The Use of Terrestrial Laser Scanning (TLS)

Journal of Coastal Research

30

1

111–119

Coconut Creek, Florida

January 2014

The Use of Terrestrial Laser Scanning (TLS) in Dune Ecosystems: The Lessons Learned Rusty A. Feagin, Amy M. Williams, Sorin Popescu, Jared Stukey, and Robert A. Washington-Allen Department of Ecosystem Science and Management Texas A&M University College Station, TX 77843-2138, U.S.A. [email protected]

ABSTRACT Feagin, R.A.; Williams, A.M.; Popescu, S.; Stukey, J., and Washington-Allen, R.A., 2014. The use of terrestrial laser scanning (TLS) in dune ecosystems: the lessons learned. Journal of Coastal Research, 30(1), 111–119. Coconut Creek (Florida), ISSN 0749-0208. This paper presents a methodology for using terrestrial laser scanning (TLS) to quantify sand dune geomorphology. As an example of the use of TLS, we present methods that were used to investigate changes in sediment and vegetation volumes after Hurricane Ike. We collected TLS data within a 100 m 3 100 m plot on the East Matagorda Peninsula, Texas, from early September 2008 (before landfall) to early October 2009 (a year after landfall). Terrestrial laser scanning-collected laser point clouds were then interpolated into several grid sizes. From several interpolated grid sizes, 0.50 m 3 0.50 m grids were determined best for analysis as they were able to compromise two competing resolutionrelated issues: gaps caused by vegetation shadows and the natural contours of the dune. We outline several additional lessons to aid coastal researchers in strengthening their own future work: the use of reference survey stakes in an unstable environment, the development of a novel method to test for errors in point cloud registration among multiple dates, how best to interpret sediment and vegetation change analysis as derived from interpolated grids, and suggestions for incorporating mass-based sedimentary and biomass-based vegetation field studies within the volumetric context of TLS analysis.

ADDITIONAL INDEX WORDS: Lidar, sand dune, geomorphology, interpolation, resolution, TLS, scanner.

INTRODUCTION Traditional techniques for evaluating coastal geomorphic changes, such as optical surveying or the use of a total station, have often lacked the ability to quickly capture the three dimensional (3-D) heterogeneity of biophysical or geomorphic structures; they are time-intensive procedures with relatively low resolution. Passive optical remote sensing technology, such as aerial photography and satellite imagery, have improved the reliability and time return on evaluating coastal processes (Brock and Purkis, 2009; Brock et al., 2002; Delgado-Fernandez, Davidson-Arnott, and Ollerhead, 2009; Klemas, 2009; Morton and Sallenger, 2003). Still, there are problems with remotely sensing geomorphic change in terms of the frequency of data collection, the resolution, and the ability to penetrate cloud cover and capture data at night (Li and Liu, 2009; McCulloh and Heinrich, 2009; Ramsey, III, Werle, and Lu, 2009). The use of lidar (light detection and ranging) in remote sensing is relatively new, yet it is an active-sensing technology that uses the round-trip time and known directionality of a DOI: 10.2112/JCOASTRES-D-11-00223.1 received 8 December 2011; accepted in revision 10 May 2012; corrected proofs received 13 August 2012. Published Pre-print online 31 August 2012. Ó Coastal Education & Research Foundation 2014

laser pulse to measure distance and position in relation to a sensor. When coupled with a global positioning system, lidar data can represent a geographic location in 3-D (X, Y, and Z). Lidar data is most commonly collected with the equipment attached to an airplane and is referred to as airborne or aerial lidar. In coastal applications the transmitted laser pulse is intercepted, backscattered, reflected, or absorbed by geomorphic features (e.g., sand), water, vegetation elements (e.g., canopy, branches, leaves, etc.) or man-made features (e.g., buildings, roads, fences, etc.) to produce a 3-D profile of a landscape (Bater and Coops, 2009; Nayegandhi et al., 2006). Accuracy and extensiveness of the profile are dependent on the parameters for data collection, particularly point spacing, which is a product of sensor scan-rate, aircraft altitude, sensor field-of-view, and flight-line spacing. This technology, in conjunction with a well-planned data collection design, is helpful in quickly providing time-sensitive data of hurricane impacts in areas where ground conditions make in-situ data difficult to collect (McCulloh and Heinrich, 2009; Sallenger, Wright, and Lilycrop, 2007). Because of the dynamic nature of coastal systems, a major advantage of airborne lidar over traditional geomorphometric techniques is that the data can be collected synoptically with high temporal frequency. It also can penetrate cloud cover and capture data at night. This results in less work in the field

112

Feagin et al.

