Author's personal copy Geomorphology 119 (2010) 135–145

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

Relationships of aeolian erosion and deposition with LiDAR-derived landscape surface roughness following wildfire Joel B. Sankey a,⁎, Nancy F. Glenn a,1, Matthew J. Germino b,2, Ann Inez N. Gironella c,3, Glenn D. Thackray d,4 a

Department of Geosciences, Idaho State University-Boise, 322 E. Front St., Suite 240, Boise, ID 83702, USA Department of Biological Sciences, Idaho State University, 921 S 8th Ave, Stop 8007, Pocatello, ID 83209-8007, USA Department of Mathematics, Idaho State University, 921 S. 8th Ave, Stop 8085, Pocatello, ID 83209-8085, USA d Department of Geosciences, Idaho State University, 921 S 8th Ave, Stop 8072, Pocatello, ID 83209-8072, USA b c

a r t i c l e

i n f o

Article history: Received 18 December 2009 Received in revised form 8 March 2010 Accepted 11 March 2010 Available online 18 March 2010 Keywords: Aeolian transport Wildfire LiDAR Surface roughness Remote sensing Shrub steppe

a b s t r a c t The reduction of vegetation by wildfire can subject stable soil surfaces to increased aeolian transport. Vegetation and associated microtopography function as surface roughness elements that influence the entrainment, transport, and deposition of sediment by wind in burned and unburned semiarid shrublands. We examined whether surface roughness derived from LiDAR can explain variability in aeolian surface change following wildfire in loess soils of a cold desert shrub steppe in SE Idaho, USA. Erosion bridges were installed in Fall 2007, following a late summer wildfire to monitor soil surface change at sites in burned and downwind unburned areas. Surface elevation measurements were made when the erosion bridges were installed and again in Fall 2008. Surface change was determined from the difference in relative elevation between the two dates. Airborne LiDAR data were acquired in Fall 2007 following erosion bridge installation. Surface roughness was calculated at 2-m raster resolution using the standard deviation of all LiDAR elevations within the 2-m cells, after elevations were detrended to remove the effects of topographic slope. Surface change varied as a function of surface roughness among burned and unburned surfaces, with net erosion occurring on the relatively smooth, burned surfaces and net deposition occurring on the rough, unburned surfaces. Site mean surface change decreased linearly as a function of the inverse of site mean surface roughness (r2 = 0.77, p b 0.00). Quantile regression analysis indicated that changes in surface roughness were related to proportionally greater changes in erosion compared to deposition. Analysis of surface change at finer spatial scales suggested that aeolian processes occurred with strong spatial patterns on burned, but not unburned surfaces. Future research to examine relationships between aeolian transport and fine spatial resolution topographic variability, for example from ground-based LiDAR systems, is recommended. © 2010 Elsevier B.V. All rights reserved.

1. Introduction Aeolian transport is an important biogeomorphic agent that can mobilize ecologically sequestered sediment, minerals, nutrients, and pollutants at local to global spatial scales (Okin et al., 2006, 2009). The reduction of vegetation by wildfire is one particularly ubiquitous process that can subject otherwise stable surfaces to increased potential for aeolian transport. Increases in wind erosion following wildfire have been reported for a large range of biomes, including cold and warm deserts, grasslands, shrublands, and forests (Ash and Wasson, 1983; Wasson and Nanninga, 1986; Zobeck et al., 1989; ⁎ Corresponding author. Tel.: +1 208 220 6571; fax: +1 208 345 8353. E-mail addresses: [email protected] (J.B. Sankey), [email protected] (N.F. Glenn), [email protected] (M.J. Germino), [email protected] (A.I.N. Gironella), [email protected] (G.D. Thackray). 1 Tel.: +1 208 345 1994; fax: +1 208 345 8353. 2 Tel.: +1 208 282 3285; fax: +1 208 282 4570. 3 Tel.: +1 208 282 3465; fax: +1 208 282 2636. 4 Tel.: +1 208 282 3560; fax: +1 208 282 4414. 0169-555X/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.geomorph.2010.03.013

Wiggs et al., 1994, 1995, 1996; Whicker et al., 2002, 2006a,b; Vermeire et al., 2005; Ravi et al., 2007, 2009; Breshears et al., 2009; Sankey et al., 2009a). Particularly in semiarid shrublands, aeolian transport is recognized as a predominant biogeomorphic process that shapes contemporary pattern and form of vegetation and soil surfaces (Breshears et al., 2003; Okin et al., 2006). Cold desert shrub steppe is one type of semiarid shrubland in which aeolian transport following wildfire has been shown to be especially influential (Sankey et al., 2009a,b). Aeolian transport has two functional components: (i) the wind's ability to entrain soil particles (erosivity) that is impeded by surface roughness components including micro- and macrotopography and vegetation; and (ii) the soil's susceptibility to this entrainment (erodibility) (Bagnold, 1941; Cornelis, 2006; Okin et al., 2006). Erosivity and erodibility interact in complex ways involving potentially nonlinear, recursive and/or self-reinforcing relationships; and the study of aeolian transport in biogeomorphic systems therefore has many existing challenges (Baas, 2007, 2008). One challenge is the development of a more clear understanding of spatial dynamics of

