were always calculated in this way. from 4 to 0.87 cm. The cap was of plastic 1 mm thick. To
minimise the errors resulting from particle rebound on
Sampler 6: SARTORIUS (isokinetic sampler) the cap front, the cap was sharply pointed over a distance
This commercially available instrument is an active of 1 cm near its end. Silicone paste was used to close any
sampler sucking in the air (and the dust) at an adjustable holes between the "lter holder and the extension cap.
D. Goossens, Z.Y. Ower / Atmospheric Environment 34 (2000) 1043}1057 1049

Fig. 6. Photography and construction scheme of the SARTORIUS sampler.

The type of "lter used was the same as in the SIERRA 3. Experimental procedure
sampler: Whatman glass micro"bre "lter, cat. No.
1827866. 3.1. The wind tunnel experiments

Sampler 7: MDCO (vertical dust yux sampler) All wind tunnel experiments were carried out in the
To have an idea of the ratio of the vertical deposition closed-return wind tunnel of the Laboratory for Experi-
#ux to the horizontal dust #ux, a vertical #ux sampler mental Geomorphology of the Katholieke Universiteit
was included in the experimental programme. From the Leuven, Belgium. The tunnel contains two test channels.
many samplers reported in the literature we selected the All experiments were conducted in the large channel,
marble dust collector (or MDCO) because this sampler is which is 7.6 m long, 1.2 m wide and 0.6 m high.
currently used world-wide and has become very popular, During the experiments, dust was added to the air
especially in desert research. It was "rst described by current by connecting an Engelhardt laboratory dust
Ganor (1975). It consists of a plastic tray with one cloud producer to the tunnel. This apparatus ensures
(or two) layers of marbles. It can be installed on the a continuous feed of dust particles to the air#ow, and
earth's surface, or attached to vertical masts. Various allows the operator to adjust dust discharge. The dust
types of trays are in use, but circular trays are better than was added to the #ow in the return section of the tunnel,
rectangular ones because these latter are direction-sensi- via a small (but wide) ori"ce in the tunnel wall. During its
tive. The tray we used in the wind tunnel was circular, passage through the tunnel, it is fully dispersed over the
with a diameter of 31.5 cm and a height of 10 cm (Fig. 7). tunnel section. The exact distribution of the dust at the
The marbles were standard glass marbles 1.6 cm in location were the experiments were carried out was accu-
diameter. Two layers of marbles were installed in rately measured so that the appropriate corrections
the MDCO. could be made when necessary. Further technical details
1050 D. Goossens, Z.Y. Ower / Atmospheric Environment 34 (2000) 1043}1057

