Frequency, magnitude, and characteristics of aeolian sediment transport: McMurdo Dry Valleys, Antarctica

Authors


Abstract

[1] Due, in part, to the challenging environment of Earth's high-latitude regions, available information on cold climate effects on aeolian processes in these areas remains limited. Data from these areas, however, provide insight into the physics of sediment transport by wind and the controls on erosive winds in proximity to ice caps and topographic influences. This study presents a 2 year record of meteorological, saltation activity, horizontal saltation flux, and particle size distribution data from four sites in the McMurdo Dry Valleys of Antarctica, 2008 to 2010. Saltation measurements revealed daily and seasonal patterns with spring and summer sediment transport events occurring between 09:00 and 24:00 hours due to thermally generated winds. Fall and winter events occur at any time of day with the strongest associated with foehn winds. Threshold wind speed at 4.2 m in all seasons for all locations was ≈10 m s−1. Saltation occurred in the temperature range −40°C to +5°C. Westerly winds in the fall/winter and easterly winds in spring/summer are associated with the majority of transport events. The sand in transport is mainly 250 to 500 µm in diameter and poorly sorted. The integrated saltation flux varies over three orders of magnitude among the sites, with the lowest mean flux recorded in the Taylor Valley (2.9 kg m−1 day−1) and the highest in the eastern Victoria Valley (2271 kg m−1 day−1) for 24 hours of continuous saltation. The percentage of time saltation active at these locations annually is ≈2%, ≈4%, and ≈13%, respectively, for the Victoria, Taylor, and Wright Valleys.

1 Introduction

[2] There have been few studies of aeolian processes in high latitudes [e.g., Malin, 1984, 1985, 1986; McKenna Neuman, 1990; Seppälä, 2004] owing, in part, to difficulties in access and the challenging nature of the environment, especially for obtaining winter measurements under conditions of extreme cold, dryness, and darkness. Measurements outside of summer periods, for the most part, must rely on automated instruments as there are few opportunities to visit otherwise. Despite the difficulties of working in this environment, it is an important place to study sediment transport by wind as it affords the opportunity to evaluate the effects of higher-density and lower-viscosity air on particle transport and the vertical distribution of grain sizes in motion, which are conditions that are difficult to replicate in laboratory studies.

[3] Field-based studies in polar deserts, and those of Antarctica in particular, offer an opportunity to probe the physics of sand transport by wind at the extremes of the terrestrial environment and provide a terrestrial analogue to Mars. In addition, studies in areas proximal to large ice caps, such as the McMurdo Dry Valleys (MDV), can provide insight to aeolian processes that were more widespread during glaciations in proglacial and ice marginal settings, and the opportunity to investigate sediment-transporting winds influenced by the presence of large ice caps, other topographic controls, and seasonal changes in meteorology.

[4] Aeolian transport is known to be an active geomorphic agent within the Dry Valleys [Malin, 1984, 1985, 1986; Lancaster, 2002, 2004; Lancaster et al., 2010; Ayling and McGowan, 2006; Bourke et al., 2009; Gillies et al., 2009] as signified by the presence of well-developed aeolian landforms [e.g., Selby et al., 1973, 1974; Calkin and Rutford, 1974; Miotke, 1982; Gillies et al., 2009, 2012]. As well, it has become increasingly recognized that aeolian processes play an important role in the ecology of the MDV [Campbell et al., 1998; Moorhead and Priscu, 1998; Nkem et al., 2006]. Wind action is believed to be the main process redistributing organic matter to Dry Valley soils [Campbell et al., 1998; Moorhead and Priscu, 1998; Nkem et al., 2006]. The deposition of sediment onto and removal of organic mats from the surfaces of the ice-covered lakes in the Dry Valleys have been attributed to wind erosion processes [Andersen et al., 1993; Wharton et al., 1993; Fritsen and Priscu, 1999; Hendy et al., 2000; Sleewaegen et al., 2002]. Deposition of aeolian sediments may also affect the radiation and hydrological balance of the glaciers and lake ice by altering their albedo [Lewis et al., 1998]. Recent research by Atkins and Dunbar [2009] indicates that wind transported sediments contribute significantly to the sea ice on South McMurdo Sound, which is subsequently transferred through the ice to deposit on the seafloor.

[5] Despite the clear importance of aeolian processes to the ecosystem and landscapes of the MDV, there is little information for characterization and quantification of the flux of material transported by wind, its temporal patterns, or the spatial extent and current activity of aeolian landforms and sediment movement patterns. The purpose of this paper is to provide new information regarding the magnitude, frequency, seasonality, and grain size characteristics of aeolian sediment transport events in three of the MDV: Taylor, Victoria, and Wright (Figure 1). The characterization of the aeolian transport system in these valleys is based on 2 years of near-continuous, 10 minute average observations of wind speed and direction, temperature, and saltation activity at four locations within these valleys. Time/event-integrated, horizontal saltation flux measurements were made with custom-designed traps. These trap-derived fluxes allow characterization of average flux rates per unit of saltation time, as well as an estimation of the total annual horizontal flux. The temporal record of saltation activity is also linked to wind direction and temperature, providing information on the daily and seasonal patterns of aeolian transport. The sediments from the vertical traps were used to characterize the nature of the grains in transport and the vertical distribution of particle mass and diameter as a function of location and surface condition.

Figure 1.

Location of the McMurdo Dry Valleys.

2 Background

[6] Transport of sediment by wind and associated depositional landforms are widespread in high latitudes, especially in those areas with low rainfall [McKenna Neuman and Gilbert, 1986; McKenna Neuman, 1993; Seppälä, 2004]. Research also suggests that aeolian processes were even more extensive during past glacial periods [Koster and Dijkmans, 1988; Pye, 1987; Rea, 1990].

[7] Although the sedimentary characteristics of cold climate aeolian deposits have been documented extensively [see Koster, 1995, for review], the processes of erosion, transport, and deposition by wind in cold climates and their relations to boundary-layer winds and surface characteristics are poorly understood. Wind tunnel studies carried out by McKenna Neuman [2003, 2004] provide clear evidence that the low-temperature and -humidity conditions present in cold deserts could have considerable effect on the threshold of particle motion and transport capacity of the wind due to air density and viscosity effects, but this has not been corroborated with field measurements. The lack of information on the dynamics of aeolian transport processes in cold climate conditions hinders effective interpretations of past climates from Pleistocene cold climate aeolian deposits and landforms [Koster and Dijkmans, 1988]. Direct observations and measurements of sediment transport are rare, and less recent data are qualitative [McKenna Neuman, 1993].

[8] Strong winds that can transport sediment of sand and dust size (1000 µm to 2 µm) are recognized as an important feature of the MDV environment [Doran et al., 2002; Nylen et al., 2004; Lancaster et al., 2010]. Geomorphic evidence for aeolian processes is widespread throughout the MDV [Malin, 1992; Morris et al., 1972]. Ventifacts (rocks shaped by wind-driven abrasion) are frequently observed [Selby et al., 1973; Gillies et al., 2009], and the Victoria Valley has extensive areas of sand dunes and sand sheets [Calkin and Rutford, 1974; Speirs et al., 2008a; Bristow et al., 2010].

[9] The first quantitative information on rates of aeolian processes in the Dry Valleys was provided by the sand transport and abrasion experiments of Malin [1984, 1985, 1992]. Although these data are limited, they do indicate great interannual and spatial variability in rates of sand transport. Malin [1984, 1985, 1986] provided data on sand transport in the eastern Victoria Valley for several summer periods and two longer periods that included a winter component. Malin’s [1985] summer time-integrated sand transport data, collected with traps of unknown sampling efficiency, showed flux relationships similar to those observed in other environments. The trap data that included the winter periods [Malin, 1985, 1986] appear to have been compromised due to poor sampling efficiency at the inlets, blockage, and overfilling. These sampling problems place a high degree of uncertainty on the reported values. Malin’s [1984, 1985] data do not provide any information on the frequency of events during his monitoring periods, and interpretation of results are hampered by the lack of measured wind data. Since that time, Lancaster [2002] has provided a preliminary assessment of rates of sand and dust deposition in the region, which largely confirm Malin's original hypothesis that high fluxes of aeolian sand occur in areas characterized by abundant surface sands (e.g., Victoria Valley, Wright Valley, Lake Brownworth).

