Efficiencies of Sediment Samplers for Wind Erosion Measurement

Transcription

1 Soil and Water Management and Conservation Aust. J. Soil Res., 1993, 31, Efficiencies of Sediment Samplers for Wind Erosion Measurement Y. Sh~o,~ G. H. ~c~ainsh,~ J. F. ~e~s' and M. R. ~ a u ~ a c h ~ A Centre for Environmental Mechanics, CSIRO, G.P.O. Box 821, A.C.T Faculty of Environmental Sciences, Griffiths University, Nathan, Qld Department of Conservation and Land Management, P.O. Box 7, Buronga, N.S.W Abstract We investigate the efficiency of three sediment samplers used for studies of wind erosion: a vertically integrating trap (an active, modified Bagnold trap for measuring vertically integrated streamwise sediment fluxes, designed for use in a portable wind tunnel); and two single-point, passive traps, the Leach trap (a small sampler of simple design, mainly for use in wind tunnels) and the Fryrear trap (a rugged sampler for field use). The vertically integrating trap is calibrated using a 'weighed sediment supply' technique in which a weighed sediment source is blown away completely during a calibration run. The single-point traps are calibrated against an accurate isokinetic sampler. The collection efficiency of all three traps is determined both in bulk and as a function of particle size. The results for overall efficiency E (for aeolian sand-sized particles) are: for the vertically integrating trap, E = ; for the Leach trap, E = 0.852~0.05 with a slight tendency to increase with wind speed; and for the Fryrear trap, E = (with or without a rain hood). Particle size analyses, carried out on the sediments collected by the traps under test and also the isokinetic sampler (assumed to have E = 1 for all particle sizes), show that the particle size distributions of the trapped sediments do not differ significantly from those of the isokinetic sampler. This unexpected result is a feature of the soils used for the tests, for which clay particles are mainly transported as small aggregates or clay skins upon sand grains. Keywords: wind erosion, wind erosion samplers, wind tunnel, soil particle size analysis, soil nutrients. Introduction There is a variety of samplers used to collect windblown particles, differing in dimension, aerodynamic behaviour and efficiency. Samplers can be classified as either active or passive: active samplers are equipped with pumping devices to maintain a flow through the intake of the sampler, while passive samplers rely on the ambient wind field to maintain the flow. Filters of fine mesh (<2 pm) can be used in active samplers, while in contrast filters of much coarser mesh (>40 pm) must be used in passive samplers to allow sufficient airflow through the sampler. As a result, active samplers are usually more accurate in collecting fine (clay, silt and dust) particles. However, since passive samplers are cheap and require no power supply, they are widely used in field experiments. A desirable property for any sampler is that it be isokinetic, meaning that the flow speed through the intake is the same as the local instantaneous ambient

