Geomorphology 105 (2009) 59–66
Contents lists available at ScienceDirect
Geomorphology j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / g e o m o r p h
Abrasion control on dune colour: Muleshoe Dunes, SW USA Kevin White a,⁎, Joanna Bullard b a b
Department of Geography, University of Reading, Whiteknights, Reading, RG6 6AB, UK Department of Geography, Loughborough University, Loughborough, Leicestershire, LE11 3TU, UK
A R T I C L E
I N F O
Article history: Accepted 30 January 2008 Available online 14 June 2008 Keywords: Aeolian processes Remote sensing Dune colour Muleshoe Dunes
A B S T R A C T The Muleshoe Dunes, an east–west trending dunefield on the border separating Texas and New Mexico, consist of two distinct components: a white (carbonate rich) component and an overlying pink (quartz rich) component. The pink component exhibits significant spatial variation in redness. The reddest sands, in the western part of the dunefield, decrease in redness towards the east. This gradient is thought to result from abrasion of an iron-rich, red clay coating as the sediments were transported eastward by Late Quaternary aeolian processes. The effects of aeolian abrasion on the spectral signature and surface texture of the sediments were examined using laboratory abrasion experiments. Changes in spectral reflectance of abrasion samples from the laboratory were compared to field samples that were abraded naturally because of sediment transport. The changes resulting from increased time of abrasion are similar to those observed with increased distance downwind in the dunefield. These results suggest that downwind abrasion can explain the pattern of dune colour in the Muleshoe Dunes, although this does not preclude other possible causes. © 2008 Elsevier B.V. All rights reserved.
1. Introduction Many continental dunefields exhibit a gradient of redness caused by an iron-rich material coating the surface of sand grains or by differing sediment sources. Redness gradients have been quantified using spectral reflectance properties along field transects in the Simpson dunefield, central Australia (Bullard and White, 2002) and mapped successfully over large areas such as the Namib dunefield, southwest Namibia, using remote sensing (White et al., 1997, 2007). This paper examines the changes in spectral reflectance properties that result from laboratory abrasion of a sample of sand from the upwind end of the Muleshoe Dunes, a small dunefield in SW USA, and compares the results with the redness gradient observed in the field. When calibrated using field or laboratory methods, remote sensing can be used to map the spatial characteristics of dunefields—such as landform distribution (Quarmby et al., 1989; Pease et al., 1999) and the geochemical properties of sediments (El-Baz, 1978; White et al., 2001; Ben-Dor et al., 2006). Bullard and White (2005) suggested that the potential exists to use remote sensing to map sedimentological processes operating in dunefields, such as aeolian abrasion, and this paper explores this concept further. A large part of the Southern High Plains (Llano Estacado) of northwestern Texas and eastern New Mexico is covered by extensive (N10 000 km2), stabilized deposits of aeolian sand. Within this region, the Muleshoe Dunes (Fig. 1) are a west–east trending belt of sand that ⁎ Corresponding author. Tel.: +44 118 3787752; fax: +44 118 9755865. E-mail addresses:
[email protected] (K. White),
[email protected] (J. Bullard). 0169-555X/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.geomorph.2008.01.019
consists of a series of individual dune fields separated by sand sheets or sand free areas (Holliday, 2001). The activity of these dunefields over the late Holocene and into historical times, and the current and potential future status has received a significant amount of research (Ahlbrandt et al., 1983; Madole,1994; Forman et al.,1995; Muhs and Holliday, 1995). Other work has focussed on the provenance of these sands (Muhs et al., 1996; Holliday, 1997). Detailed geochemical analyses point to the underlying Blackwater Draw Formation, a vast (N100 000 km2) sheet of Quaternary aeolian sands up to 27 m in thickness, as a source of the Muleshoe dune sands (Muhs and Holliday, 2001). The dunes follow the dry valley of the Blackwater Draw, a tributary of the Brazos River. The most common dune forms are simple parabolic dunes that are 200– 400 m long, associated with blowouts, coppice dunes, barchan dunes and fence row dunes (historic dunes formed along field boundaries). The orientation of the dunes indicates palaeowinds dominantly from the west, which agrees with resultant drift directions derived from modern meteorological data during winter and spring when vegetation cover is at a minimum (Muhs and Holliday, 2001). The westerly wind direction may result, in part, from funnelling winds through the Portales Valley, a re-entrant in the western High Plains escarpment (Holliday, 2001). Radiocarbon dating of buried soils indicates that the Muleshoe Dunes accumulated in several stages: after 1300 cal yr B.P., after 750–670 cal yr B.P., after 500 cal yr B.P., and during the last 200 years (Holliday, 2001). Stabilization of the Muleshoe Dunes occurs because of vegetation cover, thus, the dunes are supply-limited but can be reactivated by minor droughts (Muhs and Holliday, 2001). The Muleshoe Dunes show considerable spatial variability of colour, carbonate content, granulometry and clay content (Muhs and Holliday, 2001). Two types of dunes are recognised on the basis of
60
K. White, J. Bullard / Geomorphology 105 (2009) 59-66
Fig. 1. Location map of the Muleshoe Dunes (black rectangle on inset), showing the location of samples collected for spectral reflectance measurements (white circles) and for chemical analyses by Muhs and Holliday (2001) (white crosses).