[10] Lancaster [2004] also documented relations between surface properties (e.g., rock cover and roughness geometry) and the potential for wind transport of sediment. Lancaster et al. [2010] report on sand transport events over a rock-strewn surface in the Victoria Valley during the Austral summer. Of the five events that they measured, all were generated by winds from the east to east-north-east, with threshold shear velocities varying between 0.30 and 0.35 m s−1, calculated using the time fraction equivalent method of Stout [2004]. The rates of horizontal sand transport for these five events ranged from ≈121 to ≈1201 kg m−1 day−1, which are based on the mass of sand caught in self-orienting, wedge-shaped sand traps based on the design of Nickling and McKenna Neuman [1997].

3 Methods

3.1 Study Sites

[11] The MDV are situated in the Trans-Antarctic Mountains, bounded by the Ross Sea/McMurdo Sound to the east and the East Antarctic Ice Sheet to the west. In the Taylor, Wright, and Victoria Valleys, there are small, perennially ice-covered terminal lakes that receive water and sediments mainly from glacier melt with some direct input from snow [Fountain et al., 2009] and by aeolian processes [Andersen et al., 1993; Fritsen and Priscu, 1999; Wharton et al., 1993; Hendy et al., 2000]. The MDV are in the precipitation shadow of the Trans-Antarctic Mountains and based on current records, annual precipitation does not exceed 50 mm in water equivalent [Fountain et al., 2009].

[12] The MDV are thought to have remained essentially ice-free for much of the last 13.6 million years [Sugden and Denton, 2004]. Surfaces in the valleys comprise a suite of forms including moraines, fluvio-glacial outwash fans, sand sheets, and dunes, which experience periglacial processes. Modification of the rocky surfaces (i.e., nonsand dominated) by aeolian abrasion is apparent in many valley locations [e.g., Gillies et al., 2009].

[13] Doran et al. [2002] used 14 years of data (1986–2000) from automated weather stations in the valleys to characterize their recent climatology. They report that temperatures typically hover near freezing in the summer months, with minimums as low as −65°C in winter. The annual average air temperature varies between −15°C and −30°C [Doran et al., 2002], with annual precipitation amounts ranging from <100 mm measured in the lower elevations of the valleys [Keys, 1980] to <50 mm based on more recent data [Fountain et al., 2009].

[14] The Lake Fryxell weather station data reviewed by Doran et al. [2002] also indicate that in the Taylor Valley, during their period of observation, onshore (easterly) winds dominate during the summer months with average speeds of approximately 4 m s−1 and with maximum speeds up to 15 m s−1. The southeasterly wind flow in the summer is thermally generated with a circulation that develops due to differential heating between the low albedo sediment and rock surfaces and the high albedo glacier and sea ice surfaces to the east of the valleys, analogous to sea-breeze circulations [McKendry and Lewthwaite, 1990, 1992].

[15] During the winter months, the Taylor Valley (Lake Fryxell) data indicate a dominant wind direction with a strong westerly component. According to Nylen et al. [2004], katabatic winds are an important meteorological feature of the Dry Valleys and create a bimodal wind direction pattern. The down-valley (westerly) winds are more frequent in the winter than in the summer and are also stronger [Nylen et al., 2004]. Observed mean wind speeds during winter wind events can reach 20 m s−1, with gusts exceeding 37 m s−1 [Nylen et al., 2004].

[16] The supposition of Nylen et al. [2004] that the high-speed winter winds are katabatic in nature has been disputed by Speirs et al. [2008a, 2010, 2012], who suggest that these events are actually foehn winds [Brinkmann, 1971; McKendry and Lewthwaite, 1990]. Foehn winds are similar to katabatic winds in that they are both gravitationally driven. Both types accelerate and warm as they descend from higher altitudes, but foehn are commonly caused by topographic modification of strong airflow. In the case of the MDV, the cause is the topographical interaction of synoptically forced airflow with the Trans-Antarctic Mountains, which causes mountain wave activity that leads to foehn wind genesis. In the Victoria Valley, Ayling and McGowan [2006] report that the infrequent but high-magnitude wind events are formed when a cold pool of air over Lake Vida is replaced by topographically modified foehn, and these winds play a significant role in aeolian and geomorphological processes.

[17] The locations of the four monitoring sites were: 77°36.477′S, 163°15.101′W (Taylor Valley) and 77°31.109′S, 161°51.045′W (Wright Valley), with two sites in the Victoria Valley at 77°22.345′S, 162°14.924′W and 77°22.919′S, 162°8.966′W. The eastern Victoria Valley site became inundated by streamflow from the Victoria Glacier just days before it was visited in 2009. As a result, its position was shifted to higher ground for the second year of monitoring (77°22.325′S, 162°14.866′W). The local physical settings of each site are shown in Figure 2. The Taylor Valley site (Figure 2a) is an undulating rock-strewn surface that does not appear to change on a yearly time scale, and the available sediment for wind transport appears limited. The Wright Valley site (Figure 2b) is centered within a field of wind-formed gravel bed forms [Gillies et al., 2012]. The Victoria Valley sites (Figures 2c and 2d) are sand-dominated and undulating surfaces covered with ripples that change visibly from year-to-year. There are also areas near the measurement sites where large rocks are present protruding from the sand, but they are sparsely distributed.

Figure 2.

The local physical setting of each measurement site: (a) Taylor Valley, (b) Wright Valley, (c) Victoria Valley eastern site, and (d) Victoria Valley western site. (e) Close-up image of the horizontal saltation trap with the receptacles covered. This location is the Taylor Valley site (note the rock-strewn surface and undulating topography).

3.2 Instrumentation

[18] At each measurement location, meteorological measurements of wind speed, wind direction, and temperature were collected. A 4.2 m high tower was erected that held a combination wind speed and direction propeller type anemometer (Model 05103, R.M. Young, Inc., Traverse City, MI) at the top of the tower and three cup anemometers (1900 #40C; NRG Systems, Hinesburg, VT) spaced logarithmically between 1.2 and 4.2 m (Figure 2). Anemometers were wind tunnel calibrated. Air temperature at 2 m above ground level was measured with a laboratory-calibrated, shielded thermistor (109-L; Campbell Scientific, Inc. Logan, UT). The output from each instrument was measured at 1 Hz, and the 10 minute average values were stored on a data logger (CR1000; Campbell Scientific, Inc.). The 10 minute averaging period was necessitated by the limits of data storage.

[19] At each of the measurement locations, rates of sediment transport were determined using two techniques, one active and one passive. Modified versions of the self-orienting passive sediment trap originally designed by Fryrear [1986] were deployed at the designated sampling locations. The Fryrear trap has been used extensively for time-integrated measurements of sand-size particles moving in saltation in harsh environments [e.g., Gillette et al., 1997a, 1997b; Gillette and Chen, 2001]. These passive collectors maintain a collection efficiency of ≈90% for a wide range of wind speeds [Shao et al., 1993]. For use in Antarctica, the traps (Figure 2e) were modified to withstand the high wind speeds and sediment transport rates that can be expected. At each site, two trap arrays were deployed. Each array had four catchers spaced logarithmically at heights between ≈0.2 and ≈1.3 m. The opening of each catcher is 0.02 × 0.05 m, and the total collection volume is approximately 0.0025 m3. Unlike other Fryrear [1986] trap arrays that have been used to collect saltation samples [e.g., Gillette and Pitchford, 2004], the four catchers in this trap design are not allowed to move independently of each other. They are connected through the tail fin assembly, which keeps them all facing the same direction and aligned into the wind.