2 Y. Shao et al. air speed. An isokinetic sampler does not distort the flow streamlines entering the sampler inlet, and thus accurately samples the streamwise flux of particles through an area equal to the sampler intake area normal to the wind direction. In particular, an isokinetic sampler does not discriminate between small particles which tend to follow curved streamlines, and large particles which tend to cross curved streamlines because of their inertia; therefore, the collected sediment is an unbiased sample of the particle size distribution in the incident flow. Few passive samplers are accurately isokinetic, and active samplers are not isokinetic unless the actively driven flow through the sampler is matched to an independently sensed ambient wind speed. This paper investigates the properties of three kinds of samplers, widely used in wind erosion studies in Australia and elsewhere, which we designate as vertically integrating (modified Bagnold) traps, Leach traps and Fryrear traps. The vertically integrating trap is an active sampler for measuring streamwise sediment fluxes, designed for use in a portable wind tunnel. This trap is substantially modified from the original, passive design of Bagnold (1941), by connection to a vacuum cleaner to make the trap active, but not isokinetic. This trap was used by Leys and Raupach (1991) and Shao et al. (1993) in a portable wind tunnel to measure streamwise soil fluxes. The Leach trap (White 1982) is a small, single-point, passive trap of simple design (Fig. lb). It is mainly used in wind tunnel studies, for instance by Rasmussen and Mikkelsen (1992), White and Mounla (1991) and Shao and Raupach (l992). The Fryrear trap (Fryrear 1986) is a rugged, passive, single-point trap for field use, especially for applications involving long exposure periods. It has been modified for use in Australia by the addition of a rain hood; in this form, the Fryrear trap is the source of almost all Australian field data on dust release and nutrient enrichment on the paddock scale. Characteristics of these three traps were investigated in the portable wind tunnel of the Department of Conservation and Land Management of New South Wales (Raupach and Leys 1990). Two methods for calibrating the traps were used. The point sampling (Fryrear and Leach) traps were calibrated against an accurate, active, isokinetic sampler, while the vertically integrating trap was calibrated by weighing the sediment supply before and after erosion (a technique suggested by J. Tubb, personal communication). We evaluated both the overall efficiency of each trap, integrated over all particle sizes for the aeolian test soils used, and also the efficiency as a function of particle size. In this paper, we first describe the characteristics of the traps and the experimental procedures, then present the observational data and discuss the efficiency of the traps. Trap Characteristics The traps investigated are illustrated in Fig. 1. The vertically integrating trap (Fig. la) has a narrow opening 5 mm wide and 0.48 m tall. Thus, the trap covers more than half the tunnel height and extends above the maximum depth of saltation layer (Shao and Raupach 1992). The sampler is connected to a dust filtering device and active suction is applied to maintain air flow through the filter. However, since the trap extends vertically through an approximately

3 Sediment Samplers for Wind Erosion Measurement Filler [',"cr pump 13.9mm diameter Fig. 1. The approximate dimensions and design of the traps investigated. (a) Vertically integrating (modified Bagnold) trap. (b) Leach trap. (c) Fryrear trap. (d) Isokinetic sampler. logarithmic wind profile, it is not possible in practice to ensure isokinetic flow through the trap. The Leach trap (Fig. 1 b) is wedge shaped with a 12' angle and a frontal area of 10x20 mm2. A wire mesh of 40 pm spacing is applied at the back of the trap which allows air to be drawn through the trap by the pressure difference between the front and rear faces, but inevitably this also allows some of the particles with diameter less than 40 pm to escape. A GO0 angle baffle is placed inside the collector to decrease the pressure build-up and prevent the wire mesh being damaged by particles with large speed. The Fryrear trap follows the original design of Fryrear (1986) but was modified by McTainsh to include a rain hood. This trap has a frontal intake area of 20x50 mm2 and a larger outlet area covered by a wire mesh of GO pm spacing. The dimensions of the trap itself can be found in Fryrear (1986), while the dimensions of the trap plus rain hood are shown in Fig. lc. Experimental Procedure The performance of the traps was tested in a portable wind tunnel with a working section 7 m long, 1.15 m wide and 0.9 m tall, and a maximum flow speed of approximately 15 m s-i. A turbulent boundary layer was generated by employing a tripping fence upstream of the working section. A full description, including aerodynamic characteristics, is given by Raupach and Leys (1990).

4 Y. Shao et al. During this experiment, the tunnel was set up on a concrete base, and a false plywood floor was installed in the working section to facilitate placement of the traps, especially the vertically integrating trap which has a large base, normally buried in soil. Each of the three traps was calibrated at three different wind speeds, and for each wind speed at least three replicates were made. For all calibrations, wind speeds were measured with three Pitot tubes at heights 50, 150 and 300 mm above the wooden floor at 4.2 m from the leading edge of the working section I a - Red soil --- Wh~te soil Fig. 2. (a) Cumulative particle size distributions (CPSD) of the red soil and white soil (undispersed). (b) Corresponding particle size distribution densities (PSDD). So that the results of trap calibration can be directly applied to available field observations, the main soil used as a sediment source for calibration was a red sandy soil (hereafter referred to as the red soil) from a sand dune near Balranald, N.S.W. The cumulative particle size distribution (minimal dispersion) of the red soil and the corresponding particle size distribution density (the derivative of the cumulative distribution) are given in Fig. 2, which shows that more than 60% (in mass) of the particles have a diameter between 125 and 300 pm, while about 25% of the particles have diameters less than 125 m and about 15% of particles are larger than 300 pm. This particle-size distribution is typical of the sand dunes in the mallee country of south-east Australia (south-west N.S.W., north-west Victoria and some areas in South Australia). Although most of the calibrations were made by using the red soil, some calibrations were made with a bricklayer's sand available in the Canberra area (henceforth called the white soil). As shown from its particle size distribution (Fig. 2), the white soil has a substantially higher fraction of particles smaller than 125 pm than the red soil.