while providing a greater likelihood of detecting coastal sediment change (Blott and Pye, 2004; Brock et al., 2002; Gares, Wang, and White, 2006; Robertson et al., 2005; Sallenger et al., 2004; Shrestha et al., 2005; Stockdon et al., 2002; Woolard and Colby, 2002; Zhang et al., 2005). Aerial lidar-based results can provide advanced information in order to develop best management practices (White and Wang, 2003; Young and Ashford, 2006). Recent studies have implemented the use of aerial lidar in combination with aerial photography and ground-truth data from traditional techniques to analyze coastal changes (Camacho-Valdez et al., 2008; Claudino-Sales, Wang, and Horwitz, 2010; Houser, Hobbs, and Saari, 2008; Priestas and Fagherazzi, 2010), The disadvantages of aerial lidar collection are the costs and availability to fly the equipment and the ability to achieve a resolution fine enough to capture the processes occurring on the ground. Alternatively, field lidar, or terrestrial laser scanning (TLS) technology can be mounted on a tripod and used to collect data at local spatial scales with finer point spacing (Lichti, Pfeifer, and Maas, 2008; Moorthy et al., 2008; Straatsma, Warmink, and Middelkoop, 2008; Tao, McCormick, and Wu, 2008). For example, one study used mobile TLS mounted on a truck to study rows of trees (Rosell Polo et al., 2009 ). Some TLS systems are capable of data collection at nominal distance to targets of 150 m through 3608 in the horizontal plane and 3008 in the vertical. Terrestrial laser scanning surveys can be registered together from multiple collection sites and georeferenced to create a 3-D virtual environment. While aerial lidar is prominently represented in literature, there is a lack of methodology for TLS in coastal vegetated ecosystems. In order to test and determine the value of different methods using terrestrial lidar for coastal geomorphological work, we examined volumetric changes in sediment and vegetation in relation to Hurricane Ike using a terrestrial laser scanner. We predicted that the terrestrial lidar analysis could be used to determine the change in sediment and vegetation volumes on both embryonic and established coastal dunes from before to after Hurricane Ike’s landfall on the East Matagorda Peninsula, Texas. We conducted this study in a 100 m 3 100 m plot, over a 1 year period using a before and after experimental design, where TLS data was collected on five sampling dates: two samples in September 2008 (before and after Hurricane Ike’s landfall), a winter sample in December 2008, a spring sample in May 2009, and an October 2009 sample (a year after landfall). Specifically, our objectives were to (1) quantify sand dune sedimentary and vegetation change before and after a hurricane, through the use of a methodology/technology that can be deployed on shortnotice; (2) identify the TLS interpolation grid size that best avoids point-cloud errors induced by data gaps and shadows yet also yields well-resolved, detailed topographic features; and (3) identify future needs and directions for TLS analysis in coastal applications

METHODS Study Area The East Matagorda Peninsula is approximately 160 km south of Houston, Texas, and 160 km west of Galveston Island, Texas, in the NW Gulf of Mexico (Figure 1). Hurricane Ike, the fifth and largest hurricane in the Atlantic Ocean during 2008, reached Category 4 classification on September 4 before hitting Haiti and Turks and Caicos Islands (Davenport, 2008). Ike then proceeded to make direct landfall on Galveston Island on September 13 as a Category 2 (Berg, 2010; Kraus and Lin, 2009). Episodic events can cause catastrophic damage to both natural and anthropogenic structures along the coast (Bush, Neal, and Young, 2004; Gaddis et al., 2007; Phillips and Jones, 2006). Ike caused over 100 fatalities, about half occurring in Texas (Berg, 2010) and up to $12 billion of damages to onshore property (Schwartz, 2008) and offshore oil rigs (Rach, 2008). Environmentally, Ike caused drastic sediment erosion and overwash destruction throughout the Texas coast (Williams et al., 2009). Mean winter and summer temperatures are 128 C and 288 C, respectively, and the mean total annual precipitation is 1219 mm with 686 mm from April to September. The main hurricane season is from July to September with erosion and accretion events associated with their occurrence. Prevailing winds are from the S-SE with a peak of 15 km/h in March. The Peninsula’s soils, derived from marine sand deposits (Hyde, 1991), are characterized as very poorly to excessively drained, nonsaline to saline, and sandy to loamy textured. Paine and Morton (1989) have hypothesized that the modern development of the Peninsula is erosion-dominated, transgressing or retreating shoreward continuously, as evidenced by past hurricanes, with a transverse topographic grain (McGowen and Brewton, 1975). This has led to the Peninsula being characterized as a complex overwash feature (McGowen et al., 1976) displaying historical shoreline retreat (Paine and Morton, 1989). Peninsulas and the coastal dunes within them are dynamic environments that constantly experience accretion and erosion with a varying sediment supply to the beach from the inner shelf. The study area is located in the 640 ha Matagorda Bay Nature Park that is managed by the Lower Colorado River Authority (LCRA). The study site is approximately 1.5 km NE of the Beach Access Road that provides public access to the coast for recreational use (Figure 1). Phosphorus and protein deficient vegetation, consisting of marsh hay cordgrass, seashore salt-grass, sea-oats, and various panicums, inhabit the dunes and have mean annual production estimated at 3400 kg/ha (Hyde, 1991). The Park is in one of four principal migratory bird routes in North America. The dunes provide habitat for over 300 species of various shore and marsh birds, including a number of endangered species, making it one of the top winter birding sites in the United States as ranked by the Audubon Society. The study plot encompasses two soil series: (1) the beach (Bb), composed of marine sand and shell fragments located at the embryonic dunes; and (2) the Galveston fine sand series (GaB) comprises the weakly undulating and excessively drained established coastal dunes. Established dunes are