[20] Overfilling of a trap would result in loss of information for estimating integrated sediment flux, as visiting the sites during the year was limited to the month of January. To address this problem, we constructed a trap-closing mechanism that is triggered by the filling of the lowest receptacle to a set capacity. Within the bottom catcher, an infrared light emitting diode (LED) and sensitive photocell approximately 0.2 m apart were set in place so that they were slightly lower than the plane defined by the bottom of the trap opening. Under the control of the data logger, the LED was turned on at the beginning of each hour for 10 minutes. If the infrared photocell recorded light during the duration of the illumination, then no action was taken. Within the dark interior of the trap receptacle, the photocell could easily discern the lit LED even in full sunlight, and upon burial the loss of signal was immediate. If the photoreceptor did not register the light from the LED for 6 consecutive hours, a signal was sent from the data logger to a linear actuator that lowered four covers in place in front of the trap openings. After closing trap one, a second signal was sent to the linear actuator on the second trap, which raised the covers on that trap, allowing collection to begin. The time of the closing/opening defined the end and beginning of the collection intervals for traps 1 and 2, respectively, and was recorded by the data logger. The second trap would close if its LED/photoreceptor pair became buried by the sediment collected in its bottom catcher, and that time would be recorded to define the end of the collection interval. The second trap at each site allows for two discrete sampling intervals, should the first trap fill to capacity.

[21] The mass of sand in the catchers and the period of duration of collection are used to interpolate the integrated horizontal mass flux using the empirical formula of Shao and Raupach [1992]:

display math(1)

where a, b, and c are constants and z is height above the ground. The total mass of transport was defined for a sampling interval for the ≈1.36 m layer represented by the top trap height (geometric mean of trap opening defines top height) as

display math(2)

where zi = i (m), and Δzi = 0.01 m. The dimensions of Q are kg m−1 in the defined collection period. The protocols of Ellis et al. [2009] were followed for characterizing aeolian mass flux profiles. The collected trap samples were dry sieved with a mechanical shaker at 1/2φ intervals from −4.5φ to 4.0φ (26.5 to 0.0625 mm) and weighed to 0.001 g precision.

[22] To measure localized saltation initiation and activity (frequency) at the selected sites two, Saltation Flux Impact Responders (Safires) were deployed. Baas [2004] evaluated the performance of these piezoelectric crystal type instruments and found that the Safire presents a minimal obstruction to the wind flow and provides high-frequency omnidirectional measurements at a relatively low cost compared with other piezoelectric type sensors. For deployment in Antarctica and to maintain operation through the austral winter, one of the Safires at each site was modified to increase its ruggedness. The modified Safire was encased in a stainless-steel sheath that is able to withstand high levels of abrasion. Performance evaluations of the regular and ruggedized Safires were carried out using a cold chamber that could contain a Safire for dynamic testing at temperatures of −50°C. A mechanism that delivered controlled impacts to the sensing area of the Safire was used to evaluate the sensor performance at ≈22°C and at the minimum of ≈−50°C. Results of these tests showed that for two different kinetic energy impact levels, there was no significant difference between the mV output of the sensor for the cold and warm tests. The sensors showed repeatability of the measurements, as each was tested several times at the two kinetic energy impact levels and showed no difference in their responses.

[23] Each of the Safires was tested in a sediment transport wind tunnel to compare their mV output with measurements of horizontal sediment flux (kg m−1 s−1) with two sediment traps placed on either of the instrument. The mean diameter of the test sand was 0.34 mm. The relationships between mV and mass flux were complex (i.e., nonlinear) and were expected to provide an estimate of the horizontal flux of sediment on an event basis at each sampling location. Having the Safires collocated with the passive traps allowed for the estimation of the total time the saltation system was active during the times the traps were open and provided information on saltation activity even if both traps eventually closed.

4 Results

[24] The recovery of the meteorological, saltation activity, and sediment collection using the traps was affected by four notable circumstances. In late December 2009, the eastern Victoria Valley site data logger was submerged when the meltwater stream of the Victoria glacier changed its course and flooded the data logger. The wind speed measurements for all but the top anemometer from the first season are suspect due to damaged files on the memory card due to its submersion. The wind speed and direction data, temperature, and saltation activity data appear to be sound as they show values similar to the western Victoria Valley site.

[25] The second data recovery problem occurred for wind direction measurements in the Taylor Valley in 2008 to 2009. At some indeterminate time (or multiple times), the tower rotated due to an anchor line slipping its mooring; thus, the wind direction data for that period are unreliable and could not be used. The third problem that occurred was that there was very little sediment collected in the trap at the western Victoria Valley site during 2009, although there was ample evidence that transport had occurred. We suspect that it was prevented from rotating for some unknown reason (possibly snow accumulation) for some unknown period of time. The trap was freely moving when we arrived at the site in December 2009, and the saltation activity data indicate activity at levels approaching those observed in the eastern part of the Valley, where large masses of sand were collected. The final problem lies with the Safire saltation sensors for estimating a mass flux of sand. The sensors did register the impacts of saltating grains that provided a means to establish when the saltation system was active. There were also periods when the sensors reached their maximum voltage output in response to the saltation, exceeding their capacity to measure the vigorous saltation that was occurring in response to high wind events. Unfortunately, these data do not allow for the conversion of the mV signal to a flux value using the wind tunnel calibrations with any great confidence, but they do provide a good indication of the frequency and duration of saltation activity.

4.1 Meteorological Data: Wind Speed, Direction, and Temperature

[26] Wind conditions measured at the four observation sites are summarized in Table 1. The mean wind speeds measured at 4.2 m for the four seasons at each site range between ≈1.4 (Victoria East, fall 2009) and ≈4.9 m s−1 (Victoria West, 2009, Wright, summer 2008/2009, fall 2008, winter 2008, spring 2008). Mean seasonal wind speed for a given site does not show great variation from year to year. The greatest changes occur at the western Victoria site where mean wind speed decreased by ≈3 m s−1 between fall and winter 2008 and fall and winter 2009. A reduction of ≈2 m s−1 in mean wind speed was observed in the Wright Valley for all seasons between 2008 and 2009.

Table 1. Wind Conditions at the Four Observation Sites, 2008–2009 and 2009–2010
 Wind Speed (m s−1) at 4.2 m
 Summer 2008/2009Fall 2008Winter 2008Spring 2008
SiteMin.Max.MeanMin.Max.MeanMin.Max.MeanMin.Max.Mean
Taylor010.763.38016.782.53024.743.00014.753.94
Victoria East011.323.87014.141.61022.322.41013.594.80
Victoria West011.734.65013.234.65021.924.65013.234.93
Wright017.394.86021.074.86022.944.86017.395.27
 Summer 2009/2010Fall 2009Winter 2009Spring 2009
Taylor012.083.04019.262.63019.502.78015.683.21
Victoria East011.584.00014.451.41017.472.58012.373.90
Victoria West011.924.14013.741.660.018.011.66013.424.15
Wright011.323.01018.542.650.018.042.76015.363.24

[27] For the monitoring period, maximum 10 minute mean wind speeds are observed in the winter (June 21 through September 20) at all four sites. In the fall and winter seasons, the ratio of maximum 10 minute mean wind speed to mean wind speed is approximately two times greater than for spring and summer, indicating that the colder fall and winter periods have more instances of much lower wind speeds than the spring and summer. Wind speed over 15 m s−1 measured at 4.2 m can be considered to be quite fast and indicates that each of these sites had periods of very high wind energy.

[28] The distribution of wind speed as a function of direction at each of the sites for the combined seasons of spring/summer and fall/winter show there are seasonal changes between the warmer and colder periods. Principally, at all locations, the frequency of occurrence of winds with easterly to northerly components dominate in the spring and summer periods, whereas during the fall and winter there is an increase in the frequency and magnitude of southwesterly winds.

[29] Summer high temperatures in 2008/2009 ranged from ≈6°C in the Victoria Valley to a ≈10°C in the Wright Valley. Winter lows in 2008 reached ≈−60°C in the Victoria Valley and a high of ≈−6°C in the Taylor Valley. Mean summer temperatures in 2008/2009 were in the range of ≈−9°C (Wright Valley) to ≈−13°C (Victoria East), with mean winter temperatures in the range of ≈−32°C (Wright Valley) to ≈−41°C (Victoria Valley). Summer highs in 2009/2010 ranged from ≈2°C in the Victoria Valley to ≈4°C in the Taylor Valley. Winter lows in 2009 reached ≈−60°C in the Victoria Valley and a high of ≈−2°C in the Taylor Valley. Mean summer temperatures were in the range of ≈−9°C (Taylor Valley) to ≈−14°C (Victoria East), with mean winter temperatures in the range of ≈−31°C (Taylor and Wright Valleys) to ≈−38°C (Victoria Valley).