carbonate content; pink dunes of reddish hues (5YR or 7.5YR) that are largely free of carbonates; and overlying white dunes of light grey and light brown hues (10YR 6/2, 10YR 6/3, 10YR 6/4 and 10YR 7/2) that contain varying amounts of carbonate. The pink dunes show a systematic trend in colour across the Muleshoe Dunes, strong enough to be detected by Munsell Soil Colour determination (Muhs and Holliday, 2001), despite the low sensitivity and poor reproducibility of this technique (Anton and Ince, 1986; Bullard and White, 2002). To the west, the pink dunes have mostly reddish 5YR hues and become less red (10YR) towards the east (the downwind direction). This observation is contrary to the pattern found in other dune fields where dunes usually become redder downwind (El-Baz, 1978; Walker, 1979; Gardner and Pye, 1981; Lancaster, 1989; Bullard and White, 2002) and, together with the decreasing clay content downwind, is interpreted by Muhs and Holliday (2001) as resulting from increasing downwind abrasion and the progressive removal of red-pigmenting clay coatings from the sand grains. This paper seeks to examine the hypothesis proposed by Muhs and Holliday (2001) that the pattern of decreasing dune redness downwind, evident in the pink dunes, results from abrasion. The spectral reflectance properties of samples abraded in the laboratory are compared to samples collected at different locations in the Muleshoe Dunes. Comparison of the surface textures of laboratory abraded samples with those of samples from different locations in the Muleshoe Dunes are used to look for further evidence linking the variations in dune colour to aeolian abrasion. 2. Methodology 2.1. Laboratory abrasion A sample of red sand was collected from the western end of the Muleshoe Dunes (34.4151°N, 103.6169667°W). The sediment was abraded using an aeolian abrasion chamber of the same design as that used by Whalley et al. (1987). The sample was placed in the bottom of a large glass ‘test-tube’ abrasion chamber and the grains agitated by an air stream. Fine particles raised into suspension within the chamber are trapped by an electrostatic precipitator operating at 5 kV with a trapping efficiency of 93–95% (Bullard et al., 2004). The apparatus is designed to simulate the aeolian abrasion process that results from the saltation of particles, but it is not yet possible to equate the circular airflow in the abrasion chamber with an equivalent wind velocity in the field. Airflow in the apparatus was sufficient to lift sand particles a maximum of 8 cm above the base of the test-tube with the majority of grains rising only 5 cm The apparatus is likely to exaggerate the abrasion effect because samples are confined on three sides by glass walls and airflow is continuous rather than gusty. Eight samples of ∼ 10 g were abraded for 0, 12, 24, 48, 72, 144, 240 and 528 h. The spectral reflectance of each abraded sample was
measured using the methodology outlined below. The surface textures of sand grains of abraded samples were visualized using scanning electron microscopy and compared with samples from different locations in the Muleshoe Dunes so as to determine whether samples from farther downwind in the dunefield share characteristics of surface texture with samples exposed to greater amounts of abrasion. 2.2. Reflectance spectrometry A subset of five sand samples were collected on an east–west transect across the Muleshoe Dunes during July 2002 (Fig. 1) to represent the full range of sand redness encountered. Spectral properties of these samples, and the abraded sands from the laboratory experiments, were measured in the laboratory using a GER 3700 spectrometer (Geophysical and Environmental Research Corp., 1999). Irradiance was provided by a 1000 W high-intensity halogen lamp at an angle of 45° and a distance of 50 cm A spectralon reference panel was used, and the sample and reference panel were viewed at nadir (90°). The values used were the average of five separate readings (each separate reading is itself an average of eight repeat measurements). Spectral reflectances were calculated using the formula BRFn ¼ Rt =Rc
ð1Þ
where BRFn = bidirectional reflectance (Hapke, 1981) in waveband n, Rt is radiance from target surface and Rc is radiance from the reference panel. A number of colour indices can be derived from these data using transformations into colorimetric coordinate systems (e.g., the Helmholtz system), but simple radiometric indices calculated from broad blue, green and red bands have been shown to be good predictors of desert soil colour (Mathieu et al., 1998). A simple measure of spectral redness – the reflectance in the visible red part of the spectrum divided by the sum of visible red, green and blue reflectances – has been shown to be a good predictor of colour change that results from increasing Feoxide concentration (Bullard and White, 2002): r ¼ BRFr = BRFr þ BRFg þ BRFb