5 Sediment Samplers for Wind Erosion Measurement For the vertically integrating trap, the technique used was to place a preweighed sediment supply in the tunnel, and to blow it all away. Before a run, a small amount (about 3 kg) of sediment was spread across the tunnel near the upstream end of the working section, with careful attention to lateral uniformity. If the total sediment supply is S g per lateral metre, and all this sediment is blown away, the mass collected in a 100% efficient vertically integrating sampler with lateral width Y would be SY. Hence the efficiency of the sampler is where m is the mass actually collected in the sampler. The particle size distribution of the supply and the collected sediment were determined to find E as a function of particle size (see below). For calibration of the point sampling (Leach and Fryrear) traps, the reference was an isokinetic sampler designed by Nickling and Gillies (1991). This consists of a sampling tube (internal diameter 13.9 mm) with its open intake pointing upstream, a flow meter, a filter, and a suction pump (Fig. Id). Fibreglass filters (0.6 pm) were used for the isokinetic trap, ensuring that effectively all particles passing through the intake were collected. Thus, the isokinetic trap provides an accurate reference not only for the total streamwise soil flux and thence the efficiency of any other trap, but also for the efficiency for individual particle size fractions. In these calibrations, the source of sediment was a section of the tunnel, 1 m long, 1 m wide and 6 m upstream of the test trap, covered with a test soil to a depth of 20 mm. The efficiency of the point sampling traps E is defined as where m is the mass of sediment collected, c is the sediment concentration, u is the wind speed, A is the frontal intake area of the sampler and T is the exposure time. If an isokinetic sampler (assumed to be 100% efficient) is exposed to the same flow as the trap under test, then c can be expressed as c = (m/uat)i, where the subscript i refers to the isokinetic sampler. Hence, for a given point sampling trap exposed with the isokinetic sampler at the same wind speed and the same time, The mass m can be the total mass collected, or the mass in some particle size range, which gives sampler efficiency as a function of particle size. To find the latter, the particle size distributions of the sediment collected in both the test sampler and the isokinetic sampler were determined as described below. The technique for particle-size analysis, which follows McTainsh et al. (1993), was developed especially for analysing small samples ( g in our case) of wind-eroded material. The analysis was conducted at iq5 intervals over the range pm (38 size classes), using a Coulter Multisizer (McTainsh et al. 1993) in the range 2-75 pm; the material with particle size less than 2 pm was measured by pipette and that with particle-size larger than 75 pm by dry-sieving (100 mm diameter Endecott sieves). The very small sample sizes (as low as 0.8 g) posed