Journal of Coastal Research, Vol. 30, No. 1, 2014

Terrestrial Laser Scanning

113

Figure 1. Study site is on (a) East Matagorda Peninsula, Texas, near the city of Matagorda. The specific study area on the beach along the Gulf of Mexico is visible (b) in aerial imagery showing the surrounding geomorphic and human development features and (c) in aerial lidar imagery detailing nearby topographic elevations.

marked by dune structures that are present throughout the year and support the majority of vegetation on the coast. At the study site, the established dunes run parallel to the Gulf of Mexico, protecting landward ecosystems and structures, specifically the Intracoastal Waterway. Embryonic dunes are newly created entities that develop at different times of the year based on vegetation growth, sediment movement, and disturbance regimes. If embryonic dunes persist, they will naturally lead to established dunes with time and adequate sediment accumulation. Embryonic and established dunes differ based on location on the beach profile, vegetation growth and composition, ground water levels, elevation and slope, faunal use and habitation, and human foot and automobile traffic. At this location, the winds from Hurricane Ike reached a maximum of 176 km/h or km h-1, and the storm surge reached approximately 3.3 m above mean lower low-water level at the study area (Berg, 2010). The hurricane surge inundated the beach and left a visible scarp line that cut into the primary dune ridge.

Data Collection We first surveyed an area, stretching from the dunes to the backbeach on September 10, 2008 (before Hurricane landfall or PRE) using a Leica ScanStation 2 TLS (Leica Geosystems AG, Heerbrugg, St. Gallen, Switzerland, Scan Station 2). A handheld Global Positioning System (GPS) unit was used to initially record the general location of the TLS, while reference targets and survey stakes were used for positioning future scans. We then surveyed the same location on September 26, 2008 (POST1); December 11, 2008 (POST2); May 19, 2009 (POST3); and October 24, 2009 (POST4) (Table 1). Because of a lack of permanent structures on the beach, repeat measurements were conducted based on survey stakes (rods 1.3 m long), which were placed on the beach and in the dunes on the first sampling date. For the PRE sampling only, we were unable to locate these stakes because of the rearrangement of sand by the hurricane. Additional survey stakes were then placed in the dunes during POST2. During POST3 and POST4, the same survey stakes were used to reference the scans. Subsequently, the plots for each date were registered to one another using


114

Table 1.

Feagin et al.

Sampling dates and visual observations.

Name

Date

Focus

Visual Observation

PRE POST1 POST2 POST3 POST4

Sept 10, 2008 Sept 26, 2008 Dec 11, 2008 May 19, 2009 Oct 24, 2009

Baseline measurements Changes from Hurricane Ike Accretion or erosion since POST1 Prehurricane 2009 season Posthurricane 2009 season

Baseline Erosion, approx. 3 m scarp Little change, smoother dune contour Accretion, but not full return to baseline, reestablishment of plants Erosion from POST3, but less than after Hurricane Ike

linear transformation and then cropped to equal areas of interest (see further discussion). In the first data set (PRE), lidar scans of the study area were taken from three locations: at the beach front and on top of the dune at the east and west extents of the study area. After preliminary analysis of the data, the next survey was conducted with scans from three sites located along the beachfront: east, west, and center of the study area. This decision was made based on lidar scanner battery power and what was determined most necessary for data capture. Further preliminary data analysis determined that only the east and west beachfront scans were needed.

Data Registration Cyclone, a 3D Point Cloud Software program (Leica Geosystems, Inc., 2001–2009, version 6.0.3), was used to extract and visualize the lidar data. Same-date scans were automatically combined for coregistration by Cyclone. Three scans were merged for the PRE and POST1 surveys, and two scans were merged for POST2, POST3, and POST4. The combined point clouds were then imported into Quick Terrain Modeler (QTM, Applied Imagery, USA, 2009, version 7.0.0) as ungridded point clouds (Figure 2). Errors in registration among the five point clouds were investigated by producing linear transformation equations for the X, Y, and Z dimensions based on offsets extracted by matching concordant locations from features among the data sets. Features that were used to test this transformation included a fence, a partially buried large drift log, and a sign. The fence was present in every sample set but had been damaged by the hurricane; therefore, we used the base of the fence posts rather than their tops. The log was present in all data sets and had not appeared to move relative to the fencepost bottoms. Much of it was under sand before the hurricane and was uncovered during the storm, though apparently it had remained in the same place. The sign was available only in data sample sets after the hurricane. Transformations for PRE, POST1, POST2, and POST3 were conducted relative to POST4, and the goodness-of-fit was calculated for each using standard linear regression (Table 2). The standard error was largest for the PRE data set, likely attributable to a lack of common points because of hurricane destruction or potentially to movement of the log or other features. Next, a 100 3 100 m area of interest (AOI) was selected to set the common study area boundary because each of the collection dates contained slightly different laser point returns. This boundary was chosen based on (1) the seaward extent of the cross-shore dimension where the visible presence of car tire

tracks could alter our analysis, as could be caused by compaction of sediment or destruction of vegetation by human alteration (Figure 3); (2) along-shore boundaries of the landscape where point density dropped below an average of 1 cm in the horizontal dimension; and (3) landward extent of the cross-shore dimension where the dune ridge crest shadowed the landscape behind it. Anthropogenic structures (the fence posts and sign) were removed manually in QTM on the applicable data sets.

