Aeolian sand strip mobility and protodune development on a drying beach: examining surface moisture and surface roughness patterns measured by terrestrial laser scanning

Transient sedimentary bedforms such as slipfaceless sand strips are often present during sand transport events, particularly on beaches where surface moisture (Bauer and Davidson-Arnott, 2003; Jackson et al., 2006; Davidson-Arnott et al., 2008; Bauer et al., 2009; Davidson-Arnott and Bauer, 2009), roughness elements such as vegetation debris (Sherman, 1990; Eamer and Walker, 2010) or frozen material (Hesp and Arens, 1997) may vary both spatially and temporally and so influence aeolian transport dynamics. Little is known about the mobility characteristics of these transient features and few studies consider the influence they may have on measurements of sediment transport patterns measured using point-based anemometers, saltation probes and sediment traps. More importantly, the very nature of these mobile strips means that they will periodically expose wetter underlying surfaces as they traverse the beach, whilst the formation and movement of the sand strip itself may be partly controlled by the characteristics of the surface it travels over (Jackson and Nordström, 1997; Davidson-Arnott et al., 2005). Surface moisture is a key parameter in aeolian environments because it influences aeolian sediment transport by increasing the shear velocity threshold required to mobilize sediment (Sarre, 1988; Namikas and Sherman, 1995). This reduces surface sediment release (McKenna Neuman and Scott, 1998) and extends fetch length (e.g. Bauer and Davidson-Arnott, 2003; Davidson-Arnott et al., 2008; Bauer et al., 2009; Davidson-Arnott and Bauer, 2009; Hesp et al., 2009; Walker et al., 2009a; Walker et al., 2009b). Yet, it remains an elusive and poorly constrained physical attribute that contributes to difficulties in successfully applying predictive sediment transport models (Namikas et al., 2010).

The high degree of spatial and temporal variability of surface moisture can confound traditional measurement techniques [such as soil moisture probes and gravimetric measurements (Wiggs et al., 2004b; Yang and Davidson-Arnott, 2005)] that rely on ‘spot-based’ sampling. Whilst soil moisture probeshave proven to be very useful in elucidating the variability of surface moisture over the complete intertidal and back beach area (Oblinger and Anthony, 2008; Edwards and Namikas, 2009; Anthony et al., 2009; Namikas et al., 2010), their volumetric measurement technique means that they struggle to take account of the smaller-scale moisture variations where sharp, near-surface gradients in moisture content exist. They are therefore not ideal for determining moisture properties of a thin veneer of surface particles (Darke et al., 2009), and it is this surface moisture content which sediment entrainment and transport is most sensitive to. Further, it is specifically the drier upper region on beaches where surface moisture is typically less than 5% (gravimetrically) that the biggest detection difficulties are realized. It is in these same regions that sand strips are found which make the most significant contributions to sediment supply for dune construction (Houser, 2009) and where moisture imparted by light rain events is likely to have the most prominent impact on sand transport rates (Sherman et al., 1998).

Furthermore, along with moisture variability issues it is not easy to measure saltation and transport patterns over a large area, as this usually requires extensive systems of saltation probes and sediment traps, covering single or multiple points along a transect (Baas and Sherman, 2005, 2006). These spatially variable transport patterns are in part controlled by surface moisture (Wiggs et al., 2004a), along with the roughness of the ground they traverse over, and investigating trends over a single point or small area may not encompass the complexity, heterogeneity and large-scale relationships between components (Jackson and Nordström, 1997; Cornelis et al., 2004). Saltation, along with large-scale and micro-scale surface roughness changes can significantly influence aerodynamic roughness, shear velocity, and hence the ability of the wind to entrain sand grains (Owen, 1964). As such, in order to better understand sediment transport patterns, there is a need to examine the inter-relationships between these parameters across varying spatial and temporal scales. Synoptic remote sensing offers an attractive alternative to traditional, point-based sampling methods in this regard.

Recently remote sensing of surface brightness has been undertaken using video imagery (McKenna Neuman and Langston, 2006; Darke and Neuman, 2008; Darke et al., 2009). Whilst this enables surface moisture inference at a high spatial resolution and is particularly useful for long-term, large-scale monitoring (e.g. Delgado-Fernandez et al., 2009; Delgado-Fernandez and Davidson-Arnott, 2009), it has a number of limitations including daylight dependent operation and an inability to directly measure small changes in morphology. Similarly, whilst it can successfully measure large changes in surface moisture, it is less able to discriminate small changes (<2%) in moisture (Darke et al., 2009), particularly where sand is relatively dry (<10%) and potentially available for transport (McKenna Neuman and Langston, 2006). In contrast, terrestrial laser scanning (TLS) offers an alternative that can capture both the variation of surface moisture and beach morphology at high spatial and temporal resolution, while simultaneously also provide terrain information for sediment transport studies.