ð2Þ
where r = spectral redness, BRFr is bidirectional reflectance in the visible red part of the spectrum (400–500 nm), BRFg is bidirectional reflectance in the visible green part of the spectrum (500–600 nm) and BRFb is bidirectional reflectance in the visible blue part of the spectrum (600–700 nm). Reflectances in these broad colour bands were convolved from the high spectral resolution data using the technique of Rollin (2002). The spectral resolution of the GER 3700 instrument over the range 400–1050 nm (the region where the iron oxide pigments
K. White, J. Bullard / Geomorphology 105 (2009) 59–66
responsible for the Muleshoe Dune colour trend have distinct absorption features) is 3 nm Measurements at longer wavelengths do not influence perceived colour and were not used in this study. This provides sufficient spectral resolution to enable identification of specific absorption features, but to examine how these features change during the process of abrasion, continuum removal must be applied (Green and Craig, 1985). This is a means of normalising reflectance spectra to allow comparison of individual absorption features from a common baseline (a convex hull fitted to the spectral curve) and the approach used is detailed in Bullard and White (2002). A mathematical function is used to approximate the continuum, which can then be removed by dividing or subtracting the measured spectra from the continuum function (Clark and Roush, 1984). These continuum-removed data are suitable for measurement of individual absorption features; use of raw reflectance data is not recommended because the depth, area and asymmetry of a feature will be influenced strongly by its position on the background continuum (Clark and Roush, 1984). Convex hulls were fitted to reflectance curves using straight-line segments that connect local spectra maxima. Differences between the hulls and original spectra were subtracted from a constant (100) to obtain the hull differences. The advantage of the hull difference over the hull quotient (where the hull is divided by the original spectrum) is that the depths of features remain constant with respect to the hull. By contrast, a feature that has the same depth with respect to the hull
61
as another feature, but which has lower values than the other feature, will be preferentially enhanced with the hull quotient technique (Grove et al., 1992). 2.3. Image processing To determine if the pattern of dune colour in the Muleshoe Dunes is of sufficient magnitude to enable mapping using multispectral remote sensing techniques, a spectral mixture model was fit to an Enhanced Thematic Mapper (ETM+) scene (30 m × 30 m pixels) covering the Muleshoe Dunes (path 031 row 36 Worldwide Reference System) observed on 30/05/2001. The scene was first converted to atsatellite planetary reflectance (Markham and Barker, 1986) and a simple scene-specific empirical atmospheric correction algorithm (Switzer et al., 1981) was applied. This technique estimates the component of atmospheric reflectance from the covariance matrix of areas of homogeneous reflectance over uneven topography. Although such empirical techniques result in imperfect removal of atmospheric effects, they are appropriate in the absence of contemporary atmospheric observations that would be necessary to apply a physical atmospheric correction using radiative transfer code or in situ observations of reflectance of known targets from which atmospheric contributions can be estimated. Linear mixing models have been widely applied to map dune geochemistry (Blount et al., 1990; White et al., 1997; White et al.,
Fig. 2. (A) Spectral reflectance curves for the artificially abraded samples, (B) the same data after continuum removal, (C) spectral reflectance curves for the five field samples, (D) the same data after continuum removal. See text for discussion.
62
K. White, J. Bullard / Geomorphology 105 (2009) 59-66
2001). The reflectance of heterogeneous pixels is an area weighted average of homogenous pixels. Specifically, R ¼ f1 μ 1 þ f2 μ 2 þ fc μ c þ e
ð3Þ
where R f1 μ1 ε
the pixel multispectral reflectance the fraction of the pixel covered by the jth cover type multispectral reflectance of the individual components noise term, with zero mean and covariance matrix N
The remotely sensed proportions were extracted from the unmixed red sand image using 3 × 3 pixel windows centred on the latitude/longitude of each field sampling point used in the study by Muhs and Holliday (2001). A 3 × 3 window was used to account for uncertainty in location associated with re-sampling and to evaluate local variability. These data were compared to geochemistry data from Muhs and Holliday (2001), who derived concentrations of Fe using energy-dispersive X-ray fluorescence of bulk sediment samples taken from just below the zone of pedogenesis.