23.5 cm above the tunnel #oor. No special preparation of


the fetch was used, and under such conditions, the depth
of the boundary layer in the tunnel is of the order of 7 cm,
i.e., signi"cantly below 23.5 cm. All tests were thus car-
ried out in the free-stream region of the tunnel. Samplers
were oriented such that the average direction of dust
transport was parallel to the sampler inlet axis (i.e., the
inlet was always perpendicular to the dust #ow).
Five wind velocities were selected: 1, 2, 3, 4 and
5 m s~1. The velocities were measured with an accuracy
of 0.01 m s~1 using a digital Furness FC016 manometer.
No higher speeds could be installed in the test section
used, but 1}5 m s~1 is an acceptable range for most
deserts (especially at a height of 1 m, which is the stan-
dard height for dust concentration measurements over
land, see Sehmel, 1980). Wind speeds were measured with
a standard Pitot tube, at the location where the samplers
were later installed.
During the wind tunnel tests, the dust cloud machine
produced a dust discharge of 13.3 kg h~1. At the sam-
plers' location, airborne dust concentration was approx-
imately 2 g m~3, slightly varying with wind velocity.
For each sampler, three independent test runs were
carried out at each velocity. Before each run the sampler
was covered with a plastic sheet to prevent any dust
entering. Once the correct wind speed had been attained,
the sheet was removed and the experiment started. Each
Fig. 7. Photography and construction scheme of the MDCO run was of 4 min. After the run, the wind tunnel was
sampler.
stopped and an oblique plate connected to the tunnel
roof was immediately positioned before the sampler to
avoid any dust entering the trap. The sampler was then
of the wind tunnel and the dust cloud producer can be taken out of the tunnel, and the dust collected was
found in a description by Goossens and O!er (1988). weighed with an accuracy of 0.0001 g on a high-precision
The dust used was prepared from Belgian Brabantian balance. The horizontal dust #ux was then calculated.
loess. The loess was dried, ground and sieved through To correct the results for small deviations that could
a 63 lm sieve to exclude all sand particles. It consisted of accidentally occur in the dust concentration values, a ref-
95% silt (2}63 lm) and 5% clay ((2 lm) and had erence sampler (we used a second BSNE) was installed at
a median diameter of 30 lm (see Fig. 8). This corresponds a "xed position in the wind tunnel during each run. Also
closely to the size of modern wind-blown dust near the this sampler was initially covered by a plastic sheet and
earth's surface (Swineford and Frye, 1945; Zeuner, 1949; protected by the oblique plate after completion of the
PeH weH , 1951; Yaalon and Ganor, 1979; O!er and Goos- experiment. All #uxes measured by the test samplers were
sens, 2000; Nickling et al., 1999). Only 1% of the particles later corrected for dust concentration di!erences that
was coarser than 50 lm. could have occurred during the experiments.
Due to its size, the SIERRA sampler could not be In the BSNE runs, the dust was collected in a small
tested in the wind tunnel. Thus only "eld results are container that "tted in the sedimentation chamber of the
available for this sampler. To allow the SUSTRA to enter sampler. The dust caught by the collector was deter-
into the tunnel, the balance, the balance box and the mined by weighing the container before and after each
wind vane were removed from the instrument. This does experiment. In the MWAC runs, the complete sampler
not seriously a!ect the collector's e$ciency since the (bottle#inlet and outlet tubes) was weighed before and
balance and the box are normally buried under the soil after each experiment. In the SUSTRA runs, the dust was
surface and the vane is always turned to the wind. All directly caught on a metal plate put underneath the
other samplers were tested in their original state in the central vertical tube of the sampler. The SUSTRA bal-
wind tunnel. ance was not used, as mentioned earlier. In the WDFG
During a test, the sampler was placed in the centre of runs, the dust was collected from the sampler by cleaning
the tunnel at a fetch of 585 cm. Installation was such that the sedimentation chamber with a brush and weighing
the centre of the sampler's inlet was always at a height of the foam before and after each experiment. In the
D. Goossens, Z.Y. Ower / Atmospheric Environment 34 (2000) 1043}1057 1051

Fig. 8. Grain-size distribution of the wind tunnel dust (solid line) and the Negev dust (dotted line).

MDCO runs, dust deposition was directly determined by be connected to the anemometers measuring the instan-
weighing the sampler (with the marbles) before and after taneous wind speed at Sede Boqer. No isokinetic
the experiments. A balance with an accuracy of 0.001 measurements of the dust #ux could therefore be carried
g (instead of 0.0001 g) was used for this collector. In the out, and no absolute calibration data are available. The
SARTORIUS runs, "nally, the glass micro"bre "lters relative e$ciencies of the di!erent samplers were mea-
and the extension cap were weighed before and after each sured, however, and these can be compared to those
experiment and the quantity of dust accumulated was derived from the wind tunnel experiments.
then calculated. All samplers were installed at a height of 1 m above the
desert surface. They were installed close to each other (to
3.2. The xeld experiments avoid spatial di!erences in wind speed and airborne dust
concentration), but special care was taken to avoid any
To evaluate the wind tunnel results, several samplers mutual interaction. They could freely rotate and turn
were installed at the meteorological station of the Jacob themselves to the wind. Similar to the wind tunnel tests,
Blaustein Institute for Desert Research, Sede Boqer, Is- the balance box of the SUSTRA was removed from the
rael. The station is situated in the northern part of the instrument and the dust was directly caught on a small
Negev desert, in an area characterised by a high aeolian dish placed immediately under the sampler. Since the
dust activity. Dust monitoring at the station started in MWAC samplers are very small, two copies were instal-
1986. Since then, many measurements of sediment con- led in the "eld. MWAC data presented later in this paper
centration and particle deposition and accumulation refer to the average of both measurements. The SIERRA
have been conducted. It is an ideal location to carry out sampler was installed at a suction rate of 56 m3 h~1
the "eld calibration of the dust samplers. (41 m3 h~1 between 17 September and 15 November
All "eld measurements started in May 1998. Since the 1998), corresponding to a suction velocity of 0.38 m s~1
half-size WDFG was only available from October 1998, (0.28 m s~1 between 17 September and 15 November
no WDFG "eld calibration could be carried out. Instead 1998).
of the WDFG, the SIERRA sampler, which was too large Fig. 8 shows a grain-size distribution curve for typical
to be calibrated in the wind tunnel, was tested. The Sede Boqer dust. Grain size is similar to that of the dust
isokinetic SARTORIUS was not used since it cannot used in the wind tunnel.
1052 D. Goossens, Z.Y. Ower / Atmospheric Environment 34 (2000) 1043}1057