TLS has been used successfully to assess change over periods of weeks to years in a number of geomorphic settings such as rock cliffs (Rosser et al., 2005), gravel rivers (Hodge et al., 2009b), desert dunes (Nagihara et al., 2004; Ochoa, 2005), beach nourishment (Pietro et al., 2008), and shows huge potential for examining coastal (French and Burningham, 2009) and aeolian processes (Bullard, 2006; Baas, 2008). These instruments are time-of-flight systems that measure the distance and orientation of a surface point using a laser signal. The reader is referred to Buckley et al. (2008) and Hodge et al. (2009a) for reviews on TLS techniques and applications. Along with spatial orientation, TLS records the intensity of the reflected signal for each return point. The magnitude of this returned intensity is a function of the surface properties and the instrument position. Studies of small rock patches (Franceschi et al., 2009), salt marshes (Guarnieri et al., 2009) and burnt vegetation (Lichti, 2005) have shown the potential of TLS intensity to distinguish between different materials. Kaasalainen et al. (2008) found that saturated sand samples analysed in a controlled environment had a reduction in signal intensity of 30 to 50% relative to dry sand. The return signal intensity is influenced by a number of factors along with the physical nature of the material, including surface roughness, atmospheric conditions (i.e. rain, fog) and the distance and angle from the instrument. Of these factors, Franceschi et al. (2009)found distance to be the most important, particularly where the surface is rough and signal loss due to angular scatter is likely to be more uniform. This variation in intensity with distance has been corrected for by a number of researchers using a variety of TLS instruments which emit different wavelengths (e.g. Lichti, 2003; Lichti, 2005; Kaasalainen et al., 2008; Wang and Lu, 2009).

In this paper we calibrate the return signal intensity of a TLS to surface moisture through a controlled laboratory experiment and relate this to moisture variability on a beach during a transport event (3·5 hours) for the underlying beach surface. This paper particularly focuses on sand strips and quantifying slipfaceless bedform development perpendicular to a dune system. First, the study site is outlined in detail, followed by the method used to calibrate surface moisture measurements. Next, we analyse the effectiveness of the TLS at predicting beach surface moisture at high spatial resolutions. Finally, we interpret the development of sand strips and protodunes on a drying beach based on relationships between surface moisture and surface roughness.

The field experiment was conducted on 26 August 2009 at Ynyslas, on the central western coast of Wales, UK near Aberystwyth (site location: 52·53°N 4·06°W). The beach sediments are composed of well-sorted fine quartz sand with a mean grain size of approximately 0·26 mm. This grain size was determined using standard sieving methods from sediment collected within the top 4 mm of the beach surface during the experimental period. The study was undertaken on the northern, distal end of a spit which extends into the lower Dyfi estuary (May and Hansom, 2003). Sand overlays a gravel substrate which promotes high permeability, thereby encouraging surface drying and the availability of sand for aeolian transport (May and Hansom, 2003). Measurements were undertaken on a stretch of beach above the high tide extent, bounded to the southeast by a dune system and to the northwest by a small patch of nebkha dunes that extended towards the intertidal zone (see Figure 1a). During the measurement periods wind direction was consistently from the southwest [approximately 227° ± 8° (standard deviation), as measured at a location 1·5 km from the study site by a Davis Vantage Pro 2 weather station with anemometer and wind vane mounted 7·5 m above the ground surface]. Site measurements were undertaken after approximately 0·8 mm of precipitation was recorded earlier in the day.

The beach surface was measured three times at approximately 1·5 hour intervals between 4:20 p.m. and 7:50 p.m. using a Leica Scanstation TLS with a specified resolution of 5 mm at 25 m, positioned on an incipient foredune approximately 5 m above the scanned surface. This produced an average point cloud density of 7200 points/m². The TLS utilizes a green laser (wavelength 539 nm) and has a sampling frequency of 4000 points/s. During the experiment there was continual transport over the beach surface, which noticeably increased as the surface became drier. The surface consisted of a moist, undulating sand substrate, which was intermittently covered by drier, rippled sand strips. Whilst each scan took approximately 20 minutes to complete, we assume that the bulk sand strip mobility is slower than the scan duration, and as full surface scans covered a similar area at the same speed, influence of bedform migration during scanning should be minimal. Spatially small high-resolution patches were also scanned of both rippled sand strip and moist interstrip surface types covering areas of between 1 and 10 m² (average point cloud density of 77,900 points/m²). In each small patch moisture was measured using a Type ML2x Theta probe (Delta-T Devices, 1999) shortened to 1·5 cm by insertion through dielectric foam (Edwards and Namikas, 2009; Davidson-Arnott et al., 2008; Yang and Davidson-Arnott, 2005) and surface scrapings 4 mm deep were collected with a 20 cm wide surface sediment sampler for gravimetric analysis using the method of Wiggs et al. (2004b).