4.2 Saltation Threshold

[30] The memory requirements for storing the meteorological and saltation activity data for a 1 year time period constrained the time resolution for all instruments to 10 minute mean values. This constraint unfortunately resulted in an inability to define the wind conditions at the commencement of sand transport events, which requires 1 second data to apply the time-fraction equivalence method of Stout [2004]. The data collected from the Safires can only be used to determine whether saltation activity occurred in any given 10 minute period. Information on the conditions that supported saltation activity at the four sites can be gleaned from the 10 minute wind speed and saltation activity. The average wind speed at 4.2 m for the 10 minute period preceding saltation, and during the first 10 minutes with saltation, is used to provide an indication of the wind speed needed for saltation to be initiated. This threshold wind speed proved to be remarkably similar for all locations and all seasons (Table 2). Saltation is observed when wind speed typically exceeds 10 m s−1 at 4.2 m above the surface, regardless of ambient temperature.

Table 2. Mean Wind Speed at 4.2 m When Saltation Threshold Is Reached
 Spring/SummerFall/Winter
SiteMean (m s−1)Standard Deviation (m s−1)Mean (m s−1)Standard Deviation (m s−1)
Taylor10.281.0310.51.03
Victoria East10.070.4310.651.24
Victoria West10.220.4410.451.31
Wright10.230.6311.762.64

4.3 Wind Profiles in the Presence of Saltation

[31] During the saltation process, the sand in motion extracts momentum from the wind that reduces wind speed not only within the saltation layer, but also above it. Above the saltation layer, the horizontal wind speed profile can be described by the “law of the wall” [Prandtl, 1935; Bagnold, 1941] using the apparent roughness of the moving saltation layer (i.e., z0′ [Bagnold, 1941]), which would allow estimation of the shear velocity (u*). An examination of wind profiles during periods of saltation at the monitoring sites indicates that they are not well explained by the modified Prandlt equation. In the presence of saltation, the vertical wind speed profiles (1.2–4.2 m) typically show a power function relationship, which likely results from the lowest anemometer being within the saltation layer, where momentum is being extracted from the wind by the large saltation fluxes driven by the high-speed, dense, cold winds. This, unfortunately, precludes the determination of u* under these conditions. To effectively calculate this important parameter would have required measurement of the vertical wind speed profile well above the top of the saltation layer, which in the environments measured appears to be quite deep (≊2 m). This is an important observation for this cold environment as it is typically assumed that in warm regions the saltation cloud height does not exceed 1 m.

4.4 Horizontal Mass Flux Measurements

[32] The vertical distribution of mass for particles in saltation changes among the four sites, but in each case is well described by the Shao and Raupach [1992] distribution of mass flux decreasing exponentially as a function of height (Figure 3). In Figure 3, flux as a function of height is plotted in a normalized fashion for comparison purposes. The mass flux is normalized (NSF) by dividing the flux at each measurement height by the flux associated with the lowest trap, and height is also normalized (NH) to the height of the bottom trap. In the Taylor (Figure 3a) and Wright Valleys (Figure 3b), a greater proportion of the sand is carried higher in the vertical profile than in the Victoria Valley. For the Victoria Valley sites, a greater proportion of the mass flux is closer to the ground (Figures 3c and 3d), with the lower Victoria Valley east site having the greatest decrease in horizontal saltation flux as a function of height. For the Victoria Valley sites, the surface is composed mainly of loose sand with only a disperse scattering of large, nonerodible rocks present.

Figure 3.

(a) Normalized horizontal saltation flux (QHn/QH1) as a function of normalized height (Hn/H1), for the Taylor Valley trap data for 2008–2009 (black square) and 2009–2010 (gray square). (b) Normalized horizontal saltation flux (QHn/QH1) as a function of normalized height (Hn/H1) for the Wright Valley trap data for 2008–2009 (black diamond) and 2009–2010 (trap 1: gray diamond; trap 2: white diamond). (c) Normalized horizontal saltation flux (QHn/QH1) as a function of normalized height (Hn/H1) for the Victoria Valley western site trap data for 2008–2009 (black circle), the event of 26 January 2009 (gray circle), and 2009–2010 (white circle). (d) Normalized horizontal saltation flux (QHn/QH1) as a function of normalized height (Hn/H1) for the Victoria Valley eastern site trap for 2008–2009 (trap 1: black triangle; trap 2: gray triangle), the event of 26 January 2009 (lighter gray triangle), and 2009–2010 (trap 1: lightest gray triangle; trap 2: white triangle). The Shao and Raupach [1992] empirical formula for the integrated horizontal mass flux (i.e., equation [(2)]) is fit to the data for each data set and shown as lines.

[33] The height-integrated horizontal mass flux measurements, calculated from the collected mass in each trap receptacle, and the application of equation ((2)) for each trap for the periods 2008 to 2009 and 2009 to 2010 are provided in Table 3. The horizontal flux values are presented in three ways. First, it presented as a mean daily trap flux (kg m−1 day−1), which is based on the mass of sediment collected during the total number of days the traps were open at each site. The second is the mean saltation day flux (kg m−1 day−1), which uses the total number of days the saltation system was recorded as active by the Safire. The third flux defines the total mass of sediment passing through the defined width (i.e., 1 m) for the total number of days the saltation system was observed to be active. As listed in Table 3, there are large differences in the horizontal flux of sediment in the four locations, spanning three orders of magnitude.

Table 3. Horizontal Saltation Flux Data for the Four Sitesa
SiteNo. of Sample Days (Trap)Daily Trap Flux (kg m−1 day−1)Saltation Days% Time Saltation is ActiveSaltation Day Flux (kg m−1 day−1)Total Saltation Day Flux (kg m−1 total saltation days−1)
  1. a

    Mean is based on the total number of days in a year the measurement system was operation, daily flux is the mean of traps 1 and 2, and time is the total number of saltation days based on the Safire data.

Taylor Valley, 2008–2009      
Trap 13440.2315.44.55.077.1
Taylor Valley, 2009–2010      
Trap 13650.0315.74.30.812.5
Victoria Valley East, 2008–2009      
Trap 15651.00.61.14,620.42,855.6
Trap 29410.90.60.61,736.41,025.0
Mean36530.96.82.03,178.421,549.5
Victoria Valley East, 26 January 2009      
Trap 111,978.50.112.515,827.81,978.5
Victoria Valley West, 2008–2009      
Trap 1 (blew over)      
Trap 23470.2410.53.080.7847.1
Victoria Valley West, 26 January 2009      
Trap 11178.20.216.71,068.9178.2
Victoria Valley East, 2009–2010      
Trap 17616.20.60.82,082.91,229.5
Trap 27111.71.217.3647.0781.8
Mean35263.97.92.21,365.010,714.9
Victoria Valley West, 2009–2010      
Trap 1 (malfunctioned)      
Trap 23512.47.12.0117.3838.3
Wright Valley, 2008–2009      
Trap 13470.734.59.97.1245.3
Wright Valley, 2009–2010      
Trap 12900.028.910.034.71,002.5
Trap 2750.07.19.5113.4804.9
Mean3650.035.613.174.12,636.4

[34] The Taylor Valley has the lowest flux values with increasing amounts observed in the Wright and Victoria Valleys. Two entries in Table 3 are for a specific storm that occurred on 26 January 2009 in the Victoria Valley that lasted for 3 to 4 hours. The calculated flux rates for this event are 178.2 and 1978.5 kg m−1 day−1 at the western and eastern valley sites, respectively. Because the team was on-site during this event, the sediment was collected from the open trap after the cessation of the transport event and the traps were reset.