Fig. 9. Absolute e$ciency of the SUSTRA, BSNE, MWAC and WDFG samplers. The vertical downward dust #ux as recorded by the
MDCO sampler (relative to the horizontal dust #ux measured by the SARTORIUS) is also shown in the "gure.

The samplers were installed in May 1998 and remained sampler data are combined with the three SARTORIUS
outside until November 1998. No rain occurred in this data (nine combinations in total). Deviations are rather
period, so the results are not a!ected by undesired clean- small, and remain usually within 10% of the average values.
ing or by splash. Altogether, seven experiments of two Comparing the four horizontal #ux samplers in Fig. 9,
weeks each were carried out. During these experiments, there are large di!erences between the samplers. By far
wind speed was measured every second at di!erent levels the highest e$ciencies are recorded by the MWAC
from a meteorological mast close to the samplers. Using bottles, which show an e$ciency of more than 90% for
the logarithmic wind pro"le, appropriate values were wind speeds between 2 and 5 m s~1 (75% at a speed of
calculated for a height of 1 m (the height of the samplers' 1 m s~1). The WDFG sampler attains a maximum e$-
inlets). Average speed was then calculated for each of the ciency of 63% at 2 m s~1, but the e$ciency then dramati-
seven periods. cally drops with increasing wind speed. The wedged
shape, originally designed to create a pressure di!erence
across the sampler's front and rear openings su$cient to
drive a #ow through the gauge and to overcome the
4. Experimental results pressure drop caused by the porous foam (Hall et al.,
1994), apparently is insu$cient to create near-isokinetic
4.1. The wind tunnel calibrations #ow. Especially at high wind velocities (4 m s~1 and
more) the performance of the sampler is poor (e$cien-
Fig. 9 shows, for the four horizontal dust #ux samplers cy(30%). The BSNE sampler is less e!ective compared
tested in the wind tunnel (BSNE, SUSTRA, MWAC and to the MWAC (and, for low velocities, also to the
WDFG), the absolute e$ciency within the wind velocity WDFG), but the great advantage of this sampler is that
interval 1}5 m s~1. To have an idea of the ratio of the its e$ciency varies only slightly with wind speed. In the
vertical deposition #ux to the horizontal dust #ux, velocity interval investigated (1}5 m s~1) it is always be-
the results of the MDCO measurements are shown tween 35 and 45%, and, also important, there seems to be
in the same picture. They represent the ratio of the no general relationship with the wind. The SUSTRA
vertical deposition #ux as measured by the marble sur- sampler, "nally, is the least e$cient of all the samplers
face to the horizontal #ux as measured by the isokinetic tested. Its e$ciency increases with wind speed, but even
SARTOR assuming the former to be perfectly absorbent at a speed of 5 m s~1 it is only 15% e$cient. Also because
(which is all right for the wind velocity range used). The of its size, which makes it extremely di$cult to install
dots in the "gure represent the average of the e$ciencies several samplers in a vertical array, the SUSTRA seems
measured during the three replica experiments. The verti- to be of little merit for aeolian dust research. In sand
cal bars indicate the maximum deviation when the three research it o!ers much more possibilities, not only
D. Goossens, Z.Y. Ower / Atmospheric Environment 34 (2000) 1043}1057 1053

Table 1
Horizontal dust #uxes measured in the "eld. Data in parenthesis represent the #ux relative to the BSNE