6 Y. Shao et al. no problem for the Multisizer, but they pushed the pipette measurements of the dry fraction to the limit. Analyses were performed on minimally dispersed and dispersed samples. The minimally dispersed analyses received no pre-treatment, but as water is present some dispersion may occur. Pipette analysis was in triple-filtered (0.2 pm filters) ultra Hi-Pure water and Multisizer analysis was in an electrolyte of Isoton 11. Dispersion was by 10% tri-sodium orthophosphate (Nas Po4 12Hz 0 ) and 1 M sodium hydroxide (NaOH), with iron oxide (FeO) removal where necessary, using the method of Mehra and Jackson (1969). Overall Trap Efficiency By overall efficiency we mean the efficiency integrated over all particle sizes. Although this is the simplest measure of trap efficiency, its value depends on the particle size distribution, as discussed in the next section. Table 1. Efficiency of the vertically integrating trap for red soils S is sediment supply (g), m is the mass of sediment collected by the trap (g) and E is the trap efficiency (%) Run u = 11 ms-' u = 13 ms-' u = 12 ms-' S m E S m E S m E Mean s.d. Vertically Integrating Trap The results for the overall efficiency of the vertically integrating trap, as determined by equation (3)) are summarized in Table 1 where the wind speed u is measured at height 300 mm level. In general, the vertically integrating trap oversamples the tot a1 mass by a few percent. For each wind speed, the variation between the replicates is small. Within the wind velocity range used, trap efficiency does not clearly depend on the wind speed. Although we would expect oversampling to diminish as wind speed increases, such a trend is not obvious from the data. Averaged over all 18 runs, the overall efficiency of the trap is 102f5%. Leach Trap The Leach trap was calibrated against the isokinetic sampler at wind speeds of 9, 10.5 and 12 m s-'. During the observations, the Leach trap and the isokinetic trap were both mounted 100 mm from the floor surface, about 40 mm apart from edge to edge. The results are presented in Table 2. The overall efficiency of the Leach trap depends on wind speed. At a relatively low wind speed (9 m s-i ), the trap efficiency is 83f 1.5% compared with the

7 Sediment Samplers for Wind Erosion Measurement Table 2. Efficiency of Leach trap for red and white soils m is the mass of sediment collected by the Leach trap (g), mi is that collected by the isokinetic sampler (g) and E is the trap efficiency (%) Run u = 9 ms-' u = 10.5 ms-' u = 12 ms-' U= 10.5 ms-' Red soil Red soil Red soil White soil m mi E m m i E m m i E m m i E Mean s.d isokinetic sampler, but increases to 9Of 2% at a wind speed of 12 m s-l. Rasmussen and Mikkelsen (1992) examined the efficiency of the Aberdeen trap which is constructed from individual Leach traps in a wind tunnel experiment. For a similar range of wind speed as ours, they found that the efficiency of the Aberdeen trap is approximately 91% at heights above 35 mm and the efficiency decreases for lower levels, possibly due to smaller wind speeds at those levels. These results are in agreement with our observations. The three runs made with the white soil show that, at a wind speed of 10.5 m s-', the trap efficiency is 80&l% compared with the isokinetic sampler, in comparison with an efficiency of 86f 3% for red soil. It appears, therefore, that the efficiency of the Leach trap depends both on wind speed and the soil particle size distribution. Fryrear Trap Fryrear (1986) has already examined the overall efficiency of this trap (without rain hood) and reported an overall trap efficiency of approximately 90%. In his calibrations, Fryrear used a sediment source placed in a tray 48 mm upwind of the intake of the sampler. In this study, we re-examined the efficiency of the Fryrear trap by using the isokinetic sampler as the reference, and considered whether the rain hood would influence the trap efficiency. Table 3. Comparison between free stream flow speed ~(ms-') and flow speed through the Fryrear trap Utrap(ms- ) U Utrap Utrap/ U We first tested the dynamic properties of the Fryrear trap in the laboratory wind tunnel (Wooding 1968) of the Centre for Environmental Mechanics, CSIRO, by comparing the flow speed at the intake of the sampler and the undisturbed wind speed measured using a Pitot tube. As summarized in Table 3, the flow speed through the trap is approximately 5% lower than that of the undisturbed flow. Thus the Fryrear trap can be considered as approximately isokinetic.