To invert the linear model, the number of scene components and the μj are required. Apart from the imperfect removal of the atmospheric component, problems also exist with applying linear mixing models to complex optical materials such as Fe-oxide coatings on quartz sand grains. Spectral properties of intimate mineral mixtures have been demonstrated to mix in a nonlinear fashion (Nash and Conel, 1974; Clark and Lucey, 1984). The only case where reflectance spectra add linearly according to the areal coverage is where patches of optically isolated minerals occupy discrete parts of the field of view of the spectrometer (Clark, 1983). Pieters (1983) demonstrated that, for silicate minerals, mean length of the optical path in fine sands (b250 μm) can be up to 2 mm in the near infrared, leading to interactions with up to 50 grains, though a mean length of the optical path of 1 mm, and 20 or fewer grain interactions would be more typical of a standard particulate surface where opaque minerals were also present. Thus, the relationship between estimates of mineral composition derived from a spectral mixture model must first be compared empirically with estimates derived from ground survey (White et al., 2001). In this experiment, the proportion of red sand in each pixel is estimated and an empirical relationship is derived between these data and Fe-oxide content estimates obtained by laboratory analyses of field samples. An advantage of this approach is that errors can be evaluated using Least Squares model-fitting techniques. The area outside the dunefield was masked to omit it from further analysis. This is a necessary step to limit the number of spectral end members because the unmixing equations can only be solved for n − 1 end members, where n = the number of spectral bands (6 in the case of ETM+). Spectral outliers defining the boundaries of the data cloud were identified in principal component feature space (Smith et al., 1985). Four spectral end members were extracted, and these were interpreted as red sand, carbonates/soils, vegetation, and shade/water. To map the varying proportions of these end members throughout the study area, the generalized least squares estimator (Settle and Drake, 1993) is found by minimizing ðR−Mf Þ N−1 ðR−Mf Þ T
ð4Þ
where R N f M T
the pixel vector a matrix of errors (f1, f2,… fc)T is a vector of the scene components a matrix whose columns are the end-member spectra signifies the transpose of the matrix
Eq. (4) is minimised subject to a single constraint (fi ≥ 0 where i = 1,2,… c) to ensure no negative proportions are estimated. Negative proportions can be useful to indicate where a significant spectral component has been overlooked in the model. The R.M.S. error for each pixel, however, is also calculated, which quantifies the residuals not explained by the mixture model and enables identification of any scene components that have been overlooked or incorrectly characterised.
Fig. 3. (A) The relationship between spectral redness of the field samples and the distance east of the Pecos River of each sample point. (B) The relationship between spectral redness and duration of abrasion. (C) A line representing ‘equal redness’, enabling comparison of the effects of laboratory abrasion on sediment redness with the downwind pattern observed in the Muleshoe Dunes.
K. White, J. Bullard / Geomorphology 105 (2009) 59–66
3. Results The raw spectra (Fig. 2A) show only one distinct absorption feature over the range 400–1050 nm This Fe3+–O charge transfer band is located in the ultraviolet part of the electromagnetic spectrum. It has a well resolved absorption edge tailing into the visible, causing a fall-off in reflectance short of 550 nm (Hunt et al., 1971) and resulting in a strong minima in the hull differences between 500 and 530 nm (Fig. 2B). Charge transfers occur where the absorption of a photon causes an electron to move between ions or between ions and ligands. The band centres usually occur in the ultraviolet with the longerwavelength wing of the absorption extending into the visible. Charge transfer absorptions are the main cause of the red colour of iron oxides and hydroxides (Morris et al., 1985; Clark, 1999). Our characterisation of this feature from continuum-removed data (Fig. 2B) is imperfect, because we have only captured the longest wavelength edge of this feature. As this feature deepens, however, the apparent absorption minimum moves to longer wavelengths and increases the perceived ‘redness’ of the sample. The main red-pigmenting Fe-oxide, haematite, also has a broad crystal field transition in the near infrared at 850 nm (Hunt, 1979), but this feature occurs outside the visible part of the spectrum and does not have a direct effect on sediment colour (Bullard and White, 2002; White et al., 2007). The decrease in redness towards the east of the dunefield results from the filling of the charge transfer feature (Fig. 2B). A similar pattern is evident in the reflectance spectra of the laboratory abraded samples (Fig. 2C–D), with greater
63