Author's personal copy 136

J.B. Sankey et al. / Geomorphology 119 (2010) 135–145

aeolian processes: (i) on landscapes that span a wide range of surface roughness (e.g., bare to densely vegetated), and (ii) at multiple spatial scales that range from entire landscapes to that of individual plants and/or associated surface microtopography, such as raised mounds often located beneath woody vegetation and adjacent lower elevation interspaces (Okin et al., 2006; Ravi et al., 2007, 2009; Breshears et al., 2009). Conceptual models and a field experiment for warm desert shrublands suggest that, when vegetation is reduced by disturbance, wind removes sediment from coppices and deposits it in interspaces, which serves to redistribute biogeochemical resources and result in a more physically homogenous surface (Ravi et al., 2007, 2009; Li et al., 2008; Ravi and D'Odorico, 2009). Disturbance of vegetation by fire in cold desert shrub steppe has been suggested to homogenize the microtopography of the soil surface as well (Hilty et al., 2003), but the contributions of wind erosion and deposition to this process have not been explicitly examined to our knowledge. In semiarid shrublands, vegetation and associated microtopography function as surface roughness elements that influence aeolian transport (Wolfe and Nickling, 1993) by: (i) providing protective cover to the soil surface, (ii) altering wind flow by extracting momentum from the wind, and (iii) trapping soil particles. The protective cover provided by vegetation in undisturbed shrublands has been hypothesized to be less effective compared to other undisturbed landscapes (e.g., grasslands and forests) because (i) the relatively sparse spatial distribution of shrubland vegetation permits wind flow (in instances) to penetrate and even increase in erosive potential below the height of the mean vegetation canopy, and (ii) intershrub spaces tend to have relatively high amounts of exposed soil and relatively low cover of herbaceous vegetation (Breshears et al., 2009). Disturbance, such as wildfire, decreases the density and stature of herbaceous and woody vegetation, which increases the propensity for more erosive regimes of wind flow (Wolfe and Nickling, 1993; Breshears et al., 2009). In disturbed and undisturbed shrublands, sediment entrained by aeolian processes can be removed from transport (trapped) by surface roughness elements (Grant and Nickling, 1998; Raupach et al., 2001; Lee et al., 2002; Okin et al., 2006). Physical processes of sediment entrainment, transport, and deposition are well defined for surfaces that are vegetated (or unvegetated) with a homogenous and/or regular spatial pattern. Shrubland surfaces are often heterogeneous, however, and effects of surface roughness on aeolian processes are therefore not well understood across the continuum of disturbed to undisturbed surfaces in these environments (Okin et al., 2006). 1.1. LiDAR surface roughness Light detection and ranging (LiDAR) remote sensing technology is particularly well suited for inquiry of spatial variability of surface roughness and aeolian processes and is recognized as having great utility for the quantitative characterization of a wide range of biogeomorphic processes and systems (Tratt et al., 2008; Bauer, 2009; Pelletier et al., 2009). Digital elevation models (DEM) constructed from LiDAR point data have been used to examine the morphology and migration of desert and coastal sand dunes, for example (Bourke et al., 2009; Pelletier et al., 2009; Ewing and Kocurek, 2010). Specific to semiarid shrublands, LiDAR remote sensing has been used successfully to describe variability in microtopographic and shrub vegetation height and morphology (Menenti and Ritchie, 1994; Ritchie, 1995; Rango et al., 2000; Mundt et al., 2006; Streutker and Glenn, 2006). The term surface roughness implies a measure of topographic variability. A variety of approaches for quantitatively describing the roughness of biogeomorphic surfaces using LiDAR data have been reported. Determination of surface roughness using LiDAR has been most commonly performed by calculating a measure of the variability, often the standard deviation, of point elevations within a moving

window or pixel (Ritchie, 1995; Davenport et al., 2004; Glenn et al., 2006). LiDAR returns can be first classified as having reflected from ground or vegetation surfaces; and then roughness of bare earth, vegetation, and/or combined bare earth–vegetation calculated accordingly (Glenn et al., 2006; Streutker and Glenn, 2006). Prior to the roughness calculation, LiDAR point elevations can be detrended to remove variability because of macrotopographic slope (Davenport et al., 2004). In lieu of LiDAR point elevation, variability in derivatives of elevation (e.g., slope and aspect) can be examined using a LiDARderived DEM (McKean and Roering, 2004; Frankel and Dolan, 2007). Pelletier et al. (2009) calculated surface roughness from a LiDARderived DEM as the difference between maximum and minimum elevations within a neighborhood of pixels, for example. Surface roughness estimates derived from LiDAR were presented for vegetated surfaces in several landscapes by Ritchie (1995), who noted their potential utility in the prediction of aeolian processes. In semiarid shrublands, surface roughness from LiDAR has been used to estimate aerodynamic roughness, an important parameter for physical models of aeolian transport (Menenti and Ritchie, 1994). Recently, an inverse relationship between surface roughness derived from a LiDAR DEM and the entrainment of sand particles was employed to model the migration of coastal sand dunes (Pelletier et al., 2009). 1.2. Objectives Surface roughness elements (e.g., vegetation and microtopography) vary spatially as a function of landscape evolution processes in natural environments. As a result, wind flow and aeolian transport vary spatially. Spatial patterns associated with aeolian transport and the interaction of roughness elements with aeolian processes, in general, are not well understood in heterogeneous environments at landscape scales (i.e., spanning a continuum of disturbed and undisturbed surfaces) (Okin et al., 2006). The objectives of our study were therefore to (i) describe spatial patterns of surface change (i.e., temporal variability in the relative elevation of the soil surface indicative of deflation, inflation, or no change), and (ii) determine relationships between LiDAR-derived surface roughness and surface change, in recently burned cold desert shrub steppe. We hypothesized that (i) surface change would be significantly related to surface roughness, and (ii) the relationship would indicate that smooth, burned surfaces are characterized by erosion and that adjacent, rough unburned surfaces are characterized by deposition at the landscape scale. Additional analysis was conducted to determine whether patterns of surface change and the relationship between surface roughness and surface change, observed at a landscape scale, were consistent with observations at finer spatial scales. 2. Study area This study was conducted in shrub steppe rangelands of the eastern Snake River plain (SRP), Idaho, USA (Lat. 43°30′ N., Lon. 112°38′ W., 1650 m elevation) from Fall 2007 through Fall 2008 (Fig. 1). Geomorphology of the SRP is characterized by basalt lava flows and calderas that lie along a migratory path of volcanism that originated ∼ 16 ma in eastern Oregon and currently resides in the Yellowstone plateau (Pierce and Morgan, 1992). Surface soils have predominantly developed in aeolian sediments, which include loess deposited ∼12,000–70,000 years ago, and (to a much lesser extent) sand dunes as well as fluvial sands adjacent to the Snake River that have been reworked by wind (Busacca et al., 2004). The geomorphic setting of the study area was described in further detail in Sankey et al. (2009b). The aeolian surfaces studied included areas burned by one of two wildfires (Twin Buttes fire — 3819 ha July 2007, and Moonshiner fire — 1081 ha August 2007) and an adjacent, predominantly

Author's personal copy J.B. Sankey et al. / Geomorphology 119 (2010) 135–145

137

Fig. 1. (A) Location of study relative to Idaho, with inset showing location of Idaho relative to USA. (B) Aerial photograph of the eastern Snake River plain, Idaho, with outline of Twin Buttes and Moonshiner wildfire boundaries, study site locations (S = severely burned, M = moderately burned, and U = unburned), and bare-earth digital elevation model for LiDAR data collection area. (C) Schematic of erosion bridge locations within a 100-m radius hypothetical study site. At each site, erosion bridge transects 3 and 1 were oriented along an axis from SW–NE, and transects 2 and 4 were oriented from SE–NW. Interval values denote distance of erosion bridge inner post from the center of the study site.

downwind unburned area. The Twin Buttes fire produced a severely burned landscape with almost no vegetation remaining. The area burned by the Moonshiner fire was less severely burned, with little herbaceous vegetation but a greater presence of burned sagebrush and juniper remaining following fire. All surfaces studied were characterized by silt loam-textured soils originally deposited as loess (NCSS Web Soil Survey, 2008). Threshold wind speeds (the minimum windspeed required to initiate saltation) were measured using the method developed by Stout (2004, 2007) during Fall 2007 on severely and moderately burned surfaces and could not be determined at an unburned surface because of very low amounts of saltation (Sankey et al., 2009b). Winds exceeding the approximate measured lower limit of threshold (5 m s− 1) and middle value of threshold (8 m s− 1) predominantly trended from the SW to the NE from Fall 2007 to Fall 2008 according to windspeed and orientation data (10 m height) acquired from a weather station located 5 km NE of the study location (Fig. 2). A major source of surface variability within burned and unburned areas was the pattern of playette and coppice surface microtopography, which is typical of shrub steppe rangelands of the SRP (Hilty et al., 2003). Playettes are intershrub spaces that are generally unvegetated and have a crusted, vesicular soil surface. Coppices are

Fig. 2. Wind direction for wind speeds N5 and 8 m s− 1 from Fall 2007 to Fall 2008. Chart demonstrates that winds predominantly trended SW–NE. Wind data were acquired from the Idaho National Laboratory (collected at 10 m height); 5 and 8 m s− 1 correspond to approximate minimum and middle values, respectively, of threshold wind speed for aeolian transport previously measured by Sankey et al. (2009b) on burned surfaces at the study location.