Period Wind speed at 1 m Horizontal dust #ux (10~5 g m~2 s~1)


(m s~1)
SIERRA MWAC BSNE SUSTRA

29/05/1998}16/06/1998 2.62 25.49 (4.97) 18.28 (3.56) 5.13 (1.00) 3.36 (0.65)
16/07/1998}30/07/1998 2.68 16.91 (9.04) } 1.87 (1.00) 1.09 (0.58)
30/07/1998}17/08/1998 2.73 25.07 (4.17) 34.46 (5.73) 6.01 (1.00) 2.35 (0.39)
17/08/1998}02/09-1998 2.53 17.10 (4.05) 14.71 (3.49) 4.22 (1.00) 1.54 (0.36)
02/09-1998}15/09-1998 2.30 39.05 (4.77) 44.79 (5.48) 8.18 (1.00) 5.78 (0.71)
14/10/1998}01/11-1998 2.26 15.55 (3.18) 25.03 (5.12) 4.89 (1.00) 1.12 (0.23)
01/11-1998}15/11/1998 1.91 14.96 (9.23) 11.57 (7.14) 1.62 (1.00) 0.43 (0.27)

because sand particles are usually transported close to winds, intermittent bursts remain very important. This is
the surface, but also because they are su$ciently large to particularly the case if the dust has been locally eroded
be detected by the electronic balance. This allows the and has not yet become fully dispersed in the atmo-
operator to determine very accurately the periods of sphere. Once created, waves of high dust concentration
particle transport. The MDCO curve in Fig. 9 will be may persist for long times. The pulses not only result in
discussed later (Section 5). high horizontal dust #uxes, but also in high vertical
#uxes, and thus in a high deposition. They are of special
4.2. The xeld calibrations importance with respect to the calibration experiments
conducted in the Negev. Due to the type of the samplers,
The data for all seven "eld calibrations are sum- no instantaneous determinations of the horizontal dust
marised in Table 1. To facilitate direct comparisons be- #ux were possible. The results presented in Table 1 are
tween the samplers, the table shows, besides the absolute based on 2-week experiments, the minimum that allow
#ux data, also the relative #uxes with respect to the reliable measurements of the sediment weight. Using
BSNE (in parenthesis). The BSNE was used as a refer- larger samplers may help decrease the sampling time but
ence because it is currently one of the most frequently will not signi"cantly help solve the problem, for intermit-
used samplers in dust research and because its e$ciency tent dust transport occurs at all frequencies of the spec-
is almost velocity independent (Section 4.1). Table 1 con- trum. Even in a period of only a few minutes, there is so
"rms the general picture of the wind tunnel results, with much variation in the dust #ow that average values are of
the SUSTRA being the least e$cient sampler followed by only little (if any) signi"cance. The 2-week average #ux
the BSNE and the MWAC. As an average the SIERRA values are well documented, but it is fully unknown
sampler measures the same dust quantities as the when, and at which wind speeds exactly, the largest
MWAC bottles, but there are di!erences between the amounts of dust entered the samplers. Looking at Fig. 9,
distinct experiments. Similar variations are observed for there is already a signi"cant variation in the velocity
the other samplers although the average wind speeds are interval 1}5 m s~1, and similar variations may be ex-
more or less comparable. pected for larger (or smaller) speeds.
It can be concluded that the "eld ranking and the wind The only solution to obtain correct #ux data is to use
tunnel ranking of the samplers is identical, but despite the an automatic isokinetic sampler connected to an instan-
rather small di!erences in average wind speed, there is taneous anemometer. A much easier, and also much
some variation among the experiments. This is an impor- cheaper alternative is to use a sampler whose e$ciency is
tant conclusion, which deserves more attention. rather independent of wind speed. The BSNE is a useful
solution in this context. Its e$ciency for 30 lm dust is
always around 40% (varying between 35 and 45%), at
5. Discussion least within the wind interval 1}5 m s~1. No assessment
of the absolute e$ciency of the SIERRA sampler could
The rather large variation in the "eld data (despite of be made, but taken into account its size (and price) a set
restricted variations in average wind speed) is caused by of a few vertically mounted BSNEs is much more eco-
the unsteadiness of the dust #ow. It is well known that nomic (and may also provide more information).
boundary layer winds are usually highly unsteady (Stout Several of the samplers investigated in this study have
and Zobeck, 1997). Although gusty winds that drive been calibrated before, though not always for aeolian
suspended particles generally produce less temporal vari- dust. The BSNE was already tested by Fryrear in 1986,
ations in sediment transport than saltation driving and later by Stout and Fryrear (1989) and Shao et al.
1054 D. Goossens, Z.Y. Ower / Atmospheric Environment 34 (2000) 1043}1057