4.5 Grain Size Characteristics of the Horizontal Saltation Flux

[35] The sediment collected in the saltation traps represents, for the most part, a mixture of transport events, whereas the event of 26 January 2009 represents a single event in the Victoria Valley. The particle size distribution data for the study are presented in Table 4. Most of the sand that was transported by the wind >0.24 m above the surface in the MDV is medium in grade (72%, 0.25–0.5 mm in diameter) and is, for the most part, poor to moderately sorted (72% of all trap compartments samples). The skewness in the trap samples is dominated by symmetric and coarse skewed distributions (66.6% of all samples), and the kurtosis is dominated by the mesokurtic form (61%).

Table 4. Particle Size Characteristics of the Sediment Trap Samples
LocationCollection Heighta (m)Mean (mm)Sorting (mm)SkewnessKurtosis
  1. a

    Collection height is the distance from the surface to the geometric mean of the individual collector orifice.

  2. b

    From the individual transport event of 26 January 2009.

Taylor Valley trap 10.2640.340.406.765.0
 0.5690.350.284.741.5
 0.8700.350.304.637.7
 1.3500.450.353.420.9
Wright Valley trap 10.2640.470.695.643.0
 0.5690.430.433.925.7
 0.8700.560.512.511.4
 1.3500.700.662.411.2
Wright Valley trap 20.2641.071.822.47.5
 0.5690.280.256.688.3
 0.8700.330.262.511.2
 1.3500.330.282.911.2
Victoria Valley eastern site trap 10.2620.280.100.94.9
 0.5620.310.140.23.3
 0.8630.160.132.06.3
 1.3430.110.085.437.0
Victoria Valley eastern site trap 20.2620.240.121.35.9
 0.5620.310.190.94.8
 0.8630.140.124.138.4
 1.3430.150.185.136.4
Victoria Valley western site trap 10.2640.350.262.812.7
 0.5690.460.331.55.3
 0.8700.490.341.14.2
 1.3500.540.370.62.8
Victoria Valley western site trap 20.2640.310.182.512.8
 0.5690.390.201.99.0
 0.8700.450.222.113.5
 1.3500.500.270.64.4
Victoria Valley eastern site trap 1b0.2640.290.111.05.2
 0.5690.330.140.33.7
 0.8700.250.180.82.5
 1.3500.150.123.417.3
Victoria Valley western site trap 1b0.2640.340.202.815.9
 0.5690.420.212.514.3
 0.8700.480.221.57.1
 1.3500.590.191.04.6

[36] Three patterns of particle size change as a function of height above the surface were observed for the four sampling locations. The first pattern, observed at the Victoria Valley western site and the Taylor Valley site, is the simplest and shows an increase in mean particle size with height. This pattern is observed for the longer-term trap data, as well as the individual transport event measured on 26 January 2009. Speirs et al. [2008b] report a similar pattern for trap data they collected in the Victoria Valley. The relationship between normalized mean particle diameter and normalized collection height for the Victoria Valley western site, as well as the data of Speirs et al. [2008b], is shown in Figure 4a. The relationship is well defined by a linear equation. The data from the Taylor Valley (Figure 4a), although also well explained by a linear fit to the normalized particle diameter and normalized height data, have a slope coefficient that is only ≈0.31 that of the Victoria Valley western site slope.

Figure 4.

(a) Normalized mean particle diameter (dHn/dH1) as a function of normalized height (Hn/H4) for the Victoria Valley western site trap data (black triangles), including the data of Speirs et al. [2008b] (white triangles). The same relationship for the Taylor Valley trap data (asterisks) is also presented. (b) Normalized mean particle diameter (dHn/dH4) as a function of normalized height (Hn/H4) for the Wright Valley site for traps 1 and 2. The least squares regression derived equation does not include the data from the lowest trap receptacle (white circle and white square). (c) Normalized mean particle diameter (dHn/dH4) as a function of normalized height (Hn/H4) for the Victoria Valley eastern site for traps 1 and 2, and the 26 January 2009 event. The least squares regression derived equation does not include the data from the lowest trap receptacle.

[37] For the Wright Valley samples, the pattern of increasing grain diameter with height is observed only for the top three collection heights (Figure 4b). The trap closest to the surface can receive a small amount of large particles (largest observed between 4 and 5.65 mm), which can shift the mean particle size to a value greater than that of the next higher receptacle. Above the first collector, the mean particle diameter increases linearly with height (Figure 4b), similar to the Victoria Valley western site and Taylor Valley data. For the Wright Valley data, the slope of this relationship (0.46) is closer to that observed for the Taylor Valley (0.30).

[38] The third pattern is observed at the Victoria Valley eastern site. At this site, an increase in mean particle size between the lowest collection level and the second collection level is observed, followed by a decrease in grain size as a function of height for the top three collectors (Figure 4c). The mean of the normalized particle diameter as a function of normalized height (excluding the mean grain size data from the lowest trap) for the three sets of trap data are better fit by a power function using least squares regression than a linear relationship (Figure 4c). At this location, as noted by Speirs et al. [2008b], the mean particle diameter of the surface sediments is smaller than those farther up valley with a mean of 0.43 mm (unfortunately, our bulk samples from this site were lost). The mean particle diameter for bulk samples removed from the vicinity of the traps was 1.5 mm, corresponding to very coarse sand as defined by Folk [1974].

4.6 Daily and Seasonal Saltation Patterns

[39] The pattern of daily and seasonal saltation activity at the monitoring locations can be gleaned from the Safire data, which records every 10 minute period in which saltation activity is observed. These data can be used to examine when saltation occurs most frequently as a function of the hour of the day, wind direction, and temperature. To evaluate seasonal changes in saltation activity, the data were grouped according to the following periods: spring/summer, defined as the period September 21 through March 21, and fall/winter, March 22 through September 20. Initial grouping of these data into four seasonal categories indicated there was little difference between the patterns observed for spring and summer and for fall and winter, so these data were aggregated into only two periods.

[40] There are distinctly different patterns to the saltation activity during a 24-hour period (midnight to midnight local time) between the spring/summer and the fall/winter periods, which are similar for the four measurement locations. In the spring/summer period, there is a cycle of sediment transport that is linked closely with the hour of the day (Figure 5). At each of the four sites, sediment transport intensifies most often during the afternoon to early evening hours. This is, of course, not related to changing light conditions, but to the position of the sun relative to the orientation of the valleys. For most sunny days in the summer, there is a transition from down-valley flow before noon to an up-valley flow with winds coming into the valleys from the direction of the Ross Sea due to valley heating effects. These up-valley flows are more frequently strong enough to initiate transport events than the down-valley flows.

Figure 5.

Saltation activity as a function of hour of the day in the spring/summer periods for the four measurement locations. White bars represent 2008/2009 data, and gray bars represent the 2009/2010 data.

[41] During fall and winter transport, events occur with almost equal probability for any hour of the day (Figure 6) at the four sites. In the absence of heating effects in the darkness of fall and winter, the winds that transport sediment are likely associated with regional pressure and large synoptic-scale weather patterns, which initiate foehn winds in the valleys [Speirs et al., 2010, 2012].

Figure 6.

Saltation activity as a function of hour of the day in the fall and winter periods for the four measurement locations. Black bars represent 2008 data, and dark gray bars represent 2009 data.

[42] The occurrence of winds that initiate and sustain saltation episodes in the valleys also shows dependence on temperature (Figure 7). In the spring/summer period, most transport occurs when temperatures are above −15°C. In the fall/winter, the saltation system is most active when temperatures are around −15°C. The average percentage of time the saltation system was active at temperatures below −15°C in the fall and winter periods was 30% in the Wright Valley, 32% in the lower Victoria Valley, 29% in the upper Victoria Valley, and 12% in the Taylor Valley. No saltation was recorded at any location at temperatures below −40°C with wind speeds at ≈4.2 m in all cases <10 m s−1 under this condition.

Figure 7.

Saltation activity as a function of temperature for spring/summer and fall/winter periods for the four measurement sites. Black bars represent fall/winter periods, and white bars represent spring/summer periods.

[43] As noted earlier, winds above threshold occur most frequently when air temperature is ≥−15°C. In the majority of cases in spring and summer, there is a 0 to +1°C h−1 warming (over the preceding 6 hours) prior to a saltation event occurring and is frequently associated with foehn winds (Figure 8). A smaller percentage of spring and summer saltation events is preceded by temperature increases of more than +2.5°C h−1. In a few instances, especially at the up-valley Victoria west site and Wright Valley locations, saltation events are preceded by periods of cooling (Figure 8).