Author's personal copy 138

J.B. Sankey et al. / Geomorphology 119 (2010) 135–145

small, vegetated mounds that have a less cohesive surface soil structure. Following wildfire, coppices in the study area were vegetated with shrub and herbaceous plants: commonly in the unburned area, occasionally in the moderately burned area (Moonshiner), and rarely in the severely burned area (Twin Buttes). A survey conducted in summer 2008 estimated that approximately half of the landscape was covered by coppice and half by playette and indicated that both microtopographic units were generally anisotropic in shape (Amber Hoover, M.S. thesis in preparation, Idaho State University). Mean (SE) (n = 24) playette dimensions were 1.66 (0.16) m along the longest axis and 0.86 (0.12) m along the perpendicular axis (Amber Hoover, M.S. thesis in preparation, Idaho State University). In a study on the SRP, Hilty et al. (2003) estimated that the diameter of playettes ranged from 0.5 to 5 m, and that coppices were up to 3 m in diameter depending on whether they developed beneath one or more tightly spaced shrubs. 3. Methods 3.1. Surface change measurements Erosion bridges were installed beginning in August 2007 to monitor soil surface change at five locations in the area burned by the Twin Buttes fire (henceforth collectively referred to as severely burned study sites), three locations in the area burned by the Moonshiner fire (henceforth moderately burned study sites), and six locations in a nearby unburned area (henceforth unburned study sites) (Fig. 1). Twenty erosion bridges were installed at each study site, oriented such that samples were collected on axes perpendicular and parallel to predominant wind direction (Figs. 1 and 2). Erosion bridge positions were recorded using a Trimble GeoXT GPS receiver. Position coordinates were subsequently differentially corrected and estimated to average b1 m horizontal accuracy. Each erosion bridge consisted of two 0.7-m lengths of rebar driven ∼0.4 m into the ground, 1.2 m apart. Using a 1.2m-long carpenter's level, the height of each rebar post was carefully adjusted to ensure that a level plane existed between the tops of the two posts. Eleven measurements of the height of the carpenter's level above the ground surface were taken at 0.1-m increments along each erosion bridge (i.e., the first and eleventh measurement were separated by 1.0 m) on the installation date in Fall 2007. Basal cover, a measure of cover at the soil surface (i.e., either soil or the rooted portion of herbaceous or shrub vegetation) was recorded at each erosion bridge measurement. Erosion bridge height and basal cover measurements were repeated during 16–19 October 2008. Rate of surface change (mm y− 1) for each measurement location was estimated by subtracting the 2008 height value from the 2007 height value, dividing by the number of days between the two measurement dates, and multiplying by 365 days. In order to minimize the effects of physical and biological processes (e.g., cryoturbation, shrink/swell due to cycles of wetting and drying, bioturbation) that might have impacted relative surface elevations during the study period, measurements in Fall 2008 at several individual locations and/or entire erosion bridges were either not performed or subsequently removed from analysis because of: (i) rebar posts that were no longer in level alignment, (ii) posts that could not be located in unburned areas of dense vegetation, or (iii) observation of an animal hole or mound at the measurement location. Additionally, individual estimates of surface change indicating an absolute value N50 mm were removed from analysis because they were identified as extreme values likely to have been either erroneously recorded in the field or indicative of newly formed animal mounds, holes, and other forms of bioturbation that were not properly noted when the measurements were made. The number of measurements (and erosion bridges) in the final data set analyzed were 1094 (97) at the severely burned area, 659 (59) at the moderately burned area, 1278 (111) at the unburned area.

In order to estimate measurement error associated with our erosion bridge measurements, the repeatability of surface change measurements was assessed on the second measurement date for 10 bridges (i.e., N = 110 measurements) located in burned sites and 10 bridges located in unburned sites by calculating a standard error of the lab (SEL): sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 ∑ðY1 −Y2 Þ SEL = 2N where (Y1, Y2) are duplicate reference analysis and N is the number of replicate pairs. 3.2. LiDAR data and processing LiDAR data were acquired in November 2007 for an ∼60-km2 area covering portions of the Twin Buttes fire, Moonshiner fire, adjacent unburned area, and including all of the surface change measurement locations (Fig. 1). LiDAR acquisition was performed by the National Center for Airborne Laser Mapping (NCALM) using a Gemini ALTM© laser scanner mounted on a fixed-wing aircraft flying at ∼700 m AGL. The sensor acquired data with a pulse rate frequency of 71 kHz and a scan frequency of 40 Hz. Average downtrack and crosstrack point spacing were 0.75 and 0.74 m, respectively. The data were collected in 21 flight lines that averaged 537 m width with ∼50% overlap. Average point density of the entire data set (i.e., overlapping flight lines) was 3.63 points m− 2 and ranged from ∼1 point m− 2 in locations of no overlap to N6 points m− 2 in locations with multiple overlapping flight lines. Vertical accuracy of the LiDAR data were evaluated by NCALM using 2161 check points collected along the paved surface of U.S. Highway 20 located near the northern boundary of the LiDAR collection area. Check points were collected with a vehicle-mounted GPS antenna (Ashtech 700700.c Marine antenna). Comparison of check points with nearest neighbor LiDAR points indicated that the LiDAR data had relative vertical bias of −0.050 m with a scatter of 0.051 m. The LiDAR data consisted of up to four returns per LiDAR pulse. The first two returns were included in the analysis for the sake of processing efficiency, and the third and fourth returns comprised a small and likely insignificant fraction (≤0.1%) of the data sets. LiDAR data were spatially subset by study site, with one 200-m by 200-m subset centered on each study site. Surface roughness was characterized using previously developed and described methods and LiDAR tools (http://bcal.geology.isu.edu/envitools/index.html; Glenn et al., 2006; Mundt et al., 2006; Streutker and Glenn, 2006). In this study, surface roughness is defined as the standard deviation of all LiDAR point elevations (i.e., reflected from ground and vegetation) within an area (raster cell) of specified dimensions. LiDAR elevations were first detrended to remove the effects of coarser scale topographic slope (Davenport et al., 2004). Surface roughness was determined for each subset at 1-, 2-, 3-, and 5-m raster cell resolution. A single surface roughness value was estimated for each raster cell. 3.3. Scales of analysis Comparison of data amongst severely burned, moderately burned, and unburned areas represented the predominant and coarsest scale of analysis for this study, which is henceforth referred to as landscape scale. At the landscape scale, analyses were performed for surface change data aggregated by erosion bridge as well as aggregated by individual severely burned, moderately burned, and unburned sites and by severely burned, moderately burned, and unburned areas. Surface change data were also analyzed at two spatial scales that were finer than landscape scale. These finer spatial scales are henceforth termed among-playette/coppice scale and within-playette/coppice scale. Analyses at among-playette/coppice scale focused on individual