8 Y. Shao et al. Table 4. Efficiency of Fryrear trap for red soil, when compared with the isokietic sampler placed at the top (T), middle (M) and botton (B) of the Fryrear trap m is the mass of sediment (g) collected by the Fryrear trap, mi is that collected by the isokinetic sampler (g) and E is the trap efficiency (%) Run U= 9 ms-' u = 10.5 ms-' u = 12 ms-' m mi E m mi E m mi E T3 Mean M1 M2 Mean B B B Mean One of the difficulties of using the isokinetic sampler (with an intake area of 152 mm2 ) to calibrate the Fryrear trap (which has an intake area of 1000 mm2 ) is that the streamwise particle flux density at the intake of the isokinetic sampler is not representative of the average over the intake area of the Fryrear trap. The data in Table 4 show that if the isokinetic sampler is mounted parallel to the top edge of the Fryrear trap (runs TI-T3), then the trap efficiency is l42f 12%, while if the isokinetic sampler is mounted parallel to the middle level (runs MI and M2) or bottom edge (runs B1 to B3), the trap efficiency is lo4f 3% and 86f 4%, respectively. This result is due to the fact that the large intake area of the Fryrear trap encompasses a region of logarithmic increase in wind speed but approximately exponential decrease in particle concentration (Shao and Raupach 1992). Because of this difficulty, the efficiency of the Fryrear trap can only be approximated as where?jf and qi are the streamwise sediment flux densities measured by the Fryrear trap and the isokinetic trap, averaged over the frontal area of the Fryrear trap. It follows from the measurements (Table 4) that where qib, qim and qit are the streamwise sediment flux densities at the bottom edge, in the middle and at the top edge of the Fryrear trap, as sensed by the

9 Sediment Samplers for Wind Erosion Measurement isokinetic sampler. The exact value of iji is not known, but can be approximated as a combination of the measurements made at the three different levels: qi = Cbqib + Cmqirn + Ct qit, where Cb, C, and Ct are weights to be specified. Since soil flux decreases exponentially with height, more significant weight has to be given to the measurements at the bottom edge of the Fryrear trap. A plausible factoring of the weights is Cb = 0.7, C, = 0.2 and Ct = 0.1, which results in a trap efficiency of 95%. This estimation is consistent with the 5% reduction in wind speed at the intake of the Fryrear trap in comparison with the flow speed (Table 3). We conclude therefore that the Fryrear trap has an overall sampling efficiency larger than 86% but less than 9576, which is comparable with Fryrear's (1986) estimate of 90% using a different approach. The efficiency of the Fryrear traps with or without a rain hood do not seem to differ significantly. Particle-size Aspects of Trap Efficiency Although clay and silt particles make up only a small mass proportion of many of Australia's wind erodible soils, they contain a large proportion of soil nutrients and act to increase soil moisture storage. Thus, knowledge of trap efficiency as a function of particle size is of particular importance if traps are being used to estimate nutrient loss in wind erosion events. The major concern of this section is to establish the trap efficiency for all particle-size groups by analysing the particle-size distribution of both the trapped sediments and the parent soil. It is, however, important to recognize that the particle-size distribution of a trapped sediment can differ from that of the parent soil for several reasons: (a) trap efficiency; (b) sorting during liftoff, due to possible preferential release of certain particle-size groups; (c) sorting during transport, with larger particles concentrated at lower levels while finer particles dominate at higher levels; and (d) mechanical abrasion, causing release of clay particles due to break down of clay aggregates or clay skins around sand grains during the process of saltation. Apart from the trap efficiency, the other aspects listed above will also be reflected in the particle-size analysis of the trapped sediment. The efficiency of the point samplers (Fryrear and Leach traps) is estimated by comparing them with the isokinetic sampler which, as discussed earlier, should have an efficiency of 100% for all particle size groups. As shown in Fig. 3, the particle size distributions of the sediments collected by these two traps are very similar to that of the sediment collected by the isokinetic sampler, except for the particle size range less than 10 pm. In this range, both point samplers (especially the Fryrear trap) are less efficient than the isokinetic sampler. In comparison with the isokinetic trap, the selective efficiency of the Leach trap for clay particles (<2 pm) is about 70% and that of the Fryrear trap is about 40%. A comparison between the sampling efficiencies of the vertically integrating trap and the isokinetic sampler for all particle sizes is difficult due to the effect of sorting during transport. Fig. 4 shows that the sediments collected by the vertically integrating trap have a slightly larger proportion of clay particles than the sediments collected by the isokinetic sampler. This probably reflects the