Figure 8.

Rate of change of temperature before saltation events based on the previous 6 hours of measurements at the four sites by seasonal grouping.

[44] In the fall and winter months, more frequently higher rates of warming before the onset of winds exceeding the saltation threshold are observed, with rates up to +6.5°C h−1 (over the preceding 6 hours) (Figure 8). Similar to spring and summer, a warming of between 0°C and +0.5°C h−1 is observed frequently as a precursor to saltation; cooling prior to saltation (Figure 8) was also observed. During saltation events, the air temperature typically remains quite stable, with the majority of events warming or cooling less than 1°C h−1, regardless of the season.

4.7 Directional Transport Patterns

[45] As sediment transport was demonstrated to have daily patterns linked with the seasons and their associated winds, it follows that the direction of sediment transport will show seasonal dependence as well. The drift of sand as a function of direction and season can be evaluated by linking the percentage of time saltation was occurring, based on the Safire data, in a set time period with the wind direction during the same time. This provides a better representation of sand drift than an estimate based only on wind speed, as the Fryberger [1979] sand drift potential method uses.

[46] The seasonal pattern of sand direction movement for the Taylor Valley site for 2009 to 2010 is shown in Figure 9. The movement of sand is predominantly from the directional range of 202.5° to 225° in all seasons, accounting for 86% of all transport events. As Figure 9 shows, there is also a second, but much smaller, directional mode for 0° to 67.5°.

Figure 9.

Saltation activity as a function of direction for spring/summer and fall/winter periods for 2009–2010 at the Taylor Valley site.

[47] In the Victoria Valley, the bimodality of transport direction is more pronounced (Figures 10 and 11) than in the Taylor Valley. At the eastern location in the Victoria Valley in the 2008/2009 spring and summer transport from 90° through 112.5° dominated, which shifted to 67° through 90° in 2009/2010 (Figure 10). A small percentage of time was associated with westerly transport in both measurement years (17% in 2008/2009, 4% in 2009/2010). In the fall and winter, saltation occurs with a strong south to west component: 46% of the time in 2008 and 73% in 2009. There is saltation with an easterly component in the winter accounting for 41% in 2008 and 22% in 2009. At the western Victoria Valley site (Figure 11), spring and summer data show saltation with an east through south component dominating with 90% in 2008 and 70% in 2009, with only a limited saltation component with a south through west component: 10% in 2008 and 3% in 2009. In the fall and winter at this site, saltation from the south through west occurred 52% of the time in 2008 and 70% in 2009. In both years, saltation in winter also has a frequent (25%–38% of the time) east through south component.

Figure 10.

Saltation activity at the Victoria Valley eastern site as a function of direction for fall/winter and spring/summer periods for (a) 2008–2009 and (b) 2009–2010.

Figure 11.

Saltation activity at the Victoria Valley western site as a function of direction for fall/winter and spring/summer periods for (a) 2008–2009 and (b) 2009–2010.

[48] In the Wright Valley, the bimodal direction saltation system is dominated by sand-transporting winds from 157.5° through 247.5° throughout an annual cycle (Figure 12), with transport in an easterly direction (67.5° through 112.5°) reaching a significant proportion of the transport events only during the spring and summer period (≈35% to ≈42%).

Figure 12.

Saltation activity at the Wright Valley site as a function of direction for fall/winter and spring/summer periods for (a) 2008–2009 and (b) 2009–2010.

5 Discussion

5.1 Meteorological Data: Wind Speed, Direction, and Temperature

[49] The meteorological data collected during the 2 years of this study corroborate the results presented by Doran et al. [2002], suggesting that the environmental conditions in the MDV from 1986 through 2010 have remained relatively stable. The observed mean annual and seasonal mean temperatures are in agreement with the range of values reported by Doran et al. [2002]. Maximum wind speeds reported by Doran et al. [2002] exceed 30 m s−1, whereas at our four sites, the maximum speeds measured at ≈4 m approached, but did not exceed, 25 m s−1. Mean annual wind speeds at comparable locations are of similar magnitude. As noted by others, our data demonstrate the wind direction in the valleys is largely bimodal, principally the up- and down-valley directions. At the Taylor Valley site, wind direction was northeast and southwest, which is not up and down valley, but may reflect local topographicsteering of wind or a lake effect due to its proximity to Lake Fryxell.

5.2 Saltation Threshold

[50] As the wind speed data at 4.2 m indicate (Table 2), the entrainment threshold values for all sites were very similar. McKenna Neuman [2003] demonstrated that threshold shear velocity (u*t) can be lowered with decreasing temperatures due to the increased density and reduced interparticle cohesion forces due to the accompanying low humidity. Over the temperature range −15°C to 5°C, when saltation occurred most frequently in the MDV, air density changes from ≈1.372 to ≈1.273 kg m−3, ≈7%. At the lowest observed temperature when saltation was observed, ≈−40°C, air density increases to 1.519 kg m−3, a change of 9.7% (from −15°C to −40°C). However, we were unable to resolve a temperature effect on saltation threshold at our sites. The similarity in threshold wind speed among sites is consistent with the observation that the majority of the sand in transport is of a similar size, i.e., medium grade (0.2–0.5 mm diameter); hence, it should be mobilized under approximately the same wind shear stress range.

5.3 Horizontal Mass Flux Measurements

[51] Limited data are available from long-term studies that characterize annual saltation flux in that most measurements of sand flux are made during short-term, process-based studies. A longer-term record of sand flux in the Chihuahuan Desert of southern New Mexico is reported by Gillette and Pitchford [2004] and Bergametti and Gillette [2010]. At their measurement locations, the flux ranged from 0.001 to 0.95 kg m−1 day−1 and varied dramatically by site. For these New Mexico measurements, the record of time that saltation was active during the time of flux measurement was not recorded, and the averaging period varied from ≈4 to ≈7.6 years. The New Mexico flux rates are similar to the daily flux rates measured in the Taylor Valley, and the maximum rate of 0.95 kg m−1 day−1 for New Mexico just exceeds the saltation day flux of 0.8 kg m−1 day−1 for the Taylor Valley in the 2009 to 2010 monitoring period. By comparison, the Wright and Victoria Valleys have much greater daily flux rates than any recorded in the New Mexico studies [Gillette and Pitchford, 2004; Bergametti and Gillette, 2010].

[52] McKenna Neuman [1990] measured sediment transport rates using Leatherman traps [Leatherman, 1978] that were deployed for approximately 1 year on a sand sheet on Baffin Island in the Canadian Arctic. Although a direct comparison of transport rates between McKenna Neuman’s [1990] and those from this study is not possible due to the lower sampling efficiency of the Leatherman type traps and different units of flux, her data also indicate that rates in the Arctic, as a function of event, by season, and annually, are also very high and likely under-estimated by her methods.

[53] For the Victoria Valley, the flux rates for short-term process studies reported in the literature are similar to those measured during the storm event of 26 January 2009. For data that include measurements at all four trap heights made by Speirs et al. [2008b, Table 2] on the stoss and crest of dunes of the Victoria Valley, the flux can be estimated using equation ((2)) and ranges from ≈52 to ≈440 kg m−1 day−1 (assuming a trap opening of 10 cm2 [Fryrear, 1986]). Lancaster et al. [2010] report flux rates on the floor of the Victoria Valley, further down valley (east) of our sites ranging from ≈121 to ≈1201 kg m−1 day−1 for events that lasted between 576 and 1320 minutes. Overall, the event-based measurements of daily flux for sand-dominated surfaces reported in this study indicate a similar range. Comparison of annual flux rates based on the amount of time in a year that the saltation system is active for other areas remains limited, but clearly the data indicate that quite large amounts of sand are moved annually in the Victoria and Wright Valleys.