Author's personal copy J.B. Sankey et al. / Geomorphology 119 (2010) 135–145

burned or unburned areas (i.e., either severely burned, moderately burned, or unburned) and were generally performed for surface change data aggregated by erosion bridge. At within-playette/coppice scale, analyses focused on surface change measurements with separation distances of 0.1–1.0 m (i.e., measurements made within individual erosion bridges). The major source of surface variability at within-playette/coppice scale was that of individual coppice or playette units or the transition zone between the two types of microtopographic units. LiDAR surface roughness and relationships between LiDAR surface roughness and surface change were analyzed at the landscape and among-playette/coppice scales. The LiDAR data collected for this project did not have a sufficiently fine spatial resolution for application to within-playette/coppice scales. 3.4. Analysis of surface change and LiDAR surface roughness The inverse of LiDAR-derived surface roughness values were related to corresponding field-based surface change measurements at the landscape and among-playette/coppice scales using correlation and least squares regression (SPSS 15.0 for Windows) as well as quantile regression analysis (QUANTREG package, R: A language and environment for statistical computing). The inverse transformation was used on surface roughness to linearize its relationship with surface change as well as to stabilize the variability in surface change. Quantile regression is a useful analysis technique when a response variable is expected to vary at different rates as a function of the predictor variable, depending on what portion of the response variable distribution is analyzed (Cade and Noon, 2003). Because the response variable (surface change) encapsulated two different physical processes (i.e., erosion and deposition), we anticipated that values representative of the higher quantiles (τ) of surface change (values predominantly indicating deposition) might vary differently as a function of surface roughness compared to values representative of the lower quantiles (values predominantly indicating erosion) of surface change conditional on surface roughness. Plots of surface change and LiDAR-derived surface roughness vs. corresponding basal cover measurements were examined qualitatively. Directional semivariograms were constructed along the SW– NE and SE–NW axes (perpendicular and parallel to predominant wind direction; Figs. 1 and 2) of surface change measurements aggregated for each of the severely burned, moderately burned, and unburned areas (Geostatistical Analyst, ArcGIS 9.3). The directional semivariograms were constructed with (i) a 5-m lag spacing and maximum separation distance of 40 m and (ii) a 0.1-m lag spacing and maximum separation distance of 1.0 m, as representations of the spatial autocorrelation structure of surface change at the among- and within-playette/coppice scales, respectively.

4.2. Spatial patterns of surface change Experimental semivariograms demonstrated a random spatial pattern of surface change at the among-playette/coppice scale in severely and moderately burned areas (Fig. 5). In unburned areas, surface change at among-playette/coppice and within-playette/ coppice scales occurred with a spatially random pattern as well, indicated by experimental semivariograms that appeared to have a relatively linear-flat structure (Figs. 5 and 6). Surface change in burned areas demonstrated strong spatial autocorrelation structure at the within-playette/coppice scale (Fig. 6). The spatial patterns were directional in the severely burned area with a larger range for measurements oriented perpendicular (in comparison to parallel) to predominant wind directions. At the within-playette/coppice scale,

Site

4.1. Surface change Vegetation in the burned areas in Fall 2007 consisted of burned shrubs, with no herbaceous vegetation detected at the landscape scale Table 1 Vegetation basal cover aggregated for burned and unburned areas.

Severe burn Moderate burn Unburned

(Table 1). Some vegetation, predominantly herbaceous, had regrown in the burned areas by Fall 2008 (Table 1). Mean basal shrub cover ranged from 0 to 2% at the severely and moderately burned sites, respectively, in Fall 2007 (Table 2). Unburned sites spanned a wider range of vegetation with mean total basal cover (shrub + herbaceous) ranging from 9 to 28% amongst sites in Fall 2007 (Table 2). The severely burned area had mean (SE) rate of surface change of −2.1 (0.2) mm y− 1, indicating net deflation of the soil surface; while the moderately burned area had mean (SE) rate of surface change of −0.1 (0.2) mm y− 1 (Fig. 3). The unburned area, which was downwind of the burned areas, had mean (SE) rate of surface change of 1.5 (0.1) mm y− 1, indicating net inflation (Fig. 3). Standard error of the lab values, a measure of repeatability for surface change measurements, were smaller for burned surfaces (SEL = 1.3 mm) likely because the lack of vegetation afforded greater accuracy in repeat measurements compared to the unburned surfaces (SEL = 2.3 mm). Some of the variability in surface change observed at the among-playette/coppice scale (Fig. 3) might have been influenced by the installation date of study sites (Table 2). For example, substantially large deflation was observed for the first severely burned site to be installed (S1), whereas substantially large inflation was observed for the first installed unburned site (U1) (Fig. 3). Plots of surface change with basal cover measured in Fall 2007 demonstrate a trend from deflation to inflation with increasing shrub vegetation, in general, though the largest values of mean inflation were observed for sites with low-intermediate shrub basal cover (Fig. 4). A similar relationship was observed between surface change and basal cover when total (shrub + herbaceous) vegetation basal cover was considered for the unburned sites (Fig. 4).

Table 2 Install date and vegetation aggregated for severe (S), moderate (M), or unburned (U) sites.

4. Results

Area

139

Mean (SE) basal cover % fall 2007

Mean (SE) basal cover % fall 2008

Shrub

Herbaceous

Shrub

Herbaceous

0.5 (0.2) 1.4 (0.4) 7.6 (0.7)

0.0 (0.0) 0.0 (0.0) 8.6 (0.8)

0.8 (0.3) 1.4 (0.4) 8.2 (0.7)

5.4 (0.7) 5.3 (0.9) 10.1 (0.8)

S1 S2 S3 S4 S5 M1 M2 M3 U1 U2 U3 U4 U5 U6

Install date

8/23/2007 8/29/2007 9/18/2007 9/19/2007 10/4/2007 8/28/2007 8/29/2007 9/19/2007 8/28/2007 8/30/2007 8/31/2007 9/6/2007 10/2/2007 10/3/2007

Mean (SE) basal cover % fall 2007 Shrub

Herbaceous

0.0 (0.0) 0.4 (0.4) 0.0 (0.0) 2.3 (1.0) 0.0 (0.0) 0.4 (0.4) 1.4 (0.8) 2.3 (1.0) 4.1 (1.3) 2.2 (1.0) 14.5 (2.4) 11.8 (2.2) 6.4 (1.6) 6.8 (1.7)

0.0 (0.0) 0.0 (0.0) 0.0 (0.0) 0.0 (0.0) 0.0 (0.0) 0.0 (0.0) 0.0 (0.0) 0.0 (0.0) 7.3 (2.0) 6.4 (1.6) 10.0 (2.0) 16.4 (2.5) 6.8 (1.7) 4.5 (1.4)