Fig. 10. Relationship between the horizontal dust #ux derived from the SIERRA and the vertical dust #ux as measured by the MDCO.

(1993), although the latter authors used a slightly di!er- Finally, we consider the MDCO data. These samplers
ent design by adding a rain hood to the instrument. For do not measure the horizontal, but the vertical (depo-
sand, the overall e$ciency of the trap varied between 86 sition) dust #ux. Although the vertical rim seriously
and 96%. For the "ne particle sizes ((63 lm), Fryrear disturbs the particle #ow, MDCOs have become very
(1986) found e$ciencies between 96 and 165%, but Shao popular in dust research since they are very cheap and
et al. (1993), who used an isokinetic sampler as a refer- also easy to install. In the absence of a horizontal #ux
ence, concluded that the BSNE is only 40% e$cient for sampler, MDCOs installed at ground level may be used
particles (10 lm. This value corresponds very well to to estimate horizontal dust #uxes and airborne dust
our results with 30 lm dust (35}45%). A rather pre- concentrations, as will be illustrated later.
liminar calibration of the WDFG was published by Hall Fig. 9 shows that, in the wind tunnel, the e$ciency of
et al. (1993). E$ciency of the sampler was investigated for the MDCO decreases with increasing wind speed. The
two wind speeds (3 and 8 m s~1) and 2 grain sizes (59 and ratios of the deposition #ux (as measured by the MDCO)
136 lm). For 59 lm dust, the WDFG was 80% e$cient to the horizontal dust #ux (as measured by the isokinetic
for both wind speeds. For 136 lm sand, e$ciency varied sampler) are: 4.26% (1 m s~1), 0.93% (2 m s~1), 0.37%
between 90 and 95%, with the highest value at the lowest (3 m s~1), 0.25% (4 m s~1), and 0.22% (5 m s~1). The
wind. In our own study with 30 lm dust, e$ciency of the systematic decrease is mainly caused by the collector's
WDFG is clearly lower, and there is also a strong vari- rim, which creates a dust shadow (and, hence, a zone of
ation with wind speed, making the sampler less useful for low airborne sediment concentration) in its lee. The size
"ne dusts. For the other horizontal #ux samplers tested of the shadow is directly proportional to wind speed, as is
in the wind tunnel (SUSTRA and MWAC), no e$ciency the decrease of deposition in the sampler.
data are available from the literature, at least not for dust The relationship between the horizontal dust #ux (cal-
(sand calibrations were conducted by Sterk (1993), Bak- culated from the active SIERRA gauge) and the vertical
kum (1994) and Pollet (1995) for the MWAC sampler, deposition #ux as measured by the MDCO collector is
leading to e$ciencies between 49 and 69%, and by Spaan shown in Fig. 10. From the "gure, it is possible to
et al. (1990) for the SUSTRA, but the latter study does estimate roughly the horizontal #ux when the accumula-
not provide absolute e$ciency data). tion in the MDCO is known. In Fig. 11, the relationship
D. Goossens, Z.Y. Ower / Atmospheric Environment 34 (2000) 1043}1057 1055

Fig. 11. Relationship between airborne dust concentration at 1 m (measured by the SIERRA) and particle accumulation at ground-level
(measured in an MDCO) during a 2-yr experiment at Sede Boqer.