amounts of abrasion resulting in progressive filling of the charge transfer feature. Although the Pecos River sediments are not thought to be the provenance of the Muleshoe Dune sands, other workers have used the river as a useful reference point from which to measure geochemical variations along a west–east transect (Muhs and Holliday, 2001). A similar approach is adopted in this study for the purpose of data display. The best model fit for the relationship between spectral redness of the field samples and the distance east of the Pecos River from the point where they were collected (Fig. 3B) is a logarithmic function (explaining 92.3% of the variance). r ¼ −0:0461 ln d þ 0:6684
ð5Þ
where r = spectral redness index and d = distance east of Pecos River (km). A plot of spectral redness versus duration of abrasion is nonlinear and demonstrates the removal of red coating from abrasion (Fig. 3A). The best model fit (explaining 97.7% of the variance) is achieved with a quadratic divided by linear model, which produces an hyperbola with non-horizontal asymptote. r ¼ 0:46118 þ 0:01587=ð1 þ 0:042aÞ−0:00004431a
ð6Þ
where a = duration of abrasion (hours). A line of ‘equal redness’ can be generated from Eqs. (5) and (6), enabling a comparative relationship between the decrease in redness over a given period of abrasion and the decrease in redness with distance from the Pecos River (Fig. 3C). The upwind edge of the dunefield is 63 km east of the Pecos River and redness here equates to zero hours of laboratory abrasion. Following 48 h of abrasion, the dune redness equates to sand colour 83 km east of the Pecos River. Following 528 h of abrasion, the sediment redness equates to that of sediment 147 km east of Pecos River or 84 km from the upwind end of the dunefield. These values are still less than the change in redness observed over the whole dunefield. The nonlinear relationship at the upwind end of the dunefield (Fig. 3C) arises from the rapid loss of redness during the first 48 h of abrasion seen in Fig. 3B. Scanning electron micrographs of grain surface textures of downwind field samples (Fig. 4A) and artificially abraded samples (Fig. 4B) show very similar features. These include adhering upturned platelets (Linde and Mycielska-Dowgiallo,1980), which build-up on the surface of abraded grains as the grain surface becomes more reactive because of the formation of a surficial disrupted lattice layer, and other diagnostic features of aeolian abrasion fatigue such as arcuate, circular or polygonal cracks, and apparent shock-melting features (Mahaney, 2002). The heavy chipping and spalling evident in these micrograms helps explain the loss of Fe-oxide coating that results in decreasing spectral redness. The results of applying the spectral mixture model are shown in Fig. 5. Fig. 5A shows the visible blue band of the ETM+ image, for spatial reference. The R.M.S. error image (Fig. 5F) shows little spatial pattern, though the striping noise inherent in whiskbroom sensors is evident and suggests that the model has accounted for most of the spectral variability present in the area. The map of red sand proportions (Fig. 5B) is positively correlated with laboratory estimates of Fe concentration (Fig. 6). This relationship is statistically significant (r2 =0.588) and provides some confidence in the observed systematic downwind decrease in redness evident in Fig. 5B. The other maps of proportion estimates (Fig. 5C–E) are included for completeness, but are not calibrated against independent data and are not considered further here. 4. Discussion
Fig. 4. (A) SEM micrograph of grain surface texture of field sample from the east of the Muleshoe Dunes, showing fracturing of surface and build-up of upturned plates, (B) similar features developed after 528 h of artificial abrasion in the laboratory. See text for discussion.
On the basis of downwind trends in granulometry and clay content, Muhs and Holliday (2001) hypothesised that the redness gradient evident in the pink Muleshoe Dunes is a function of aeolian
64
K. White, J. Bullard / Geomorphology 105 (2009) 59-66
Fig. 5. Proportions estimate images produced by mixture modelling; (A) contrast stretched ETM+ band 1 image for reference, (B) red sand, (C) carbonates/soils, (D) vegetation, (E) shade/moisture, (F) R.M.S. error (dark tones indicate small R.M.S. errors, light tones indicate high R.M.S. errors).
abrasion processes during sand transport. If we assume that Muleshoe Dune sand colour is entirely a function of abrasion resulting from eastward transport, and that the empirical relationships derived from
our field data and our laboratory simulations are an appropriate representation of the abrasion process, then the change in redness after 528 h of artificial abrasion is equivalent to that of 84 km distance
K. White, J. Bullard / Geomorphology 105 (2009) 59–66
65
variation likely results from gradually varying concentrations of Feoxide coatings on the sand grains from west to east. The results of this research demonstrate that the decrease in dune redness may be explained by removal of the Fe-oxide coating by aeolian abrasion, similar to that simulated successfully in the laboratory. The resulting patterns in spectral redness can be remotely sensed by unmixing multispectral Landsat ETM+ data. Acknowledgements
Fig. 6. Relationship between remotely sensed redness index and Fe-oxide content (p.p.m.).