Author's personal copy 140

J.B. Sankey et al. / Geomorphology 119 (2010) 135–145

Fig. 3. Mean (with standard error bars) surface change and LiDAR-derived surface roughness measurements for severely burned, moderately burned, and unburned areas and for individual study sites within these areas. Surface roughness was calculated as the standard deviation of all LiDAR point heights (ground + vegetation) per raster cell (2-m resolution) after the point heights were detrended for topographic slope.

semivariogram models for burned surfaces had ranges b0.85 m (Fig. 6). This suggests that patterns of surface change occurred within landscape units (i.e., length scales) b0.85 m in length and/or width on burned surfaces. 4.3. LiDAR surface roughness At the landscape scale, the severely burned area had (on average) small surface roughness values, the unburned area had large surface roughness values, and the moderately burned area had intermediate values (Fig. 3). Roughness increased with increasing shrub basal cover at the landscape scale, in general, though the largest mean roughness values were observed for unburned sites with low-intermediate shrub basal cover (Fig. 4). A similar relationship was observed between surface roughness and basal cover when total (shrub + herbaceous) vegetation basal cover was considered for the unburned sites (Fig. 4). Site S1 had small roughness values relative to the other four severely burned sites, and site U2 had large roughness values relative to the other unburned sites, providing examples of variability in LiDAR roughness at the among-playette/coppice scale (Fig. 3). Relationships of surface roughness values amongst sites and amongst the three surface types were consistent for rasters processed at 1-, 2-, 3-, and 5m cell size. 4.4. LiDAR surface roughness and surface change

Fig. 4. Mean (with standard error bars) surface change and LiDAR-derived surface roughness (refer to Fig. 3) vs. shrub vegetation basal cover (Fall 2007) for severely burned, moderately burned, and unburned study sites.

Surface change was strongly, linearly and negatively related to the inverse of surface roughness derived from LiDAR at the landscape scale (Fig. 7). The strongest relationship (r2) using least squares regression was observed with raster cell size of 2 m, though results were very similar at 1-, 3-, and 5-m raster cell size (Table 3). The relationship implies that smoother surfaces (burned areas) were

Author's personal copy J.B. Sankey et al. / Geomorphology 119 (2010) 135–145

141

Fig. 5. Autocorrelation structure of surface change measurements at the among-playette/coppice scale, as depicted by directional semivariograms constructed for erosion bridge data aggregated by severely burned, moderately burned, and unburned areas with 5-m lag and 40-m maximum separation distance. Semivariograms depict the spatial dependence (semivariance: y-axis) of samples as a function of separation distance (distance: x-axis). Smaller semivariance indicates greater relative spatial dependence (autocorrelation).

characterized by deflation, while rougher surfaces (unburned) were characterized by inflation. The modeled relationship suggests that a transition from deflation (i.e., erosion) to inflation (i.e., deposition) occurred at a surface roughness of ∼ 0.05 m (i.e., inverse surface roughness ∼ 20 m− 1; Fig. 7). Although the first sites installed in the severely burned and unburned areas appeared to be potentially influential and outlying points, when these points were removed from analysis the model fit only improved very slightly (r2 = 0.78, p b 0.00, for 2-m raster cell size). At the landscape scale, quantile regression analysis of erosion bridge mean surface change versus inverse surface roughness indicated that changes in surface roughness corresponded to proportionally greater changes in surface change for lower τ values (i.e., erosion) in comparison to higher τ values (i.e., deposition), where τ values signify quantiles of surface change conditional on surface roughness (Fig. 8). The estimated slopes (β1) for τ = 0.10 and 0.25 (β1 = −0.27 and −0.21, respectively) were 2–4 times as large as the estimated slopes for τ = 0.50, 0.75, and 0.90 (β1 = −0.10, −0.08,

and − 0.06, respectively), for example (Fig. 8). Quantile regression slope and intercept coefficients were significant for τ = 0.1, 0.25, 0.5, 0.75, 0.90 [all p b 0.05 and SE(β1) = 0.06, 0.03, 0.03, 0.02, 0.02, respectively]. Overall, findings suggested that the effect of surface roughness on aeolian transport was greater for erosion versus deposition processes. Relationships between surface roughness and surface change at the among-playette/coppice scale were consistent with, though not as strong as, observations at the landscape scale. Mean surface change (aggregated by erosion bridge) was negatively related with inverse surface roughness; and the relationship was significant within the unburned and severely burned areas, but not the moderately burned area [severe burn: Pearson correlation coefficient (R) = − 0.24, p = 0.01; moderate burn: R = − 0.21, p = 0.11; unburned: R = −0.22, p = 0.02]. When performed within the burned and unburned areas, quantile regression analysis of surface change versus inverse surface roughness produced models with intercept and/or slope coefficients that were not statistically significant (p N 0.05).

Author's personal copy 142

J.B. Sankey et al. / Geomorphology 119 (2010) 135–145

Fig. 6. Autocorrelation structure of surface change measurements at the within-playette/coppice scale, as depicted by directional semivariograms (refer to Fig. 5) constructed for erosion bridge data aggregated by severely burned, moderately burned, and unburned areas with 0.1-m lag and 1.0-m maximum separation distance.

5. Discussion 5.1. Surface change — landscape scale A simple calculation suggests that 1.0 × 108 and 1.3 × 105 kg of soil might have been mobilized by wind from burned surfaces at the Twin Buttes and Moonshiner fires, respectively (mean surface change × area burned × assumed bulk density of 1.25 g/cm3). Results imply that a net loss of soil occurred from the burned areas and that

Table 3 Site mean surface change (y) vs. site mean inverse LiDAR surface roughness (x).

Fig. 7. Relationship of mean surface change aggregated by site with the inverse of LiDAR-derived surface roughness (refer to Fig. 3) amongst severely burned, moderately burned, and unburned sites.

Raster cell size

Model

r2 (p)

MSE

SE (estimate)

1m 2m 3m 5m

y = − 0.3x + 4.9 y = − 0.3x + 5.6 y = − 0.3x + 4.8 y = − 0.3x + 4.2

0.76 0.77 0.75 0.73

1.03 0.99 0.88 1.18

1.02 0.99 0.94 1.09

(0.00) (0.00) (0.00) (0.00)

Author's personal copy J.B. Sankey et al. / Geomorphology 119 (2010) 135–145

143

5.2. Surface change — finer scales

Fig. 8. Mean surface change aggregated by erosion bridge vs. the inverse of surface roughness (refer to Fig. 3) for corresponding nearest neighbor LiDAR pixels amongst severely burned, moderately burned, and unburned areas. Values for τ represent the quantile of the surface change distribution analyzed conditional on the inverse of surface roughness; values for β are the quantile regression slope coefficients.