between airborne dust concentration at 1 m (measured below 15% even at the highest speeds tested. All percent-
by the SIERRA) and particle accumulation at ground- ages are for the 30 lm dust shown in Fig. 8.
level (in an MDCO) is shown for a 2-yr experiment In the "eld experiments, the passive WDFG was re-
carried out earlier at Sede Boqer (Goossens and O!er, placed by the active SIERRA. Highest #uxes were ob-
1995). Also here, concentrations can roughly be esti- tained using the MWAC and SIERRA samplers (both
mated from the MDCO data. instruments have similar e$ciencies) followed by the
BSNE and the SUSTRA.
The ranking of the samplers with respect to their
6. Conclusions e$ciency is identical in the wind tunnel and in the "eld.
Despite of rather small variations in average wind speed,
Wind tunnel and "eld experiments were carried out to the numerical data of the distinct "eld experiments show
determine the e$ciency of six aeolian dust samplers. In considerable scatter due to the intermittent character of
the wind tunnel, four samplers designed to measure the the air (and dust) #ow.
horizontal dust #ux (BSNE, MWAC, SUSTRA and The most recommendable sampler in the test is the
WDFG) and one sampler designed to measure the verti- BSNE. Although its e$ciency is much lower than that of
cal deposition #ux (MDCO) were calibrated against an the MWAC, it is nearly constant and does not seem to
isokinetic sampler. Calibration was carried out at "ve depend on wind velocity. The BSNE is a low-cost instru-
wind velocities, ranging from 1 to 5 m s~1. The most ment, easy to install, and due to its small size and weight
e$cient sampler is the MWAC, which shows e$ciencies it is very suited to measure vertical #ux pro"les.
of over 75% (and usually over 90%) in the velocity range In the absence of horizontal #ux samplers, the MDCO
investigated. The WDFG has an e$ciency of about collector can be used as an alternative to assess approx-
50}60% at low wind speeds ((3 m s~1), but becomes imate values of the horizontal dust #ux and sediment
much less accurate at higher winds (e$ciency (30% for concentration. Correlation diagrams between MDCO
winds '4 m s~1). The e$ciency of the BSNE is approx- and SIERRA records are shown in the Figs. 10 and 11,
imately 40% (varying between 35 and 45%) and seems to but these are only valid for 30-lm particles and for the
be independent of wind speed. The SUSTRA sampler, conditions in force during the tests. For other particle
"nally, is not very accurate, at least for dust. Its e$ciency sizes and circumstances, appropriate calibrations have to
systematically increases with wind speed, but remains be carried out.
1056 D. Goossens, Z.Y. Ower / Atmospheric Environment 34 (2000) 1043}1057