Overall Sample Data, Sediment, and Vegetation Change Analysis Our analysis looked at three variables for the landscape: overall sample volume (the combination of sediment and vegetation volume), sediment volume, and vegetation volume. The analyses were performed on a gridded surface interpolated from the point cloud using the Convert Data Type tool in QTM at resolutions of 0.05 m, 0.10 m, 0.50 m, 1.00 m, and 5.00 m. For each interpolated grid the overall sample data volume changes were then computed using the volume calculation tool in QTM by subtracting a given interpolation from the interpolation of the previous sampling date (i.e., POST2 overall sample data change ¼ [interpolated POST2 data] [interpolated POST1 data]). Next, we used QTM to extract the sediment from the overall sample data. The above ground level (AGL) tool in QTM finds the ground level from the point cloud by filtering the data set, leaving only the lowest Z return points within a given resolution bin. Sedimentary volume changes were then computed using the volume calculation tool by subtracting the interpolated ground data from the previous sampling date [i.e., POST2 sediment change¼ (POST2 ground data) – (POST1 ground data)]. Vegetation volumes were then calculated by the AGL procedure where all points above ground level are interpolated and the floor is considered to be the ground level. To produce volume change, the vegetation amounts were then determined by subtracting the vegetation volume from the previous sampling date, mimicking the process done for the other two variables. For further investigation, we sectioned the datasets into two separate spatial locations: embryonic and established dunes. To do this, we manually separated the AOI for all scans based on a subjective interpretation of where the PRE dune-ridge slope began. This location was held spatially consistent across all datasets. We subsequently analyzed each section, as described previously, for overall sample data, sediment, and vegetation changes.



Table 2.

115

Quantification of error in data registration.

Regression Statistics (Goodness-of-Fit with Post4 XYZ Locations) Sample PRE

POST1

POST2

POST3

Variable

R square

STD error

X Y Z X Y Z X Y Z X Y Z

0.999 0.993 0.969 1.000 1.000 0.984 1.000 1.000 0.968 1.000 1.000 0.997

0.694 0.964 0.229 0.176 0.214 0.109 0.114 0.107 0.120 0.025 0.038 0.064

RESULTS

Figure 2. Overall sample data (raw point) data for the five data samples: (A1) PRE – September 10; (A2) POST1 – September 26; (A3) POST2 – December 11; (A4) POST3 – May 19; (A5) POST4 – October 24. Example of raw point data for only the (B1) established and (B2) embryonic dunes.

After looking at the initial results, two issues became apparent. First, interpolated maps at the two finest grid sizes (0.01 m 3 0.01 m and 0.05 m 3 0.05 m) were ‘‘bumpy’’ because of shadows in the original laser point cloud (shadows were caused by a lack of returns from behind dense vegetation, dune ridges, or mounds). Conversely, the interpolation at the coarsest grid size (5.00 m 3 5.00 m) lacked the ability to resolve transitions in the dune structure. Based on errors introduced by shadow gaps for fine grid sizes and lack of ability to resolve transition in coarse grid sizes, both 1.00 3 1.00 m and 0.50 3 0.50 m grids were determined best for subsequent analyses. At the 0.50 m 3 0.50 m grid resolution, Hurricane Ike caused 201.64 m3 of overall sample data loss and 92.25 m3 of sediment loss immediately after the hurricane within the 100 3 100 m area. After initial loss to the overall sample data and sediment variables, there was recovery of nearly half of the overall sample data and sediment volume between POST1 and POST2, followed by more erosion in POST3 and then slight accretion throughout the summer months (Figures 4a and b). Over a 1year time period (PRE to POST4), 139.05 m3 of overall sample data and 144.74 m3 of sediment loss occurred. At the 1.00 m 3 1.00 resolution, the direction of change coincided with the 0.50 m results; however, the magnitude of change was different. Vegetation change analysis at the 0.50 m 3 0.50 m resolution showed only slight vegetation loss immediately after the hurricane with only 47.53 m3 lost, but by POST2 vegetation recovery had begun and continued to increase through POST3 and POST 4 (Figure 4c). Over a 1-year time period (PRE to POST4), 25.47 m3 of vegetation was lost. It should be noted that while overall sample area loss would be expected to be the sum of sediment loss plus vegetation loss, the values do not add up. This is likely attributable to interpolation errors because of shadowing when distinguishing vegetation during the AGL procedure, as well as interpolation errors attributable to the averaging of all points to find elevation in the total data set (see ‘‘Discussion’’ for more details). For both the embryonic and dune-ridge portions of the landscape, the overall and sedimentary changes exhibited loss immediately after the hurricane. Recovery occurred during POST2, and then losses began by POST3 through POST4 (Figures 5a and b). In general, the embryonic dunes initially


116

Figure 3.