along a downwind transect in the field. In reality, attempts to simulate physical processes in the laboratory are subject to severe limitations. In this case, aeolian sediment transport is much more episodic than the continuous abrasion experienced in the laboratory experiment, and involves periods of storage and weathering. These results indicate, however, that it is possible to explain the pattern of dune colour in the Muleshoe Dunes by aeolian abrasion. The similar surface textures observed in field samples and artificially abraded samples, also indicate that aeolian abrasion mechanisms may account for the observed pattern. Even though the aeolian abrasion hypothesis of Muhs and Holliday (2001) might explain the observed pattern of dune redness, other potential causes must also be considered. First, as mentioned above, the source of the Muleshoe Dunes is thought to be the underlying Blackwater Draw Formation, and it is possible that the pattern of decreasing redness to the east may reflect changes in the Blackwater Draw source sediments. This is considered an unlikely explanation because previous research on the Blackwater Draw Formation has not identified any systematic spatial variation in sediment characteristics (Holliday, 2001). Second, whilst spectral reflectance has been proven to be an effective way of quantifying sediment redness, the redness value can be affected by sediment characteristics other than the Fe-oxide content. For example, the size and sorting of sediments can affect reflectance to the extent that sediments appear to be different colours when this difference is simply an artefact of the distribution of particle-sizes (Okin and Painter, 2004; Sanchez-Maranon et al., 2004). For this to explain the redness gradient in the Muleshoe Dunes, however, a systematic variation of particle size is needed from west to east throughout the dunefield, and sediment analyses by Muhs and Holliday (2001) demonstrate that this is not the case. Another complicating factor is that, in the remote sensing analysis, the apparent distribution of one scene component can be controlled by variation in another; for example, greater amounts of vegetation cover in the eastern part of the Muleshoe Dunes could give the appearance of reduced sediment redness. Laboratory analysis of field samples, however, demonstrates that the decreasing redness of the pink sand component towards the east is a real pattern, quantifiable by laboratory spectrometry. Ongoing sedimentological research seeks to address the problems outlined above using quantitative scanning electron microscopy (Pirrie et al., 2004) of abraded and non-abraded grains. The research reported here, however, demonstrates that abrasion remains a candidate process to explain the observed pattern of dune redness in the Muleshoe Dunes. 5. Conclusions The pink Muleshoe Dunes in SW USA show a marked pattern of decreasing redness in an easterly direction (downwind). The colour
The authors would like to thank Dan Muhs (US Geological Survey) and Vance Holliday (University of Wisconsin) for useful discussions and making available their geochemical data, Tom Gill (University of Texas at El Paso) for all his help with logistics and collection of field data, Christa Pudmenzky (Griffith University, Australia) for her assistance with the laboratory abrasion experiments and Jeff Settle (NERC-ESSC) for help with data analysis. This research was funded by The Leverhulme Trust with additional funding from Loughborough University. We are also grateful to Grady Blount and Patrick Pease for their careful and helpful reviews of the original manuscript. References Ahlbrandt, T.S., Swinehart, J.B., Maroney, D.G., 1983. The dynamic Holocene dunefields of the Great Plains and Rocky Mountain basins, USA. In: Brookfield, M.E., Ahlbrandt, T.S. (Eds.), Eolian Sediments and Processes. Elsevier, New York, pp. 379–406. Anton, D., Ince, F., 1986. A study of sand colour and maturity in Saudi Arabia. Zeitschrift für Geomorphologie, N. F. 30, 339–356. Ben-Dor, E., Levin, N., Singer, A., Karnieli, A., Braun, O., Kidron, G.J., 2006. Quantitative mapping of the soil rubification process on sand dunes using an airborne CASI hyperspectral sensor. Geoderma 131, 1–21. Blount, G., Smith, M.O., Adams, J.B., Greeley, R., Christensen, P.R., 1990. Regional aeolian dynamics and sand mixing in the Gran Desierto: evidence from Landsat Thematic Mapper images. Journal of Geophysical Research 95, 15463–15482. Bullard, J.E., White, K., 2002. Quantifying iron oxide coatings on dune sands using spectrometric measurements: an example from the Simpson-Strzelecki Desert, Australia. Journal of Geophysical Research 107 B, ECV5 1–11. doi:10.1029/ 2001JB000454. Bullard, J.E., White, K., 2005. Dust production and the release of iron oxides resulting from the aeolian abrasion of natural dune sands. Earth Surface Processes and Landforms 30, 95–106. Bullard, J.E., McTainsh, G.H., Pudmenzky, C., 2004. Aeolian abrasion and modes of fine particle production from natural red dune sands: an experimental study. Sedimentology 51, 1103–1125. Clark, R.N., 1983. Spectral properties of mixtures of montmorillonite and dark carbon grains: implications for remote sensing minerals containing chemically and physically adsorbed water. Journal of Geophysical Research 88, 10,635–10,644. Clark, R.N., 1999. Spectroscopy of rocks and minerals, and principles of spectroscopy. In: Rencz, A.N. (Ed.), Manual of Remote Sensing for the Earth Sciences. John Wiley and Sons Ltd., New York, pp. 3–58. Clark, R.N., Lucey, P.G., 1984. Spectral properties of ice-particulate mixtures and implications for remote sensing. Journal of Geophysical Research 89, 6341–6348. Clark, R.N., Roush, T.L., 1984. Reflectance spectroscopy: quantitative analysis techniques for remote sensing applications. Journal of Geophysical Research 89 (B7), 6329–6340. El-Baz, F., 1978. The meaning of desert color in Earth orbital photographs. Photogrammetric Engineering and Remote Sensing 44, 69–75. Forman, S.L., Oglesby, R., Markgraf, V., Stafford, T., 1995. Paleoclimatic significance of late Quaternary eolian deposition on the Piedmont and High Plains, central United States. Global and Planetary Change 11, 35–55. Gardner, R., Pye, K., 1981. Nature, origin and palaeoenvironmental significance of red coastal and desert dune sands. Progress in Physical Geography 5, 514–534. Geophysical and Environmental Research Corp., 1999. GER 3700 User Manual Release 2.1. Milbrook, New York. Green, A.A., Craig, M.D., 1985. Analysis of aircraft spectrometer data with logarithmic residuals. Jet Propulsion Laboratory Publication 85–41, 111–119. Grove, C.I., Hook, S.J., Paylor, E.D., 1992. Laboratory reflectance spectra of 160 minerals 0.4 to 2.5 micrometers. Jet Propulsion Laboratory Publication 92–2 406 pp. Hapke, B., 1981. Bidirectional reflectance spectroscopy 1: theory. Journal of Geophysical Research 86, 3039–3054. Holliday, V.T., 1997. Origin and evolution of lunettes on the High Plains of Texas and New Mexico. Quaternary Research 47, 54–69. Holliday, V.T., 2001. Stratigraphy and geochronology of upper Quaternary eolian sand on the Southern High Plains of Texas and New Mexico, USA. Geological Society of America Bulletin 113, 88–108. Hunt, G.R., 1979. Spectral signatures in the 0.4-1.1 micron region. Multispectral Resource Sampler 3, 97–116. Hunt, G.R., Salisbury, J.W., Lenhoff, C.J., 1971. Visible and near-infrared spectra of minerals and rocks: III. Oxides and hydroxides. Modern Geology 2, 195–205. Lancaster, N., 1989. The Namib Sand Sea. Balkema, Rotterdam.