10 Y. Shao et al. Fig. 3. (a) Cumulative particle-size distributions (minimal dispersion) of the red sediments collected in the point-sampling traps: Fryrear trap (FT), Leach trap (LT) and isokinetic trap (IT). (b) Corresponding particle size distribution densities. sorting effect during transport: the isokinetic sampler was placed 100 mm above ground level, within the saltation sediment stream, and would therefore have collected samples enriched in saltation material, whereas the vertically integrating trap sampled from ground level up to 480 mm, collecting a more representative sample of the total sediment including the fine particles which disperse upward more rapidly. An alternative is to compare the particle-size distributions of the trapped sediment and the parent red soil. As can be seen from Fig. 4a, the sediments collected by the vertically integrating trap have a similar cumulative particle size distribution to the parent soil. In particular, this trap is the most efficient in the modal size range of the soil ( pm) which supplies most of the saltation particles. However, a different result is obtained for the very finest particles (< 10 m, including clays). Fig. 4b shows that sediments collected by the vertically integrating and isokinetic traps have a higher clay content than the parent soil (minimal dispersion), especially for the vertically integrating trap. This result may be related to abrasion during saltation: for the red soil used, clay particles may often be transported as small clay aggregates or as clay skins on sand grains which break down during saltation transport. Unfortunately, the exact amount of clay particles released by abrasion could not be determined in our experiment, so a definitive conclusion about the efficiency of the vertically integrating trap for very fine particles cannot be made. Although the quantity of clay released by abrasion could not be measured in this study, it is a process worthy of more careful scrutiny. A measure of the potential amount of clay particles which could be released by abrasion is

11 Sediment Samplers for Wind Erosion Measurement Fig. 4. (a) Cumulative particle-size distributions (minimal dispersion) of the parent red soil and the sediments collected in the vertically integrating (modified Bagnold) trap (BT) and the isokinetic traps (IT). (b) Corresponding particle size distribution densities. obtained from particle-size analysis performed on both dispersed and minimally dispersed samples. Fig. 5 shows the particle-size distribution of the red soil in a dispersed and minimally dispersed form. This soil has a minimally dispersed clay content of only 0.6% but a dispersed clay content of 6.96%. The significant increase in the clay proportion (<2 pm) in the dispersed sample, and decrease in the proportion of particles between 20 and 200 pm, also indicates that clays are probably carried as coarser aggregates or as clay skins upon the sand grains. The colour of the sand component of these soils also changes dramatically with dispersion. In their natural (minimally dispersed) condition, the colour (Munsell) of the pm (modal) sand fraction is 2-5YR4/4 (reddish brown), whereas after dispersion the soil colour changes to 10yr7/2 (light grey) as a result of the exposure of the white quartz grain surfaces after the removal of the red clay skins. This kind of change in colour has been frequently observed in south-west N.S.W. after wind erosion events. Fig. 5 also shows the particle size distribution of sediments collected by the Fryrear trap in a dispersed and minimally dispersed condition. Again, there is a significant increase in the clay fraction following dispersion. Dispersion has the same effect upon the colour of the trap sediments. The colour (Munsell) of the pm fraction of the sediment changes from 5YR4/6 (yellowish red) to 5YR.712 (pinkish grey). The inefficiency of the Fryrear trap in the 2-10 pm range is again apparent from a comparison of the minimally dispersed particle-size distributions of the trap and parent soil. In general, however, the particle-size collection efficiency of the Fryrear trap is quite high for the red soil. As Fig. 5

12 Y. Shao et al. Fig. 5. (a) Cumulative particle size distributions of the red soil in a dispersed and minimally dispersed condition, together with those of the red sediments of the Fryrear trap (FT) in a dispersed and minimally dispersed condition. (b) Corresponding particle size distribution densities. shows, most of the dispersed clays were held as silt and fine sand-sized aggregates (8-80 pm) when the soil was in its natural condition. These aggregated clays, plus the clays existing as clay skins upon sand grains, are more efficiently collected by non-isokinetic samplers than if the clays were being transported as fine particles (<2 pm). Conclusions In this study, we investigated the efficiency of three sediment traps for wind erosion measurements: an active, vertically integrating (modified Bagnold) trap and two passive, point-sampling traps (Leach and Fryrear). The Leach and Fryrear traps were calibrated against an accurate isokinetic trap; the vertically integrating trap was calibrated by a technique of weighing the sediment supply. We also investigated the particle-size collection efficiency of the traps by carrying out detailed particle size analyses. The following conclusions can be made: (i) For a wind speed around 10 m s-' (measured in the middle of the tunnel), the overall efficiency of the vertically integrating trap is 102&5%, for an aeolian sand. (ii) The overall efficiency of the Leach trap for a wind speed around 10 m s-i (same for the Fryrear trap) is around 85% for an aeolian sand, increasing with wind speed. The Leach trap has an efficiency of 70% for particles less than 10 pm. (iii) The overall efficiency of the Fryrear trap is 90155% for an aeolian sand. The Fryrear trap is less efficient for particles smaller than 100 pm and