Acknowledgements Goossens, D., O!er, Z.Y., 1995. Comparisons of day-time and


night-time dust accumulation in a desert region. Journal of
The SUSTRAs used in the test were kindly provided Arid Environments 31, 253}281.
by W. Schaefer (Institute of Soil Technology, Bremen, Hall, D.J., Waters, R.A., 1986. An improved, readily available
dustfall gauge. Atmospheric Environment 20, 219}222.
Germany). The half-scale WDFG was constructed by
Hall, D.J., Upton, S.L., 1988. A wind tunnel study of the
I. Hanby (Newark, U.K.) and was kindly provided
particle collection e$ciency of an inverted frisbee used as
by G. McTainsh (Gri$th University, Nathan, Australia). a dust deposition gauge. Atmospheric Environment 22,
Technical assistance by D. Klepach (Jacob Blaustein 1383}1394.
Institute for Desert Research, Sede Boqer, Israel) and J. Hall, D.J., Upton, S.L., Marsland, G.W., 1993. Improve-
Meersmans (Katholieke Universiteit Leuven, Belgium) is ments in dust gauge design. In: Couling, S. (Ed.), Measure-
also greatly acknowledged. ments of Airborne Pollutants. Butterworth, Heinemann
(Chapter 11).
Hall, D.J., Upton, S.L., Marsland, G.W., 1994. Designs for
References a deposition gauge and a #ux gauge for monitoring ambient
dust. Atmospheric Environment 28, 2963}2979.
Bakkum, A.W.G., 1994. The behaviour of an arti"cial soil crust Janssen, W., Tetzla!, G., 1991. Entwicklung und Eichung einer
in a simulated sand storm. Department of Irrigation and Soil registrierenden Suspensionsfalle. Zeitschrift fuK r Kulturtech-
and Water Conservation, Agricultural University Wagenin- nik und Landesentwicklung 32, 167}180.
gen, pp. 40. KoK hler, A., Fleck, U.W., 1963. Untersuchungen zur Festlegung
Belyaev, S.P., Levin, L.M., 1972. Investigation of aerosol aspira- eines StandardmessgeraK tes fuK r Staubniederschlag. Inst. Me-
tion by photographing particle tracks under #ash illumina- teorol. Abschlussber. No. J. 84, 2.
tion. Journal of Aerosol Science 3, 127}140. Kuntze, H., Beinhauer, R., Tetzla!, G., 1990. Quanti"cation
BuK cher, A., 1986. Recherches sur les poussières mineH rales of soil erosion by wind, I. Final Report of the BMFT
d'origine saharienne. Ph.D. Thesis, UniversiteH de Reims- project. Project No. 0339058 A, B, C. Institute of Meteoro-
Champagne-Ardenne, pp. 290. logy and Climatology, University of Hannover, Germany
Clements, T., Stone, R.O., Mann, J.F, Eyman, J.L., 1963. A study (in German)
of windborne sand and dust in desert areas. US Army, Leys, J.F., Raupach, M.R., 1991. Soil #ux measurements using
Natick Laboratories Mass., Earth Science Division, Tech- a portable wind erosion tunnel. Australian Journal of Soil
nical Report ES-8, Project Ref. 7]83-01-008. Research 29, 533}552.
Clough, W.S., 1975. The deposition of particles on moss and Littmann, T., 1997. Atmospheric input of dust and nitrogen into
grass surfaces. Atmospheric Environment 9, 1113}1119. the Nizzana sand dune ecosystem, north-western Negev,
Drew, R.T., Lippmann, M., 1978. Calibration of air sampling Israel. Journal of Arid Environments 36, 433}457.
instruments. In: Air Sampling Instruments for Evaluation of Nickling, W.G., Gillies, J.A., 1993. Dust emission and transport
Atmospheric Contaminants, 5th Edition. American Confer- in Mali, West Africa. Sedimentology 40, 859}868.
ence of Governmental Industrial Hygienists, Cincinnati, pp. Nickling, W.G., McKenna Neuman, C., 1997. Wind tunnel
I1-I38. evaluation of a wedge-shaped aeolian sediment trap.
Fryrear, D.W., 1986. A "eld dust sampler. Journal of Soil and Geomorphology 18, 333}345.
Water Conservation 41, 117}120. Nickling, W.G., McTainsh, G.H, Leys, J.F., 1999. Dust emis-
Furtado, V.C., 1978. Air movers and samplers. In: Air Sampling sions from the Channel Country of western Queensland,
Instruments for Evaluation of Atmospheric Contaminants, Australia. In press.
5th Edition. American Conference of Governmental Indus- O!er, Z.Y., Goossens, D., 2000. Airborne particle accumulation
trial Hygienists, Cincinnati, pp. K3}K52. and composition at di!erent locations in the northern Negev
Ganor, E., 1975. Atmospheric dust in Israel. Sedimentological desert. Zeitschrift fuK r Geomorphologie, in press.
and meteorological analysis of dust deposition. Ph.D. Thesis, O!er, Z.Y., Goossens, D., Shachak, M., 1992. Aeolian deposition
Hebrew University of Jerusalem. of nitrogen to sandy and loessial ecosystems in the Negev
Goodman, G.T., Inskip, M.J., Smith, S., Parry, G.D.R., Burton, Desert. Journal of Arid Environments 23, 355}363.
M.A.S., 1979. The use of moss-bags in aerosol monitoring. Orange, D., Gac, J.-Y., Probst, J.-L., Tanre, D., 1990. Mesure du
In: Morales, C. (Ed.), Saharan Dust. Mobilization, Trans- deH po( t au sol des aeH rosols deH sertiques. Une meH thode simple
port, Deposition. Wiley, New York, pp. 211}232. de preH lèvement: le capteur pyramidal. Comptes Rendus de
Goossens, D., O!er, Z.Y., 1988. Loess Erosion and Loess Depos- l'AcadeH mie des Sciences, Paris 311, 167}172.
ition in the Negev Desert: Theoretical Modelling and Wind PeH weH , T.L., 1951. An observation on wind-blown silt. Journal of
Tunnel Simulations. The Jacob Blaustein Institute for Desert Geology 59, 399}401.
Research, Desert Meteorology Papers Ser. A, No. 13, pp. 65. Pollet, I., 1995. Meten van windsnelheden en zandtransport in
Goossens, D., O!er, Z.Y., 1990. A wind tunnel simulation and een windtunnel. M.Sc. Thesis, Universiteit Gent, pp. 153.
"eld veri"cation of desert dust deposition (Avdat Experi- Pye, K., 1992. Aeolian dust transport and deposition over Crete
mental Station Negev Desert). Sedimentology 37, 7}22. and adjacent parts of the Mediterranean Sea. Earth Surface
Goossens, D., O!er, Z.Y., 1993. Eolian deposition of dust over Processes and Landforms 17, 271}288.
symmetrical hills: an evaluation of wind tunnel data by Ralph, M.O., Barrett, C.F., 1976. A wind-tunnel study of the
means of terrain measurements. Zeitschrift fuK r Geomor- e$ciency of deposit gauges } interim report. Warren Spring
phologie 37, 103}111. Laboratory, Laboratory Report LR 235 (AP), pp. 19.
D. Goossens, Z.Y. Ower / Atmospheric Environment 34 (2000) 1043}1057 1057

