Precise Airborne LiDAR Surveying For Coastal Research And Ge(4)

models, we have classified ALSM data using algorithmsdeveloped by TopScan GmbH (Petzold et al, 1999) and by theUT Center for Space Research (Neunschwander et al, 2000).The TopScan algorithm identifies points as either “ground” or“non-ground” by iteratively improving an initial terrain surface.The initial terrain surface is generated from the minimum valueof elevation points within a large, moving window. All theelevation points that exceed a specified threshold above theterrain are classified as non-ground points and removed. Using asmaller moving window, the remaining elevation points are usedto create a new terrain surface. The ALSM data are againcompared to a threshold value and the non-ground points areremoved. This process is repeated for a set number of iterations.The window size and threshold values are terrain-dependent andrequire a high level of user interaction.

The UT method classifies elevation points as ground, vegetation,or buildings using an image-based processing algorithm. TheALSM data are gridded to create a high-resolution topographicimage. The average topographic surface is estimated andsubtracted from the high-resolution image. The resultingresidual image contains the high-frequency content of thevegetation and the building edges. The lower envelope of high-frequency residuals represents the ground surface in the ing the lower envelope, an initial ground surface is estimated.A gradient-based method is used to detect and remove any largebuildings remaining in the estimated ground surface. Afterinterpolating across gaps, the final ground surface is used toclassify the ALSM data. Building classification is accomplishedby first detecting planar surfaces representing roofs. Thebuilding boundaries are delineated by extending the edges usinga gradient-flood fill method. The building surface is then used toclassify ALSM points as man-made features. A building outlinecan be distorted by laser multi-pathing, therefore ALSM first-returns are used for building classification.

Figure 8 is a 1m × 1m DEM constructed from first-return ALSMdata of the Mayan ruins at Copan, Honduras. An aerialphotograph is shown for comparison. Figure 9 is a 1m × 1mDEM of the Copan ruins constructed from last-return ALSMdata filtered to remove the trees using the envelope detector andgradient based method developed at UT. The elevation pointsrepresenting the Mayan archeological structures were classifiedand added to the ground points before the DEM was computed.For comparison is a site map constructed from a HarvardUniversity ground survey.

Figure 8. Left: ALSM DEM of Copan. Right: aerial photograph.

188

Figure 9. Left: vegetation-filtered ALSM. Right: ground survey.

3 COASTAL MAPPING

3.1 Texas Gulf Shoreline Change Project

In 1999, with the support of the Texas General Land Office, theBEG developed the Texas Shoreline Change Project. The project’sgoal is to establish a state-of-the-art regional shoreline-monitoringand shoreline-change analysis program that will help solve coastalerosion and storm hazard problems along the bay and Gulfshorelines of Texas. ALSM is a key component of the TexasShoreline Change Project; it is important in identifying "criticalcoastal erosion areas" and in the monitoring of historical shorelineerosion rates.

During 2000 we mapped the entire Texas Gulf shoreline using theOptech 1225 system from Sabine Pass, at the Texas-Louisianaborder, to the mouth of the Rio Grande River, a distance of over600 kilometers. We mapped the shoreline in three sections: SabinePass to Freeport (212km), Freeport to Corpus Christi (215km), andCorpus Christi to the Rio Grande (174km). During a typicalshoreline survey, the aircraft flew two to four passes along theshoreline with parallel swaths overlapping by about 50 percent.The survey altitude varied from 450m to 760m AGL and theground speed was usually held to 51m/sec (100 knots). Theresulting ALSM coverage of the beach, dunes, and back-barrierarea is 500m to 700m wide and has an average ground pointspacing of <1m.

Three ground GPS receivers, Ashtech Z-12 or Trimble 4000SSi,operated during the ALSM mapping. One GPS receiver wassituated at each end of the 200km section of coastline and the thirdwas located approximately in the middle of the survey area. Six ofthe nine GPS base stations occupied benchmarks at NOAA orTexas Coastal Ocean Observation Network (TCOON) tide gauges.These gauges are at Sabine Pass, Port Bolivar, Port O’Connor, PortAransas, Port Mansfield, and South Padre Island. The remainingthree GPS ground stations were monuments established by eitherthe NGS, the U.S. Army Corps of Engineers, or UT.

GPS data processing was conducted in the International TerrestrialReference Frame 1997 (ITRF97) and the ALSM elevation pointswere output in Universal Transverse Mercator (UTM) coordinatesand height above the GRS-80 ellipsoid (HAE). The ALSM datawere compared to GPS ground surveys for the estimation of