some soil particles were likely transported to the unburned area that was located downwind over the course of one year, post-fire. Surface change measurement locations in the severely burned area were positioned north of measurement locations in the unburned area due to logistical constraints, which limited our scope of inference concerning the transport of sediment from the burned to unburned areas. We did visually observe clouds of dust that traveled from the burned to the unburned areas during the course of the year-long study, however. Localized erosion and deposition occurred within burned and unburned areas, as well (Fig. 8). The small net deflation observed within the Moonshiner fire and net inflation observed within the unburned area might not have been different than no change when considered with regard to measurement error (SEL). The mean rate of surface deflation observed in the severely burned area (− 2.1 mm y− 1) appeared smaller than that measured by Whicker et al. (2002) who estimated a mean of − 5.8 mm 162 d− 1 from six erosion bridges in a burned semiarid shrubland in New Mexico. A mean inflation of 1.9 mm 162 d− 1 was estimated by Whicker et al. (2002) from six erosion bridges in an unburned area, which appears larger than the range observed at the unburned sites we studied (0.81–3.54 mm y− 1). To what extent inflation at the unburned site studied by Whicker et al. (2002) was resultant of background atmospheric deposition versus having been transported from the burned area is unclear. The net inflation we observed in the unburned area appears to be substantial relative to background atmospheric deposition that might be expected in the absence of local wildfire and subsequent aeolian transport. Analyses and reviews of dust deposition rates in the western USA measured with sediment traps have found ranges of 1–48 g m− 2 y− 1 for suspension-sized particles at sites across seven states (Reheis and Kihl, 1995), and 2– 20 g m− 2 y− 1 for long-term monitoring sites in the southern Great Basin and Mojave desert, USA (Reheis, 2006). Dividing the rates presented by Reheis and Kihl (1995) by a particle density of 2.65 × 106 g m− 3 suggests a range of background atmospheric deposition of 0.0004–0.02 mm (in units of thickness) per year. In contrast, the net rate of inflation we observed in the unburned area was 1.5 mm y− 1. How deposition rates measured with sediment traps such as those used by Reheis and Kihl (1995) translate to vertical accumulation of sediment across an actual landscape surface, however, is uncertain as such measurements represent a lower limit of background atmospheric dust that might accumulate in soil (M.C. Reheis, United States Geological Survey, personal communication, 2009).

Aeolian transport resulted in zones of erosion and deposition that were comparable in size or smaller (average dimensions b0.85 × 0.85 m as indicated by semivariogram ranges; Fig. 6) than the surface microtopography in burned areas (e.g., mean playette dimensions = 1.66 × 0.86 m). The zones of erosion and deposition were anisotropic in shape on the severely burned surfaces, with a long axis oriented perpendicular to prevailing winds. In contrast, erosion and deposition occurred in unburned areas with a random spatial distribution. Based on these results, erosion and deposition appeared to occur with a strong spatial pattern at scales comparable to and finer than the microtopography on burned, but not unburned surfaces. Our findings at finer scales appear analogous to those of Whicker et al. (2002) who found differences in surface change resultant from wind erosion in shrub versus intershrub patches, in burned but not unburned areas. Our findings at finer scales appear counter to the conceptual model for aeolian transport in burned shrublands and shrub-encroached grasslands presented by Ravi et al. (2007, 2009) and Ravi and D'Odorico (2009). In their model, the microtopography of burned landscapes was homogenized by the removal of sediment by wind from the raised coppice surfaces and deposition in the lower interspaces, whereas our findings suggested that substantial erosion and deposition might have occurred within burned coppices and interspaces in our study environment. Similarly, our findings also appear counter to those of Hilty et al. (2003) who found that the regular pattern of coppice and playette microtopographic units evolved to a flatter, disturbed surface following burning in a shrub steppe landscape in Idaho, similar to our study area. Hilty et al. (2003), however, analyzed the evolution of surfaces over timescales of multiple years, post-fire; and the evolutionary mechanism they proposed included a combination of burning, livestock trampling, cheatgrass (Bromus tectorum L.) invasion, and water erosion, though not aeolian transport. The areas we examined were not grazed during the course of the study, and cheatgrass has not invaded the study areas yet. The degree to which post-fire aeolian transport serves to increase either the heterogeneity or homogeneity of surface microtopography through the redistribution of sediment is important, as it can directly influence the cycling and spatial distribution of biologically important nutrients, as well as the evolution of plant communities that affects surface roughness and provides feedback to aeolian processes (Su et al., 2006; Ravi et al., 2007; Li et al., 2008).

5.3. Surface change and LiDAR surface roughness Characterization of surface (ground + vegetation) roughness with LiDAR indicated that severely burned surfaces were generally smoother than moderately burned surfaces, which were in turn smoother than unburned surfaces. Results suggested that LiDAR detected variability in roughness that existed largely because of the degree of vegetation present and/or absent amongst the three types of surfaces. Surface change varied as a function of surface roughness at the landscape scale, and the modeled relationship supported the hypothesis that smooth, burned surfaces are characterized by deflation and that adjacent rough, unburned surfaces are characterized by inflation. The transition from erosion to deposition in the modeled relationship (Fig. 7) occurred at surface roughness ∼0.05 m (i.e., inverse surface roughness ∼ 20 m− 1) that approximately corresponded with the threshold in roughness between burned and unburned surfaces. Streutker and Glenn (2006) found burned and unburned shrub steppe to have mean vegetation roughness of ∼ 0.05 and 0.09 m, respectively, in a nearby study location. The modeled relationship appeared in accord with the inverse relationship between surface roughness and sand entrainment that Pelletier et al. (2009) employed for modeling coastal dune migration.

Author's personal copy 144

J.B. Sankey et al. / Geomorphology 119 (2010) 135–145

Quantile regression analysis indicated that incremental changes in surface roughness were related to proportionally greater corresponding changes in erosion compared to deposition. Quantile regression analysis suggested that deposition increased gradually with increased roughness, whereas erosion increased rapidly with decreased roughness. Erosion is generally expected to increase with decreased vegetation cover and density, though complexity in these relationships exists at lower densities of vegetation because of the spatial structure of turbulence and related potential for increased erosivity adjacent to individual roughness elements (e.g., shrubs) (Fryrear, 1985; Findlater et al., 1990; Lee, 1991a,b; Raupach et al., 1993). Deposition, in contrast, generally occurs in shrublands when sediment is trapped by vegetation and is greatest at intermediate levels of vegetation porosity (Grant and Nickling, 1998; Raupach et al., 2001; Lee et al., 2002; Okin et al., 2006). While we did not relate surface change or surface roughness to vegetation porosity, we did relate these variables to field measurements of vegetation basal cover, a measure of vegetation abundance. Results implied that both surface roughness and sediment deposition were greatest for low to intermediate values of vegetation basal cover. Breshears et al. (2009) suggested that aeolian transport is greatest for intermediate levels of shrub cover in undisturbed but not disturbed shrublands. Our plotted data suggested that the greatest deposition corresponded with intermediate values of surface roughness (e.g., inverse surface roughness ∼ 15–30 m− 1 in Fig. 8), though we did not specifically model a nonlinear relationship for the upper quantiles of surface change. We also did not specifically model separate relationships for undisturbed and disturbed surfaces. All of the surfaces on which we detected deposition were likely subjected to sediment transported from burned surfaces that were upwind as well as background atmospheric deposition. Relationships between surface roughness and surface change at the among-playette/coppice scale were consistent with, though not as strong as, at the landscape scale. The low correlation might be resultant of several sources of measurement error, including the average: vertical (∼5 cm) and horizontal (b1 m) errors in LiDAR measurements, horizontal error (b1 m) in erosion bridge GPS locations, and vertical error (∼1–2 mm) in erosion bridge surface change measurements. We also acknowledge that erosion bridges are designed to measure surface change irrespective of whether the change was driven by wind or water transport. Some water-driven transport certainly might have occurred in erosion bridge locations during the study, however, based on our previous research (Sankey et al., 2009a,b) and observations in the field sediment transport was predominantly driven by wind in the burned environment that we studied. In light of the findings of spatial patterns of aeolian transport at finer-than-landscape scales, future research to examine relationships of aeolian transport with surface roughness from higher resolution LiDAR (e.g., increased point density or ground-based LiDAR systems) is needed. 6. Conclusion LiDAR characterization of the roughness of burned and unburned surfaces suggested that the reduction of vegetation by fire produced surfaces that were relatively smooth compared to unburned surfaces in shrub steppe at a landscape scale. Surface change because of subsequent aeolian transport varied as a function of surface roughness at the landscape scale. Burned surfaces that were smooth were characterized by net erosion, and the adjacent unburned surfaces that were rough were characterized by net deposition. Furthermore, erosion appeared to vary at a greater rate as a function of surface roughness compared to deposition amongst burned and unburned surfaces. This study also found that predominant patterns of surface change occurred at and within the scale of microtopography on burned surfaces. These findings are important for understanding