Ralph, M.O., Hall, D.J., 1989. Performance of the British Sterk, G., 1993. Description and calibration of sediment sam-
Standard Directional Dust Gauge. Proceedings of the Aero- plers. Sahelian Wind Erosion Research Project, Report III.
sol Society's Third Annual Conference, 20}22 March 1989, Department of Irrigation and Soil and Water Conservation,
West-Bromwich, U.K. Wageningen Agricultural University, pp. 31.
Sehmel, G.A., 1980. Particle and gas dry deposition: a review. Stout, J.E., Fryrear, D.W., 1989. Performance of a wind-
Atmospheric Environment 14, 983}1011. blown particle sampler. Transactions of the ASAE 32,
Shao, Y., McTainsh, G.H., Leys, J.F., Raupach, M.R., 1993. 2041}2045.
E$ciencies of sediment samplers for wind erosion measure- Stout, J.E., Zobeck, T.M., 1997. Intermittent saltation. Sedimen-
ment. Australian Journal of Soil Research 31, 519}532. tology 44, 959}970.
SkaK rby, L., 1977. Correlation of moss analysis to direct measure- Swineford, A., Frye, J.C., 1945. A mechanical analysis of wind-
ment by deposit gauge of cadmium, lead, and copper. blown dust compared with analyses of loess. American Jour-
Swedish Water and Air Pollution Research Laboratory, nal of Science 243, 249}255.
Publ. B377a, pp. 24. White, B.R., 1982. Two-phase measurements of saltating turbu-
Spaan, W.P., van Dijk, P., Hollemans W., Eppink, L.A.A.J., lent boundary layer #ow. International Journal of Multi-
1990. Wind erosion measurements on Schiermonnikoog, Re- phase Flow 5, 459}473.
port II. Department of Irrigation and Soil and Water Con- Wilson, S.J., Cooke, R.U., 1980. Wind erosion. In: Kirkby, M.J.,
servation, Wageningen Agricultural University, pp. 20. Morgan, R.P.C. (Eds.), Soil Erosion. Wiley, Chichester, pp.
Steen, B., 1979. Techniques for measuring dry deposition. Sum- 217}251.
mary of WMO Expert Meeting on Dry Deposition, April Yaalon, D.H., Ganor, E., 1979. East Mediterranean trajectories
18}22, 1977, Gothenburg. In: Morales, C. (Ed.), Saharan of dust-carrying storms from the Sahara and Sinai. In: Mo-
Dust. Mobilization, Transport, Deposition. Wiley, New rales, C. (Ed.), Saharan Dust. Mobilization, Transport, De-
York, pp. 287}289. position. Wiley, New York, pp. 187}193.
Steen, B., 1977. A new simple isokinetic sampler for the deter- Zeuner, F.E., 1949. Frost soils on Mount Kenya, and the rela-
mination of particle #ux. Atmospheric Environment 11, tion of frost soils to aeolian deposits. Journal of Soil Science
623}627. 1, 25}30.

You might also like