Feagin et al.

On-site location of scanner, survey reference stakes, and features used during the data registration process.

lost less overall volume and sediment volume than the established dunes. By POST3, both areas had similar overall sample data volumes, but the embryonic dunes had higher sediment volumes than the established dunes. Vegetation was also lost immediately after the hurricane and continued to decrease through POST2 (Figure 5c). However, the established dunes exhibited vegetation recovery to PRE levels by POST3 and continued to increase through POST4, while the embryonic dunes continued to decrease through POST2, only showing slight increases through POST3 and POST 4 but never reaching PRE levels.

DISCUSSION Summary of Findings at the Example Study Area Using the 0.50 m 3 0.50 m analysis as an example, sediment loss occurred immediately after Hurricane Ike. The sediment volume then recovered to over 75% of PRE levels by POST2 (December) but fell again by POST4 (October). Therefore, the sediment volumes experienced a greater net loss over the course of the year than during the hurricane erosion event alone. The greatest vegetation volume loss was experienced after Hurricane Ike, with a gradual recovery beginning by POST2 (December). Increased vegetation volume during this winter season could be the growth of perennial plants or the emergence of annual plants (Udo and Takewaka, 2007). The data in this study supports a cyclic process of erosion and accretion throughout the year in which sediment and vegetation have different cycles instead of a continuous recovery after Hurricane Ike, as seen in other studies (Priestas and

Fagherazzi, 2010). However, multiple years of data would be needed to determine if this is a typical yearly cycle and, if so, how hurricanes affect this cycle.

Lessons Learned: Sedimentary Volume and Change TLS, aerial lidar, or point cloud data users should be careful when using point data to produce an interpolated digital elevation model (DEM), particularly when all laser returns are averaged within a given grid cell. The result can often overestimate ground height in locations of high vegetation density and alter estimates of landscape change over time. We thus consider our sediment change results as the most accurate because they are based on the points of data at the lowest elevations, which were then interpolated. There often is a smooth slope between ground data points, making the interpolation more reliable than the overall sample data or vegetation data, where the interpolated vertical profile is often more jagged and more randomly dependent upon the laser reflecting off of individual plants. We suggest using groundtruthing of sediment data to verify accuracy of TLS results, such as what was done in aerial lidar studies (Claudino-Sales, Wang, and Horwitz, 2010; Priestas and Fagherazzi, 2010). Using our dataset as an example, the choice of the resolution of analysis will affect the mismatch between the overall landscape volume and the sediment volume. Interestingly in our dataset, more sediment volume was lost than overall sample data volume throughout the study duration. This may have been attributable to a net gain of vegetation throughout the year; however, it was more likely caused by holes in the data being filled in during the interpolation. Figures 4 and 5



117

Figure 4. Differences in volumetric changes from previous dates at two resolutions for (a) overall sample data, (b) sediment, and (c) vegetation data sets.

Figure 5. Differences in volumetric changes from previous dates for embryonic and established dunes at 1 m resolution for (a) overall sample data sets, (b) sediment data sets, and (c) vegetation data sets.

reflect this in the POST1 results; the 0.50 m 3 0.50 m interpolated data provides results where the sum of sediment and vegetation loss is more similar to overall sample data change than the data interpolated to 1.00 m 3 1.00 m. This is important to keep in mind when designing the experiment in order to capture data that will be adequate to use at a finer resolution.

POST1 date in Figure 4. In comparison, the 0.50 m 3 0.50 m results appear to be more closely additive (i.e., overall sample data change ¼ sediment change þ vegetative change). As mentioned earlier in ‘‘Results,’’ one wants to balance between detecting transitions in the structure that would be missed by a coarser resolution analysis while also minimizing the errors induced by shadows at finer scales. The 0.50 m 3 0.50 m resolution appears to best satisfy this balance based upon our data set. Moreover, the vegetation dataset has interpolation errors introduced by both the ground-level estimation procedure and the interpolation of the vegetation points themselves. While a variety of data manipulations were attempted to compensate for this issue, no appropriate solutions were discovered using

Lessons Learned: Vegetation Volume and Change While TLS data can provide multiple point returns, vegetation can still be hard to separate from the ground. This complexity can lead programs or operations, such as QTM’s AGL procedure, to introduce errors into the vegetation layers. Again, this can be seen at the 1.00 m 3 1.00 m resolution for the


118

Feagin et al.

QTM. This is a limitation of using lidar data and the computer hardware and QTM software that was discovered in this study and needs to be taken into consideration for future studies. The authors suggest further ground-truthing, use of survey-grade GPS, permanent benchmarks, and advanced data manipulation to determine how to adequately manage these errors. Also, decisions need to be made ahead of time to determine a better scanning scheme in which point spacing is set to reduce gaps, allowing for better interpolation at finer scales.