feedbacks between soil processes and vegetation establishment, as well as longer-term evolution of post-fire surfaces in shrublands of cold desert environments such as the Snake River plain. Future research that further elucidates the relationships of aeolian erosion and deposition processes with surface roughness at microtopographic to landscape scales and over longer time periods is required, as these processes can drive the redistribution of sediment within and between disturbed and undisturbed landscapes. Such processes have important implications for landscape connectivity and cycling of sediment, nutrients, and pollutants within and amongst geomorphic, ecological, and management boundaries. Acknowledgments This material is based on work supported in part by the U.S. Army Research Laboratory and the U.S. Army Research Office under grant number W911NF-07-1-0481. LiDAR data were provided through a seed grant from the National Center for Airborne Laser Mapping (NCALM) supported by NSF. Additional support was provided by an Inland Northwest Research Alliance Ph.D. research fellowship. S.M. Stoller Corporation provided logistical and technical support at the Idaho National Laboratory. The authors specifically wish to acknowledge Dr. Roger Blew and Jeremy Shive of S.M. Stoller Corporation for their support of this project. Amber Hoover and Eli Eversole assisted in field measurements. References Ash, J.E., Wasson, R.J., 1983. Vegetation and sand mobility in the Australian desert dunefield. Z. Geomorphol. Suppl. 45, 7–25. Baas, A.C.W., 2007. Complex systems in aeolian geomorphology. Geomorphology 91, 311–331. Baas, A.C.W., 2008. Challenges in aeolian geomorphology: investigating aeolian streamers. Geomorphology 93, 3–16. Bagnold, R.A., 1941. The Physics of Blown Sand and Desert Dunes. Methuen, New York. Bauer, B.O., 2009. Contemporary research in aeolian geomorphology. Geomorphology 105, 1–5. Bourke, M.C., Ewing, R.C., Finnegan, D., McGowan, H.A., 2009. Sand dune movement in the Victoria valley, Antarctica. Geomorphology 109, 148–160. Breshears, D.D., Whicker, J.J., Johansen, M.P., Pinder, J.E., 2003. Wind and water erosion and transport in semi-arid shrubland, grassland and forest ecosystems: quantifying dominance of horizontal wind driven transport. Earth Surf. Process. Land. 28, 1189–1209. Breshears, D.D., Whicker, J.J., Zou, C.B., Field, J.P., Allen, C.D., 2009. A conceptual framework for dryland aeolian sediment transport along the grassland–forest continuum: effects of woody plant canopy cover and disturbance. Geomorphology 105, 28–38. Busacca, A.J., Beget, J.E., Markewich, H.W., Muhs, D.R., Lancaster, N., Sweeney, M.R., 2004. Eolian sediments. In: Gillespie, A.R., Porter, S.C., Atwater, B.R. (Eds.), The Quaternary Period in the United States. Elsevier, Amsterdam, The Netherlands, pp. 275–310. Cade, B.S., Noon, B.R., 2003. A gentle introduction to quantile regression for ecologists. Front. Ecol. 1, 412–420. Cornelis, W.M., 2006. Hydroclimatology of wind erosion in arid and semiarid environments. In: D'Odorico, P., Porporato, A. (Eds.), Dryland Ecohydrology. Springer, Dordrecht, The Netherlands, pp. 141–161. Davenport, I.J., Holden, N., Gurney, R.J., 2004. Characterizing errors in airborne laser altimetry data to extract soil roughness. IEEE Trans. Geosci. Remote Sens. 42, 2130–2141. Ewing, R.C., Kocurek, G., 2010. Aeolian dune-field pattern boundary conditions. Geomorphology 114, 175–187. Findlater, P.A., Carter, D.J., Scott, W.D., 1990. A model to predict the effects of prostrate ground cover on wind erosion. Aust. J. Soil Res. 28, 609–622. Frankel, K.L., Dolan, J.F., 2007. Characterizing arid region alluvial fan surface roughness with airborne laser swath mapping digital topographic data. J. Geophys. Res. 112 (F02025)doi:10.1029/2006JF000644. Fryrear, D.W., 1985. Soil cover and wind erosion. Trans. ASAE 28, 781–784. Glenn, N.F., Streutker, D.R., Chadwick, D.J., Thackray, G.D., Dorsch, S.J., 2006. Analysis of LiDAR-derived topographic information for characterizing and differentiating landslide morphology and activity. Geomorphology 73, 131–148. Grant, P.F., Nickling, W.G., 1998. Direct field measurement of wind drag on vegetation for application to windbreak design and modelling. Land Degrad. Dev. 9, 57–66. Hilty, J.H., Eldridge, D.J., Rosentreter, R., Wicklow-Howard, M.C., 2003. Burning and seeding influence soil surface morphology in an artemisia shrubland in southern Idaho. Arid Land Res. Man. 17, 1–11. Lee, J., 1991a. The role of desert shrub size and spacing on wind profile parameters. Phys. Geogr. 12, 72–89. Lee, J., 1991b. Near-surface wind flow around desert shrubs. Phys. Geogr. 12, 140–146.