There are a number of inherent problems associated with gravimetric sampling and Theta probe measurements in beach environments. Theta probes are unable to measure moisture directly on the beach surface, but make an average reading over the top 1·5 cm. So although the surface and near-surface samples follow similar moisture trends, on areas of the beach where the surface dries rapidly, the near-surface response is lagged and the Theta probe measurement tends to reflect the averaged near-surface rather than the actual surface characteristics. Similar conclusions were made by (Darke et al., 2009) and are illustrated in Figure 2 where there is a weak linear relationship (R² = 0·69) between gravimetric and probe voltage measurements for surfaces with an average moisture content of less than 4%. Our study focuses on the influence of rain on the upper beach area (<13% moisture) in contrast to other beach studies that use Theta probes where the measurement focus is on the influence of tidal cycles on beach moisture conditions attributes over much larger areas, extending from the swash zone to the base of bordering dunes (e.g. Namikas et al., 2010). Over these larger areas there is substantially more near-surface moisture variation and the underlying sediment may have considerably higher mean moisture content (up to 30%), even if the surface layer is dry enough for transport to occur.

Gravimetric surface scrapings also include errors due to the uneven rippled characteristics of beach surfaces where crests and troughs of ripples may have very distinct moisture values yet are sampled in a single scrape. In our experiments Theta probe readings in an adjacent ripple crest and trough indicated a difference of 20 mV (1–3% moisture difference based on probe calibration). Further errors are possible because of contamination by drier saltating grains.

Despite these inherent errors, both gravimetric and Theta probe measurements have the ability to elucidate beach surface moisture patterns, and have been used with particular success (R² = 0·99) during periods of reduced or nil sediment transport over a large range of moisture values (Edwards and Namikas, 2009; Namikas et al., 2010). As Figure 2 indicates, the success of both the probes and gravimetric surface scraping is reduced during transport events and over a narrow moisture range. To limit the influence of these factors, inherent on an active, back beach area, we undertook a controlled experiment to calibrate TLS signal return intensity and surface moisture, before applying the derived relationships to the field data.

Distance at which a sample is scanned influences the signal return intensity, so we developed a calibration curve using a controlled experiment. Dry sand collected from the beach surface at Ynyslas in the area where the field experiment took place was placed in a 1 cm deep tray and scanned at distances between 20 m and 35 m to develop a distance calibration relationship. A section of scanned points 20 cm wide were extracted from the TLS point cloud for each distance increment. A non-linear response of intensity has been previously observed at distances less than 20 m (Lichti, 2005; Lichti, 2003; Wang and Lu, 2009; Franceschi et al., 2009). However, at distances of between 20 and 35 m, the target distance range in this study, behaviour between measured dry sediment sample intensity and Euclidean distance between sample and TLS is strongly linear (R² = 0·99) as shown in Figure 3.

Two separate controlled experiments was also undertaken to calibrate TLS signal intensity to gravimetric moisture using Ynyslas sand. Two 1 cm deep trays were filled with sand, positioned along side each other and scanned at a distance of 20 m from an elevated location, emulating the position of the TLS on the foredune above the scanned beach surface in the field. A single scan was completed over approximately two minutes, after which 4 mm surface scrapings were bagged and moisture measured gravimetrically. The control tray remained dry, whilst well-mixed sediment with varying moisture content (0–10%) was placed in the second tray for scanning and sampling. In addition, Theta probe measurements were taken of each moist sediment sample from a beaker containing the same sediment as the tray. This process was repeated until probe voltage measurements were above those found in situ in the field (110 mV). The first experiment was undertaken during humid, overcast weather conditions, whilst the second took place on a dry, windy, sunny day.