13 Sediment Samplers for Wind Erosion Measurement 531 has an efficiency of 40% for particles less than 10 pm. For the red soil, the Fryrear trap is a quite useful sampler for field observations with an acceptable efficiency for all particles sizes. This is because, despite mechanical abrasion, a considerable fraction of the fine particles are transported in the form of small aggregates or clay skins on sand grains. (iv) Our evaluation of the fine particle collection efficiency of the traps is made less certain by two factors. Firstly, the combination of very low clay contents and very small samples meant that the pipette measurements of clays were at the limit of accuracy. Secondly, our assessments of trap efficiency as a function of particle size (especially for the vertically integrating trap) used a direct comparison of the particle size distribution for the parent soil and trapped sediment. This comparison is complicated by the aggregated nature of the red soil, which appears to change as a result of saltation impacts during transport, to an extent which could not be quantified in this study. (v) Our particle-size analysis shows that mechanical abrasion appears to be an important mechanism for the release of fine clay particles during saltation. In this study, the exact quantities of clay release due to abrasion could not be determined, but this is a process worthy of further scrutiny. Acknowledgments We are grateful to Dr W. G. Nickling for his constructive suggestions at an early stage of the project and for making his isokinetic sampler available. We also wish to thank Mr A. Lynch for his help with the particle-size analysis. This study was supported by a CSIRO/Griffith University Collaborative Research Grant. References Bagnold, R. A. (1941). 'The Physics of Blown Sand and Desert Dunes.' (Methuen: London.) Fryrear, D. W. (1986). A field dust sampler. J. Sod Water Cons. 41, Leys, J. F., and Raupach, M. R. (1991). Soil flux measurements using a portable wind erosion tunnel. Aust. J. Soil Res. 29, McTainsh, G. H., Lynch, A. W., and Hales, R. (1993). Particle-size analysis of soils, dusts and other very small samples using a coulter multisizer. J. Sediment. Petrol. (in press). Mehra, 0. P., and Jackson, M. L. (1969). Iron oxide removal from soil and clays by a dithionite-citrate system buffered with sodium bicarbonate. Clays Clay Miner Nickling, W. G., and Gillies, J. (1991). Sahelian aerosols (Mali). Final Tech. Rep. No. 3. Department of Geography, University of Guelph, Ontario, Canada. Rasmussen, K. R., and Mikkelsen, H. E. (1992). The aeolian transport rate profile and the efficiency of sand traps. Sedimentology (in press). Raupach, M. R., and Leys, J. F. (1990). Aerodynamics of a portable wind erosion tunnel for measuring soil erodibility by wind. Aust. J. Soil Res. 28, Shao, Y., and Raupach, M. R. (1992). The overshoot and Equilibrium of saltation. J. Geophys. Res. 97, Shao, Y., Raupach, M. R., and Findlater, P. A. (1993). The effect of saltation bombardment on the entrainment of dust by wind. J. Geophys. Res. (in press). White, B. R. (1982). Two-phase measurements of saltating turbulent boundary layer flow. Int. J. Multiphase Flow 5,

14 532 Y. Shao et al. White, B. R., and Mounla, H. (1991). An experimental study of froude number effect on wind-tunnel saltation. Acta Mechanica Suppl. 1, Wooding, R. A. (1968). A low-speed wind tunnel for model studies in micrometeorology. CSIRO Aust. Div. of Plant Industry Technical Paper No. 25. Manuscript received 3 March 1993, accepted 31 May 1993