66
K. White, J. Bullard / Geomorphology 105 (2009) 59-66
Linde, K., Mycielska-Dowgiallo, E., 1980. Some experimentally produced microtextures on grain surfaces of quartz sand. Geografiska Annaler Series A 62, 171–184. Madole, R.F., 1994. Stratigraphic evidence of desertification in the west-central Great Plains within the past 1000 yr. Geology, 22, 483–486. Mahaney, W.C., 2002. Atlas of Sand Grain Surface Textures. Oxford University Press,, Oxford. Markham, B.L., Barker, J.L., 1986. Landsat MSS and TM post-calibration dynamic ranges, exoatmospheric reflectances and at-satellite temperatures. EOSAT Landsat Technical Notes, pp. 3–8. Mathieu, R., Pouget, M., Cervelle, B., Escadafel, R., 1998. Relationships between satellitebased radiometric indices simulated using laboratory reflectance data and typic soil colour of an arid environment. Remote Sensing of Environment 66, 17–28. Morris, R.V., Lauer, H.V., Lawson, C.A., Gibson Jr., E.K., Nace, G.A., Stewart, C., 1985. Spectral and other physiochemical properties of submicron powders of hematite (−Fe2O3), maghemite (−Fe2O3), maghemite (Fe3O4), goethite (−FeOOH), and lepidochrosite (−FeOOH). Journal of Geophysical Research 90, 3126–3144. Muhs, D.R., Holliday, V.T., 1995. Evidence of active dune sands on the Great Plains in the 19th Century from accounts of early explorers. Quaternary Research 43, 198–208. Muhs, D.R., Holliday, V.T., 2001. Origin of late Quaternary dune fields on the southern High Plains of Texas and New Mexico. Geological Society of America Bulletin 113, 75–87. Muhs, D.R., Stafford Jr., T.W., Cowherd, S.D., Mahan, S.A., Kihl, R., Maat, P.B., Bush, C.A., Nehring, J., 1996. Origin of the late Quaternary dunefields of northeastern Colorado. Geomorphology 17, 129–149. Nash, D.B., Conel, J.E., 1974. Spectral reflectance systematics for mixtures of powdered hypersthene, labradorite and ilmenite. Journal of Geophysical Research 79, 1615–1621. Okin, G.S., Painter, T.H., 2004. Effect of grain size on remotely sensed spectral reflectance of sandy desert surfaces. Remote Sensing of Environment 89, 272–280. Pease, P.P., Bierly, G.D., Tchakerian, V.P., Tindale, N.W., 1999. Mineralogical characterization and transport pathways of dune sand using Landsat TM data, Wahiba Sand Sea, Sultanate of Oman. Geomorphology 29, 235–249. Pieters, C.M., 1983. Strength of mineral absorption features in the transmitted component of near-infrared reflected light: first results from RELAB. Journal of Geophysical Research 88, 9534–9544. Pirrie, D., Butcher, A.R., Power, M.R., Gottlieb, P., Miller, G.L., 2004. Rapid quantitative mineral and phase analysis using automated scanning electron microscopy
(QemSCAN); potential applications in forensic geoscience. In: Pye, K., Croft, D.J. (Eds.), Forensic Geoscience: Principles, Techniques and Applications, Geological Society, vol. 232. Special Publications, London, pp. 123–136. Quarmby, N.A., Townshend, J.R.G., Millington, A.C., White, K., Reading, A.J., 1989. Monitoring sediment transport systems in a semiarid area using Thematic Mapper data. Remote Sensing of Environment 28, 305–315. Rollin, E.M., 2002. Applying filter functions to GER1500 spectral reflectance data using a spreadsheet. NERC EPFS Advice Note G15/01/Tec/02. NERC Equipment Pool for Field Spectroscopy. University of Southampton, U.K. Sanchez-Maranon, M., Soriano, M., Melgosa, M., Delgado, G., Delgado, R., 2004. Quantifying the effects of aggregation, particle size and components on the colour of Mediterranean soils. European Journal of Soil Science 55, 551–565. Settle, J.J., Drake, N.A., 1993. Linear mixing and the estimation of ground cover proportions. International Journal of Remote Sensing 14, 1159–1177. Smith, M.O., Johnson, P.E., Adams, J.B., 1985. Quantitative determination of mineral types and abundances from reflective spectra using principal components analysis. Journal of Geophysical Research Supplement 90, C797–C804. Switzer, P., Kowalick, W.S., Lyon, R.J.P., 1981. Estimation of atmospheric path radiance by the covariance matrix method. Photogrammetric Engineering and Remote Sensing 47, 1469–1476. Walker, T.R., 1979. Red colour in eolian sand. In: McKee, E.D. (Ed.), A Study of Global Sand Seas. USGS Professional Paper, vol. 1052, pp. 62–81. Whalley, W.B., Smith, B.J., McAlister, J.J., Edwards, A.J., 1987. Aeolian abrasion of quartz particles and the production of silt-size fragments: preliminary results. In: Frostick, L., Reid, I. (Eds.), Desert Sediments: Ancient and Modern. Geological Society Special Publication, vol. 35, pp. 129–138. White, K., Walden, J., Drake, N.A., Eckardt, F., Settle, J.J., 1997. Mapping the iron oxide content of dune sands, Namibia, using Landsat Thematic Mapper data. Remote Sensing of Environment 62, 30–39. White, K., Goudie, A.S., Parker, A., Al-Farraj, A., 2001. Mapping the geochemistry of the northern Rub' Al Khali using multispectral remote sensing techniques. Earth Surface Processes and Landforms 26, 735–748. White, K., Walden, J., Gurney, S.D., 2007. Spectral properties, iron oxide content and provenance of Namib dune sands. Geomorphology 86, 219–229. doi:10.1016/j. geomorph.2006.08.014.