Lessons Learned: Ground-Truthing of Sedimentary and Vegetation Parameters Although we could calculate volumetric changes remotely using TLS, we could not determine changes in the compaction of sediment. After Hurricane Ike, the remaining sediment appeared more compact, as the loose sediment was removed by erosion. Ground-truthing of sediment volumes could be an extension of this project to calculate mass change more accurately. Alternately, one could determine the soil mass by multiplying the range moist bulk density estimates from the Matagorda Bay County Soil survey for the Beach (1.35 to 1.50 g cm3) by the volumes involved. One could also calculate the biomass of vegetation. Identification of the plants in the area, the percent coverage of each plant species, and their field-estimated height could be used to determine the biomass through the use of allometric equations. Species of vegetation each have a unique physiognomy that could be interpreted differently through lidar analysis. For example, sea-oats (Uniola paniculata), observed on top of the established dune, are tall plants that protrude into the air and would easily be picked up by lidar and distinguished from the ground, while morning glories (Ipomoea spp.) and seashore dropseed (Sporobolus virginicus), commonly observed on the embryonic dunes, grow low to the ground in clumped or creeping formations (Britton and Morton, 1989; Tiner, 1993). Vegetation is well known to alter sedimentary accretion and erosion (Camacho-Valdez et al., 2008), so more studies are needed into this area, and TLS may prove particularly useful in this regard. Moreover, studies such as the one described in Stockdon, Doarn, and Sallenger (2009) are needed to segregate specific biophysical regions of the beach-dune continuum into discrete units for analysis, for example, the embryonic dunes vs. the dune ridge. Capitalizing on new methods and techniques for computational analysis of lidar is imperative in producing the quick and reliable results for coastal evaluations (Ali and Mehrabian, 2009; Hart and Knight, 2009; Kempeneers et al., 2009; Leigh, Kidner, and Thomas, 2009; Palaseanu-Lovejoy et al., 2009; Yates et al., 2008). For episodic events, such as Hurricane Ike, quick and reliable analysis of the situation is critical in making decisions that can result in life-saving procedures. Terrestrial laser scanning can provide a quick and accurate method of predicting coastal changes, analyzing impacts, and developing recovery plans (Klemas, 2009; Ramsey, III, Werle, and Lu, 2009). Terrestrial laser scanning can be used prior to natural events to determine vulnerable areas, especially in dune structures on coasts, in order to take preemptive measures for protection (Hart and Knight, 2009).

CONCLUSION Our research provides insight into methods for collecting and analyzing geomorphological changes on a coastal ecosystem using TLS. While the use of TLS has the potential to quantify geomorphic changes faster, more reliably, and at finer scales than before, it is imperative that considerations of point cloud registration, resolution, errors introduced by interpolation, and the interaction of these factors be studied. In our example, 0.50 m 3 0.50 m grids were determined best for analysis as they were able to handle data gaps yet also accurately map sand dune topography. Further investigations are necessary to link mass-based sediment changes and vegetation biomass changes to the volumetric analyses made possible with TLS.

ACKNOWLEDGMENTS We would like to thank Mark Karnauch and Vince Mendieta for field assistance in collecting the terrestrial data, and we want to express our appreciation to Dr. Kaiguang Zhao for his expert assistance in data analysis. We would also like to thank the anonymous reviewers for their comments and critiques that helped to make this a stronger and more focused manuscript.

LITERATURE CITED Ali, T. and Mehrabian, A., 2009. A novel computational paradigm for creating a Triangular Irregular Network (TIN) from LiDAR data. Nonlinear Analysis 71, e624–e629. Bater, C.W. and Coops, N.C., 2009. Evaluating error associated with lidar-derived DEM interpolation. Computers & Geosciences, 35, 289–300. Berg, R., 2010. Tropical Cyclone Report (AL092008) 1–14 September 2008. National Hurricane Center of NOAA/National Weather Service, http://www.nhc.noaa.gov/pdf/TCR-AL092008_Ike_ 3May10.pdf (accesed May 30, 2012). Blott, S.J. and Pye, K., 2004. Morphological and sedimentological changes on an artificially nourished beach, Lincolnshire, UK. Journal of Coastal Research, 20, 214–233. Britton, J.C. and Morton, B., 1989. Gulf Beaches and Barrier Islands. In: Shore Ecology of the Gulf of Mexico. Austin, Texas: University of Texas Press, 387p. Brock, J.C. and Purkis, S.J., 2009. The emerging role of lidar remote sensing in coastal research and resource management. In: Brock, J.C. and Purkis, S.J. (eds.), Coastal Applications of Airborn Lidar. Journal of Coastal Research, Special Issue No. 53, pp. 1–5. Brock, J.C.; Wright, C.W.; Sallenger, A.H.; Krabill, W.B., and Swift, R.N., 2002. Basis and methods of NASA airborne topographic mapper lidar surveys for coastal studies. Journal of Coastal Research, 18, 1–13. Bush, D.M.; Neal, W.J., and Young, R.S., 2004. After the storms: geologists’ perspectives for architects on building in coastal zones. Architectural Record, 192, 65–66. Camacho-Valdez, V.; Murillo-Jimenez, J.M.; Nava-Sanchez, E.H., and Turrent-Thompson, C., 2008. Dune and beach morphodynamics at Cabo Falso, Baja California Sur, Mexico: response to natural, Hurricane Juliette (2001) and anthropogenic influence. Journal of Coastal Research, 24, 553–560. Claudino-Sales, V.; Wang, P., and Horwitz, M.H., 2010. Effect of Hurricane Ivan on coastal dunes of Santa Rosa Barrier Island, Florida: characterized on the basis of pre- and poststorm LIDAR surveys. Journal of Coastal Research, 26, 470–484. Davenport, S., 2008. Weather News: Hurricane Ike. Weather, 63, 290. Delgado-Fernandez, I.; Davidson-Arnott, R., and Ollerhead, J., 2009. Application of remote sensing technique to the study of coastal dunes. Journal of Coastal Research, 25, 1160–1167. Gaddis, E.B.; Miles, B.; Morse, S., and Lewis, D., 2007. Full-cost accounting of coastal disasters in the United States: implications for planning and preparedness. Ecological Economics, 63, 307–318.