The scraped surface sediment samples from the controlled experiment were dried at 105 °C for 24 hours and their gravimetric moisture content calculated. As in the distance experiment, a section of scanned points 20 cm wide corresponding to the position and size of the sediment surface sampler were extracted from the TLS point cloud for each moisture increment. A strong power relationship exists for each of the two control experiment data sets between gravimetric water content and average intensity (R² equals 0·93 and 0·97 for the humid and dry days, respectively). When the results from both these experiments are combined, the power relationship (R² = 0·92) is:

where W is the gravimetric moisture content (as a percentage) and S is the TLS return signal intensity. This relationship is particularly strong over the lower gravimetric moisture content range between 0 and 6% (Figure 4) where other studies suggest the threshold for grain entrainment is likely to occur in many environmental situations (Namikas and Sherman, 1995). Standard deviations are small (mean of 0·004, <13% of the distance calibrated intensity range, denoted by the error bars) suggesting that intensity is able to discriminate gravimetric moisture content to within 1–2% in low moisture sediment samples. Such a result is encouraging given that sand entrainment and transport is seen to be sensitive to changes in moisture content within this range (Wiggs et al., 2004b; Bauer et al., 2009).

Signal intensity values from the time series and high spatial resolution patch data collected from the Ynyslas experiment were converted to moisture content after first correcting for distance attenuation using the relationship derived from the controlled experiments (Figure 4). The TLS returned signals both from the beach surface, and from the saltation cloud. Whilst the laser footprint (<4 mm) is much larger than a single grain (D₅₀ = 0·25 mm), if enough grains are saltating at the moment the laser beam passes over them, a signal related to the saltating cloud will be returned. This paper focuses on the relationship between surface moisture and sedimentation, so it was necessary to remove points that were not associated with the beach surface using a filtering method. The filter, based on angle of repose, was applied to all TLS cloud data to compare the angular orientation of point pairs within a 0·01 m² area and identify true surface location. The filter separated out any higher points where the angle between pairs exceeded the angle of repose (defined as 35°). Due to the nature of the scanning, where data collection over a large area was required in a short time period, it was not possible to carry out multiple scans from the same or different locations as is generally the method applied to reduce mixed pixels in fluvial landscapes where gravel bed geometries are measured between, and not during, transport events (Hodge et al. 2009a, 2009b).

Average moisture content determined from the 24 small, high-resolution patches scanned at Ynyslas is shown in Figure 5. There is a wide variation in surface moisture derived from the TLS even in these small patches. This is particularly noticeable in patches with higher surface moisture values and this is probably due to the greater uncertainty in measurement at lower spectral intensities. As expected, the TLS under-predicts the surface moisture compared to the gravimetric method, as indicated by points in Figure 5 which are positioned above the 1:1 relationship. This is because the gravimetric method provides an average value of moisture over a depth of 4 mm, whilst the TLS responds to moisture at the surface. In this way, the gravimetric method may over-predict the moisture of the surface, particularly where undulations and ripples are present as discussed earlier. While the agreement between the gravimetric moisture and surface TLS moisture is weak (R² = 0·36), the large standard deviations for each patch indicate the variability of surface moisture within each field patch which the TLS is able to discriminate. Particularly weak agreement between measurements occur when a thin veneer of dry surface grains covers and adheres to the wet underlying substrate, considerably increasing the gravimetric near-surface measurement. The strong relationship between TLS and gravimetric moisture (R² = 0·92) in the controlled experiment indicates that with careful calibration, the TLS is able to determine surface moisture characteristics of beach sediments.

The development of sand strips and protodunes after rain is related to saltation, surface roughness and moisture patterns. Following light rain there is a period of drying when the system is controlled by variation in erosivity and dominated by bedform development. As the beach surface becomes drier, sediment erodibility assumes control, and bedform migration dominates. TLS is a useful tool to examine the relationships between aeolian deposition patterns and surface moisture characteristics in areas where the spatial and temporal nature of bedform development limit the use of traditional methods. Further research is needed to investigate these relationships over a longer time period, along with saltation cloud attributes, aerodynamic measurements and traditional sediment collection to quantify active transport patterns. Overall, this research suggests that the inherent complexities of beach surface moisture interactions with surface roughness and saltation dynamics are important to consider when calculating sediment transport trends.

This study was partly funded by a British Society for Geomorphology (BSG: Wiley-Blackwell ESPL Fund) research grant awarded to Nield, and Squirrell was supported by a School of Geography, University of Southampton summer student bursary. Mike Bailey and Countryside Council for Wales are acknowledged for their permission to use Ynyslas as an experimental field site. Alan Cole is thanked for kindly providing meteorological data. Angela Harris gave invaluable assistance at Ynyslas and Richard Pomeroy, Chris Hackney and Julian Leyland helped with the controlled experiments. This research was improved by insightful discussions with Stephen Darby, Gary Kocurek, Julian Leyland, Joseph Wheaton and Jadu Dash. Robin Davidson-Arnott, an anonymous reviewer and the Earth Surface Processes and Landforms Editors are thanked for their comments which strengthened the focus of this paper.