[54] The vertical saltation flux profiles measured at the MDV sites are well described by an exponential function [Shao and Raupach, 1992]. The exponential form of the vertical saltation mass flux relationship has been observed by others as well [e.g., Bagnold, 1941; Chepil, 1945; Sørensen, 1985; Williams, 1964]; however, there is no general agreement that this holds for all vertical flux profiles. Vertical mass flux profiles have also been observed to deviate from the exponential form [e.g., Anderson and Haff, 1991; McEwan and Willetts, 1991; Butterfield, 1999]. Comparable field data using traps of similar design and efficiency have presented vertical flux profiles that are well described by the exponential form [e.g., Namikas, 2003], although Stout and Zobeck’s [1996] measurements were presented as having the form of a power function. The literature appears to indicate a consensus of support for the exponential form to explain the vertical characteristics of the flux profile.

[55] The characterization of the vertical flux profile is critical for developing accurate numerical modeling of the saltation process mass transport rates. The vertical profile of mass flux represents the integration of grain trajectories, distribution of ejection angles, and speeds of grains splashed by particle impacts. Clearly, the MDV trap data (Figure 3) show that the normalized vertical flux profiles, although exponential in form in all cases, vary considerably between sites with respect to the rate of change of sediment flux with height from the surface. As the particle size range of the sand in transport for all sites is similar (poor to moderately well-sorted sand, 0.25–0.5 mm in diameter), as is the wind speed range during saltation, the differences in the mass flux profiles are likely caused by the constitution of the surfaces, principally their roughness and hardness. These properties affect the coefficient of restitution (ε) of particles rebounding from the surface, which is defined as the ratio of postcollision particle speed to precollision particle speed [Gordon and McKenna Neuman, 2009]. At the Victoria Valley sites, the surface is unconsolidated sand with coarser sands observed at the western site. Gordon and McKenna Neuman [2009] have shown that loose beds absorb more momentum and energy from impacting sand particles, which tends to lower the average ejection angle. For the Victoria Valley sites, 50% of the integrated horizontal flux (i.e., NSF = 0.5) moves below NH of ≈0.9 for the eastern site and NH ≈1.9 for the western site. That a greater proportion of the horizontal saltation flux is carried higher above the surface at the Victoria Valley western site may be, in part, due to the coarser grain size, which provides more opportunity for particles in saltation to strike larger grains on the surface and rebound with a higher ε. In the Wright and Taylor Valleys, the surfaces are much harder, or consolidated, resulting in less particle momentum being absorbed at impact. At these two sites a greater proportion of the mass flux moves much higher above the surface than at the two sandy sites. In the case of the Wright Valley, where the measurement site is covered by particles of diameter 2 to 16 mm [Gillies et al., 2012], 50% of the mass in horizontal transport moves below NH ≈2.4. At the Taylor Valley site, 50% of the mass moves below NH ≈2.5; this may be due, in part, to the presence of larger rocks and gravel particles. These components of the surface sediments provide hard areas for saltating particles to rebound from. The MDV can provide a unique opportunity to study the saltation process as it is affected by environmental conditions, if the logistical constraints on deploying instrumentation to measure in real time (at least 1 Hz) can be overcome.

5.4 Grain Size Characteristics of the Horizontal Saltation Flux

[56] As described earlier, the sand in the saltation traps is dominated by the medium grade, and this remains stable for each of the four sites that were monitored. The description of the vertical profile of the mean grain sizes in saltation in the MDV, unlike the mass flux that generally followed an exponential form, varies quite markedly between the measurement sites. The observed patterns of changing mean grain size as a function of height above the surface (Figure 4) follow closely those reported by Williams [1964]. In Williams’s [1964] wind tunnel study, he observed that mean grain diameter decreased rapidly within 0.05 m above the bed, then decreased more slowly (or remained relatively constant) between 0.05 and 0.16 m, and for a large number of his tests, mean grain size then began to increase with increasing height. It is likely that because of our trap design, which limited the height at which the bottom receptacle could be placed (>0.24 m), the data from the Victoria west and Taylor Valley sites represent collection above the inflection point described by Williams [1964] where mean grain size begins to increase with height. The lowest trap receptacles for the Victoria east and Wright Valley sites appear to be influenced by saltation dynamics that are in some way more complex than the Victoria west and Taylor Valley sites.

[57] The characteristics of the particle ejection process and the resultant trajectories between the surface and ≈0.3 m are leading to two very different responses in mean grain size between 0.24 and 0.29 m between sites. In the Wright Valley, there is the possibility that a few very coarse particles could enter the bottom trap receptacle, dramatically increasing the mean grain diameter. At the Victoria Valley east site, mean grain diameter within approximately the same height zone is less that that moving between 0.49 and 0.54 m, and unlike the other sites, the mean grain diameter decreases with height between 0.54 and 1.37 m, which follows some of Williams’s [1964] tests, as well as the results of Li et al. [2008], who report an exponential decrease of particle diameter with height for a wind tunnel study.

[58] The different patterns of mean grain size as a function of height above the surface, and the notable change in the patterns for the Victoria Valley east and Wright Valley sites, suggest that there may be two well-defined species of grain in transport, which correspond to what Andreotti [2004] called saltons (high-energy grains) and reptons (grains ejected from the sand bed by the impact of saltons); this may, in part, be affecting the grain size distribution below 0.24 m. We suggest that in the natural setting of an unconstrained boundary and saltation layer in the MDV, the length dimensions of the particles in the phases of transport may be much greater than has been observed in wind tunnel studies. This is supported by other researchers examining saltation processes and aerodynamic effects in wind tunnels [e.g., Sherman and Farrell, 2008] and by field studies such as that by McGowan and Sturman [1997]. They measured aeolian transport up to 4 m above ground level over a coarse-grained bed in an alpine environment with high wind speeds (>30 m s−1) and also observed that the trajectory height of saltating particles exceeded 2 m, similar to what we observed in the MDV. Their mean grain size data are only for heights ≥0.5 m, which show a nonlinear decrease with height. Explaining how the different grain size versus height relationships observed in field and wind tunnel studies arise remains a research challenge. As noted earlier, the MDV provide an ideal location for examining the saltation system across a wide variety of conditions, but the logistics of measurement are formidable.

5.5 Daily and Seasonal Saltation Patterns

[59] Transport patterns of sediment by wind have not been well documented in many areas using real-time instrumentation, especially in remote areas where logistical support is difficult. The high-latitude location of the MDV creates very different seasonal variations in the interaction between solar heating, thermal instability, atmospheric turbulence, and wind strength compared to midlatitude locations where aeolian sediment transport occurs. In the spring and summer months, the daily pattern of aeolian transport at the MDV (Figure 5) is similar to the pattern observed by Stout [2010] in Texas, with both having maxima associated with a time of day. In both locations, there is an increase in saltation activity after 09:00, and in the case of the MDV this lasts through to 24:00, whereas in Texas, activity subsides after 18:00 coinciding with a decrease in solar radiation. In both locations, saltation can occur at any time of the day, but in the MDV during the fall and winter, the diurnal pattern is completely lost (Figure 6).

[60] The meteorology that potentially drives MDV saltation events changes seasonally. In the winter, high wind speeds have been attributed to katabatic [Doran et al., 2002; Nylen et al., 2004] and foehn winds [McKendry and Lewthwaite, 1990; Speirs et al., 2008a, 2010, 2012]. In the spring and summer, the thermally induced up-valley and the westerly down-valley flow, as described by McKendry and Lewthwaite [1992], can reach speeds sufficient to transport sediment. In these cases, there is minimal warming (0 to +1°C h−1 over the preceding 6 hours) prior to a saltation event occurring (Figure 8). Some saltation events in spring and summer are preceded by temperature increases of >2.5°C h−1, which could indicate that gravity-generated winds are responsible for the transport. In a few instances, especially at the up-valley Victoria Valley and Wright Valley locations, some saltation events are preceded by periods of cooling (Figure 8).