Gares, P.A.; Wang, Y., and White, S.A., 2006. Using LIDAR to monitor a beach nourishment project at Wrightsville Beach, North Carolina, USA. Journal of Coastal Research, 22, 1206–1219. Hart, D.E. and Knight, G.A., 2009. Geographic Information System assessment of tsunami vulnerability on a dune coast. Journal of Coastal Research, 25, 131–141. Houser, C.; Hobbs, C., and Saari, B. 2008. Posthurricane airflow and sediment transport over a recovering dune. Journal of Coastal Research, 24, 944–953. Hyde, H.W. 1991. Soil Survey of Matagorda County, Texas. College Station, Texas: USDA Natural Resources Conservation Service in cooperation with Texas Agricultural Experiment Station. Kempeneers, P.; Deronde, B.; Provoost, S., and Houthuys, R., 2009. Synergy of airborne digital camera and lidar data to map coastal dune vegetation. In: Brock, J.C. and Purkis, S.J., Coastal Applications of Airborne Lidar. Journal of Coastal Research, Special Issue No. 53, pp. 73–82. Klemas, V.V., 2009. The role of remote sensing in predicting and determining coastal storm impacts. Journal of Coastal Research, 25, 1264–1275. Kraus, N.C. and Lin, L., 2009. Hurricane Ike along the Upper Texas coast: an introduction. Shore & Beach, 77, 3–8. Leigh, C.L.; Kidner, D.B., and Thomas, M.C., 2009. The use of LiDAR in digital surface modelling: issues and errors. Transactions in GIS, 13, 345–361. Li, J. and Liu, H., 2009. Improved hurricane track and intensity forescast using single field-of-view advanced IR sounding measurements. Geophysical Research Letters, 36, 1–4. Lichti, D.; Pfeifer, N., and Maas, H-G., 2008. ISPRS Journal of photogrammetry and remote sensing theme issue ‘‘Terrestrial Laser Scanning.’’ ISPRS Journal of Photogrammetry and Remote Sensing, 63, 1–3. McCulloh, R.P. and Heinrich, P.V., 2009. Lidar imagery and geologic mapping in Louisiana and similar humid-subtropical and coastalplains regions. NewInsights, 19, 4–6. McGowen, J.H. and Brewton, J.L., 1975. Historical Changes and Related Coastal Processes, Gulf and Mainland Shorelines, Matagorda Bay Area, Texas. Austin, Texas: Bureau of Economic Geology. McGowen, J.H.; Brown, L.F.; Evans, T.J.; Fisher, W.L., and Groat, C.G., 1976. Environmental Geologic Atlas of the Texas Coastal Zone, Bay City-Freeport Area. Austin, Texas: Bureau of Economic Geology. Moorthy, I.; Miller, J.R.; Hu, B.; Chen, J., and Li, Q., 2008. Retrieving crown leaf area index from an individual tree using ground-based lidar data. Canadian Journal of Remote Sensing, 34, 320–332. Morton, R.A. and Sallenger, A.H., Jr., 2003. Morphological impacts of extreme storms on beaches and barriers. Journal of Coastal Research, 19, 569–573. Nayegandhi, A.; Brock, J.C.; Wright, C.W., and O’Connell, M.O., 2006. Evaluating a small-footprint, waveform-resolving Lidar over coastal vegetation communities. Photogrammetric Engineering and Remote Sensing, 12, 1408–1417. Paine, J.G. and Morton, R.A., 1989. Shoreline and Vegetation Line Movement, Texas Gulf Coast, 1974–1982. Bureau of Economic Geology Circular 89-1. Austin, Texas: University of Austin. Palaseanu-Lovejoy, M.; Nayegandhi, A.; Brock, J.; Woodman, R., and Wright, C.W., 2009. Evaluation of airborne lidar data to predict vegetation presence/absence. In: Brock, J.C. and Purkis, S.J., Coastal Applications of Airborne Lidar. Journal of Coastal Research, Special Issue No. 53, pp. 83–97. Phillips, M.R. and Jones, A.L., 2006. Erosion and tourisms infrastructure in the coastal zone: problems, consequences and management. Tourism Management, 27, 517–524. Priestas, A.M. and Fagherazzi, S., 2010. Morphological barrier island changes and recovery of dunes after Hurricane Dennis, St. George Island, Florida. Geomorphology, 114, 614–626. Rach, N., 2008. September hurricanes hit gulf rigs. Oil & Gas Journal, 106, 64–65. Ramsey, III, E.; Werle, D., and Lu, Z., 2009. A case of timely satellite image acquisitions in support of coastal emergency environmental