Author's personal copy J.B. Sankey et al. / Geomorphology 119 (2010) 135–145 Lee, S., Park, K., Park, C., 2002. Wind tunnel observations about the shelter effect of porous fences on the sand particle movements. Atmos. Environ. 36, 1453–1463. Li, J., Okin, G.S., Alvarez, L., Epstein, H., 2008. Effects of wind erosion on the spatial heterogeneity of soil nutrients in two desert grassland communities. Biogeochemistry 88, 73–88. McKean, J., Roering, J., 2004. Objective landslide detection and surface morphology mapping using high-resolution airborne laser altimetry. Geomorphology 57, 331–351. Menenti, M., Ritchie, J.C., 1994. Estimation of effective aerodynamic roughness of Walnut Gulch watershed with laser altimeter measurements. Water Resour. Res. 30, 1329–1337. Mundt, J.T., Streutker, D.R., Glenn, N.F., 2006. Mapping sagebrush distribution using fusion of hyperspectral and LiDAR classifications. Photogramm. Eng. Remote Sens. 72, 47–54. NCSS Web Soil Survey, 2008. Accessed 9/22/2008. http://websoilsurvey.nrcs.usda.gov/ app/. Okin, G.S., Gillette, D.A., Herrick, J.E., 2006. Multi-scale controls on and consequences of aeolian processes in landscape change in arid and semi-arid environments. J. Arid Environ. 65, 253–275. Okin, G.S., Parsons, A.J., Wainwright, J., Herrick, J.E., Bestelmeyer, B.T., Peters, D.C., Frederickson, E.L., 2009. Do changes in connectivity explain desertification? Bioscience 59, 237–244. Pelletier, J.D., Mitasova, H., Harmon, R.S., Overton, M., 2009. The effects of interdune vegetation changes on eolian dune field evolution: a numerical-modeling case study at Jockey's Ridge, North Carolina, USA. Earth Surf. Process. Land. 34, 1245–1254. Pierce, K.L., Morgan, L.A., 1992. The track of the Yellowstone hot spot: volcanism, faulting, and uplift. In: Link, P.K., Kuntz, M.A., Platt, L.B. (Eds.), Regional Geology of Eastern Idaho and Western Wyoming. Geological Society of America, Boulder, CO, USA, pp. 1–53. Rango, A., Chopping, M., Ritchie, J., Havstad, K., Kustas, W., Schmugge, T., 2000. Morphological characteristics of shrub coppice dunes in desert grasslands of southern New Mexico derived from scanning LiDAR. Remote Sens. Environ. 74, 26–44. Raupach, M.R., Gillette, D.A., Leys, J.F., 1993. The effect of roughness elements on wind erosion threshold. J. Geophys. Res. 98 (D2), 3023–3029. Raupach, M.R., Woods, N., Dorr, G., Leys, J.F., Cleugh, H.A., 2001. The entrapment of particles by windbreaks. Atmos. Environ. 35, 3373–3383. Ravi, S., D'Odorico, P., 2009. Post-fire resource redistribution and fertility island dynamics in shrub encroached desert grasslands: a modeling approach. Landscape Ecol. 24, 325–335. Ravi, S., D'Odorico, P., Zobeck, T.M., Over, T.M., Collins, S., 2007. Feedbacks between fires and wind erosion in heterogeneous arid lands. J. Geophys. Res. 112 (G04007) doi:10.1029/2007JG000474. Ravi, S., D'Odorico, P., Wang, L., White, C.S., Okin, G.S., Macko, S.A., Collins, S.L., 2009. Post-fire resource redistribution in desert grasslands: a possible negative feedback on land degradation. Ecosystems 12, 434–444.

145

Reheis, M.C., 2006. A 16-year record of eolian dust in southern Nevada and California, USA: controls on dust generation and accumulation. J. Arid Environ. 67, 487–520. Reheis, M.C., Kihl, R., 1995. Dust deposition in southern Nevada and California, 1984– 1989: relations to climate, source area, and source lithology. J. Geophys. Res. 100 (D5), 8893–8918. Ritchie, J.C., 1995. Airborne laser altimeter measurements of landscape topography. Remote Sens. Envrion. 53, 91–96. Sankey, J.B., Germino, M.J., Glenn, N.F., 2009a. Aeolian sediment transport following wildfire in sagebrush steppe. J. Arid Environ. 73, 912–919. Sankey, J.B., Germino, M.J., Glenn, N.F., 2009b. Relationships of post-fire aeolian transport to soil and atmospheric conditions. J. Aeolian Res. 1, 73–85. Stout, J.E., 2004. A method for establishing the critical threshold for aeolian transport in the field. Earth Surf. Process. Land. 29, 1195–1207. Stout, J.E., 2007. Simultaneous observations of the critical threshold of two surfaces. Geomorphology 85, 3–16. Streutker, D.E., Glenn, N.F., 2006. LiDAR measurement of sagebrush steppe vegetation heights. Remote Sens. Environ. 102, 135–145. Su, Y.Z., Li, Y.L., Zhao, H.L., 2006. Soil properties and their spatial pattern in a degraded sandy grassland under post-grazing restoration, Inner Mongolia, northern China. Biogeochemistry 79, 297–314. Tratt, D.M., Neff, J.M., Valinia, A., 2008. Analysis of laser remote sensing technology needs in the Earth sciences: a decadal-scale outlook. J. Appl. Remote Sens. 2 (023546)doi:10.1117/1.3036940. Vermeire, L.T., Wester, D.B., Mitchell, R.B., Fuhlendorf, S.D., 2005. Fire and grazing effects on wind erosion, soil water content, and soil temperature. J. Environ. Qual. 34, 1559–1565. Wasson, R.J., Nanninga, P.M., 1986. Estimating wind transport of sand on vegetated surfaces. Earth Surf. Process. Land. 11, 505–514. Whicker, J.J., Breshears, D.D., Wasiolek, P.T., Kirchner, T.B., Tavani, R.A., Schoep, D.A., Rodgers, J.C., 2002. Temporal and spatial variation of episodic wind erosion in unburned and burned semiarid shrubland. J. Environ. Qual. 31, 599–612. Whicker, J.J., Pinder, J.E., Breshears, D.D., 2006a. Increased wind erosion from forest wildfire: implications for contaminant-related risks. J. Environ. Qual. 35, 468–478. Whicker, J.J., Pinder, J.E., Breshears, D.D., Eberhart, C.F., 2006b. From dust to dose: effects of forest disturbance on increased inhalation exposure. Sci. Total Environ. 368, 519–530. Wiggs, G.F., Livingstone, I., Thomas, D.S.G., Bullard, J.E., 1994. Effect of vegetation removal on airflow patterns and dynamics in the southwest Kalahari desert. Land Degrad. Rehabil. 5, 13–24. Wiggs, G.F., Thomas, D.S.G., Bullard, J.E., Livingstone, I., 1995. Dune mobility and vegetation cover in the southwest Kalahari. Earth Surf. Process. Land. 20, 515–529. Wiggs, G.F., Livingstone, I., Thomas, D.S.G., Bullard, J.E., 1996. Airflow and roughness characteristics over partially vegetated linear dunes in the southwest Kalahari desert. Earth Surf. Process. Land. 21, 19–34. Wolfe, S.A., Nickling, W.G., 1993. The protective role of sparse vegetation in wind erosion. Prog. Phys. Geogr. 17, 50–68. Zobeck, T.M., Fryrear, D.W., Pettit, R.D., 1989. Management effects on wind-eroded sediment and plant nutrients. J. Soil Water Conserv. 44, 160–163.