[61] In the fall and winter months, more frequently greater rates of warming before the onset of saltation are observed, with rates up to +6.5°C h−1 (over the preceding 6 hours) (Figure 8). Delta change in temperature of ≈+25°C to ≈+50°C between minimum and maximum air temperatures preceding and during high wind events were considered by Speirs et al. [2010] to be, in part, diagnostic of foehn winds. Changes of temperature in this range prior to saltation events were observed at the four MDV sites in the fall and winter periods. Similar to spring and summer, a warming of 0 to +0.5°C h−1 was observed frequently as a precursor to saltation, as was a cooling prior to saltation, but this occurred less frequently (Figure 8), suggesting meteorological events besides gravity-induced winds can also initiate and sustain sediment transport in the fall and winter.

[62] As saltation is strongly coupled to wind speeds >10 m s−1, we make use of the study of Speirs et al. [2012] to evaluate how representative the measured horizontal saltation flux values may be for this study. According to Speirs et al. [2012], 91% of days with foehn winds have mean wind speed >10 m s−1, suggesting that for sand transport events, the variability in days with winds >10 m s−1 will be a major factor affecting the annual magnitude of the sand flux. Speirs et al. [2012] provide a characterization of the annual distribution of foehn winds (their Figure 5) in the Taylor, Wright, and Victoria Valleys, and relationships between mean seasonal air temperature and the percentage of foehn days per season (their Figure 6). This information can be used to evaluate how the conditions that prevailed during our study compare to a longer 12 year record.

[63] Using seasonal mean temperature data, we can estimate foehn wind frequency at our sites and compare this to the longer data record (1996–2008) [Speirs et al., 2012]. For 2008 to 2009 for the Taylor Valley, the Speirs et al. [2012] relationships indicate that ≈20% foehn days could be expected in the fall months (March, April, and May) and 32% for the winter months (June, July, and August). The 12 year record suggests in the Taylor Valley that the range of mean values is ≈25% to ≈40% and ≈30% to ≈40%, respectively, for the fall and winter months.

[64] In the Wright Valley, the percentage of foehn days as predicted from the Speirs et al. [2012] relationship for the fall months (mean seasonal temperature ≈−29°C) is ≈20%, which is close to the mean range of ≈18% to ≈22%. The same is true for the winter months in 2008 to 2009 (≈33%) compared to a range of ≈30% to ≈35% shown by Speirs et al. [2012]. In the Victoria Valley, our mean temperature data indicate that there should be a greater percentage of foehn days in this valley in the fall (≈20% compared with the mean monthly range of ≈5%–≈10%) and winter (≈33% compared with the mean monthly range ≈15%), suggesting the frequency of sand-transporting winds may have been more frequent in the Victoria Valley as compared to the mean conditions. This, by no means comprehensive, evaluation of mean seasonal temperatures and percentage frequency of foehn days provides some confidence that the conditions conducive to sand transport in the fall and winter months during our study were not at the extreme ends of the variability range associated with the percentage frequency of foehn days. As the wind conditions approximate the mean seasonal foehn frequencies defined by Speirs et al. [2012] for fall and winter periods, it suggests that the observed range of horizontal saltation flux also represents more typical conditions (i.e., nonextreme years). In the spring and summer months, our mean temperatures and wind speeds are close to the mean values reported by Doran et al. [2002], which suggest that the fluxes reported for these seasons in this article are typical of the prevailing conditions.

[65] Due to differences in supply of sand between the three valleys studied, the relative amount of sand moving should remain consistent as this is more closely linked to sediment supply rather than wind energy. Unless new and large sand sources are uncovered in the Taylor Valley and Wright Valleys, or the input of sediment brought into the Wright by the Onyx River increases dramatically, the relative amounts of sand movement among the valleys studied should remain similar to that observed in this study.

[66] The role of snow in affecting aeolian transport in these three Dry Valleys is likely minimal. As reported by Fountain et al. [2009], precipitation inputs in the form of snow decrease inland with a relatively strong gradient with water equivalent (WEQ) values in the Victoria and Wright Valleys near 40 and 20 mm, respectively. The Taylor Valley site is around 30 mm WEQ [Fountain et al., 2009]. The extreme dryness of the MDV environment and foehn wind events combine to sublimate and erode the snow cover, making its long persistence unlikely, and thus remove its capacity to protect the surface from erosive winds.

6 Conclusions

[67] The aeolian sediment transport system in the MDV has been demonstrated to be very dynamic with daily and seasonal patterns of transport that are driven by winds that reflect seasonal changes in meteorology. Transport events in the spring and summer are caused primarily by valley heating effects, and in the fall and winter by gravity-driven warming winds that bear the signature of foehn winds as described by Speirs et al. [2008a]. In the winter, long periods of low wind speeds and stable very cold conditions are punctuated by dramatic warming and rapid increase in wind speed, which can lead to periods of saltation lasting up to several days with only minutes to hours of quiescent periods during the event. In all seasons and at any time of day there is clear evidence that winds can reach speeds sufficient for sediment transport in the absence of any strong thermal signature in the air temperature, and transport can even occur following a cooling trend. The meteorological conditions that give rise to these events remain to be resolved, but the event periods when this occurred can be identified for interested parties.

[68] The mass flux of sand-sized sediment being moved in the MDV studied during this research period, and by extension over long periods of time assuming similar conditions have prevailed, is substantial. The magnitude of the flux in each valley is linked to the available sand supply. The pattern of movement, especially for observations in the Wright and Victoria Valleys, indicate that the sand is moved on a seasonal basis up and down the valleys, with the majority of sediment moved during fall and winter periods. The observation of significant transport in the Victoria Valley in the fall and winter periods refutes the observation of Speirs et al. [2008a] that aeolian transport is largely confined to the period of November to February. Based on data from Speirs et al. [2012] of the frequency of foehn winds in the fall and winter, and the spring and summer mean wind speed records, the reported saltation fluxes and the patterns of aeolian sediment transport are typical of the environmental conditions that have prevailed for the period 1986 to 2010.

[69] The measurements of mass flux and mean particle diameter as a function of height obtained by this research indicate that unconstrained aeolian sediment transport in a cold, dry environment has very different characteristics from those obtained in wind tunnel testing. It appears from field testing, including these new data from the MDV, that the vertical and horizontal length scales of particles in saltation and being splashed into saltation and reptation are much greater than is exhibited in wind tunnels and in models of saltation that are based on reconciling modeled behavior with wind tunnel measurements. The relative roles of fluid air density, viscosity, and surface characteristics (e.g., hardness/consolidation, grain size distribution) in affecting the vertical profiles of the horizontal saltation flux remain to be elucidated. The MDV offer an exceptional opportunity to study fundamental aeolian processes for a wide variety of environmental conditions both in the boundary and saltation layer, as well as a range of surface conditions (e.g., loose sand, highly elastic bounding surfaces); the access to these sites is feasible due to the support offered by the countries participating in science research in the MDV and Antarctica in general, but collecting appropriate measurements at the time scales of interest remains a scientific and instrumentation challenge.

Notation
d

particle diameter (mm)

dH1

mean particle diameter at height of first trap receptacle (mm)

dHn

mean particle diameter at height of trap receptacle number n (mm)

H1

height of first trap receptacle (m)

Hn

height of trap receptacle number n (m)

LED

light-emitting diode

MDV

McMurdo Dry Valleys

NH

normalized height

NSF

normalized saltation flux

Q

vertically integrated horizontal saltation flux (kg m−1 t−1)

QH1

horizontal saltation flux (kg m−1 t−1) at height of first trap receptacle

QHn

horizontal saltation flux (kg m−1 t−1) at height of trap receptacle number n

R2

correlation coefficient

u*

wind shear velocity (m s−1)

z

height above the surface (m)

z0

apparent aerodynamic roughness height in the presence of saltation (m)

Acknowledgments

[70] This research was supported by the U.S. National Science Foundation's Office of Polar Programs Grant ANT-063621 to J. A. Gillies. W. G. Nickling acknowledges support from the National Sciences and Engineering Council, Canada. We are also grateful for the outstanding logistical and technical support provided by the U.S. Antarctic Program. We also thank Becky Peace (BFC, USAP) for her invaluable assistance in the field, and the USAP volunteers who joined us at various times. Finally, we thank Mario Finoro and Sandy McClaren (Dept. of Geography, University of Guelph) for their outstanding technical assistance in helping to create the instruments that withstood the Antarctic weather.

Ancillary