119

response management. Journal of Coastal Research, 25, 1168– 1172. Robertson, V, W.; Zhang, K.; Whitman, D., and Leatherman, S., 2005. Shoreline and beach volume change before and after the 2004 hurricane season, Palm Beach County, Florida. Shore & Beach, 73, 79–84. Rosell Polo, J.R.; Sanz, R.; Llorens, J.; Arno, J.; Escola, A.; Ribes-Dasi, M.; Masip, J.; Camp, F.; Gracia, F.; Solanelles, F.; Palleja, T.; Val, L.; Planas, S.; Gil, E., and Palacin, J., 2009. A tractor-mounted scanning LIDAR for the non-destructive measurement of vegetative volume and surface area of tree-row plantations: a comparison with conventional destructive measurements. Biosystems Engineering, 102, 128–134. Sallenger, A.; Wright, C.W., and Lillycrop, J., 2007. Coastal-change impacts during Hurricane Katrina: An overview. Coastal Sediments ’07 (ASCE, Reston, Virginia), pp. 1–9. Sallenger, A.H.; Wright, C.W.; Guy, K., and Morgan, K. 2004. Assessing storm-induced damage and dune erosion using airborne lidar: examples from Hurricane Isabel. Shore & Beach, 72, 3–7. Schwartz, E., 2008. Full Rebound from Hurricane Could Take Years, Say Officials. In: Engineering News-Record, p. 15. Shrestha, R.L.; Carter, W.E.; Sartori, M.; Luzum, B.J., and Slatton, K.C. 2005. Airborne laser swath mapping: quantifying changes in sandy beaches over time scales of weeks to years. ISPRS Journal of Photogrammetry & Remote Sensing, 59, 222–232. Stockdon, H.F.; Asbury, H.; Sallenger, J.; List, J.H., and Holman, R.A., 2002. Estimation of shoreline position and change using airborne topographic lidar data. Journal of Coastal Research, 18, 502–513. Stockdon, H.F.; Doarn, K.S., and Sallenger, Jr., A.H., 2009. Extraction of lidar-based dune-crest elevations for use in examining the vulnerability of beaches to inundation during hurricanes. In: Brock, J.C. and Purkis, S.J., Coastal Applications of Airborne Lidar. Journal of Coastal Research, Special Issue No. 53, pp. 59– 65. Straatsma, M.W.; Warmink, J.J., and Middelkoop, H., 2008. Two novel methods for field measurements of hydrodynamic density of floodplain vegetation using terrestrial lidar scanning and digital parallel photography. International Journal of Remote Sensing, 29, 1595–1617. Tao, Z.; McCormick, M.P., and Wu, D., 2008. A comparison method for spaceborne and ground-based lidar and its application to the CALIPSO lidar. Applied Physics B: Lasers & Optics, 91, 639–644. Tiner, R.W., 1993. Field Guide to Coastal Wetland Plants of the Southeastern United States. Amherst: The University of Massachusetts Press. Udo, K. and Takewaka, S., 2007. Experimental study of blown sand in a vegetated area. Journal of Coastal Research, 23, 1175–1182. White, S.A. and Wang, Y., 2003. Utilizing DEMs derived from LIDAR data to analyze morphologic change in the North Carolina coastline. Remote Sensing of Environment, 85, 39–47. Williams, A.M.; Feagin, R.A.; Smith, W.K., and Jackson, N.L., 2009. Ecosystem impacts of Hurricane Ike on Galveston Island and Bolivar Peninsula: perspectives of the Coastal Barrier Island Network (CBIN). Shore & Beach, 77, 71–76. Woolard, J.W. and Colby, J.D., 2002. Spatial characterization, resolution and volumetric change of coastal dunes using airborne LIDAR: Cape Hatteras, North Carolina. Geomorphology, 48, 269– 287. Yates, M.L.; Guza, R.T.; Gutierrez, R., and Seymour, R.A., 2008. Technique for eliminating water returns from lidar beach elevation surveys. Journal of Atmospheric and Oceanic Technology, 25, 1671–1682. Young, A.P. and Ashford, S.A., 2006. Application of airborne LIDAR for seacliff volumetric change and beach-sediment budget contributions. Journal of Coastal Research, 22, 307–318. Zhang, K.; Whitman, D.; Leatherman, S., and Robertson, W., 2005. Quantification of beach changes caused by Hurricane Floyd along Florida’s Atlantic Coast using airborne laser surveys. Journal of Coastal Research, 21, 123–134.
