G51B-0431
An Evaluation of the Uncertainty Resulting from Human Editing for Airborne Lidar Point Clouds
In the process of producing DEM from point clouds obtained with airborne lidar, automated filtering is usually applied first to separate ground points from the rest. Then, the filtered result is manually edited. The quality of the automated filtering would affect the efficiency of manual editing, but not the quality. This study investigates the difference among individual operator in terms of the amount of add/remove points, as well as quality of the result. Because the automated filtering package used in this study requires operator to select proper parameters, there are uncertainties resulting from human choices as well. It is found that in flat terrain, the differences among individuals are not as significant as in the complex terrain with high reliefs. In general, the differences among operators are less than 5% for the deviation larger than one meter. It is also found that manual editing for type 1 error makes the terrain surface higher.
G51B-0432
A High-altitude, Advanced-technology Scanning Laser Altimeter for the Elevation for the Nation Program
In January of this year the National Research Council's Committee on Floodplain Mapping Technologies recommended to Congress that an Elevation for the Nation program be initiated to enable modernization of the nation's floodplain maps and to support the many other nationwide programs reliant on high-accuracy elevation data. Their recommendation is to acquire a national, high-resolution, seamless, consistent, public-domain, elevation data set created using airborne laser swath mapping (ALSM). Although existing commercial ALSM assets can acquire elevation data of sufficient accuracy, achieving nationwide consistency in a cost-effective manner will be a challenge employing multiple low-flying commercial systems conducting local to regional mapping. This will be particularly true in vegetated terrain where reproducible measurements of ground topography and vegetation structure are required for change-detection purposes. An alternative approach using an advanced technology, wide-swath, high-altitude laser altimeter is described here, based on the Swath Imaging Multi-polarization Photon-counting Lidar (SIMPL) under development via funding from NASA's Instrument Incubator Program. The approach envisions a commercial, federal agency and state partnership, with the USGS providing program coordination, NASA implementing the advanced technology instrumentation, the commercial sector conducting data collection and processing and states defining map product requirements meeting their specific needs. An Instrument Synthesis and Analysis (ISAL) study conducted at Goddard Space Flight Center evaluated an instrument compliment deployed on a long-range Gulfstream G550 platform operating at 12 km altitude. The English Electric Canberra is an alternative platform also under consideration. Instrumentation includes a scanning, multi-beam laser altimeter that maps a 10 km wide swath, IMU and Star Trackers for attitude determination, JPL's Global Differential GPS implementation for position determination not reliant on local ground stations, and airborne magnetometer and gravimeter measurements to refine knowledge of the medium-wavelength component of crustal magnetics and the geoid. The laser altimeter utilizes multiple short- pulse (1 nsec) laser transmitters operating at 532 nm, photon counting detectors, and high precision timing electronics to achieve 15 cm range precision per single detected photon. Scanning is accomplished using dual, counter-rotating, transmissive wedges, producing a self-calibrating, overlapping, bow-tie scan geometry that yields a point cloud with a nominal density of three discrete elevation measurements per square meter. Using a single platform to conduct long-range deployments to clear-sky areas and map 16,000 sq km areas per flight mission, complete seamless, consistent national coverage would be accomplished in 3 years. This approach would achieve the goals for the Elevation for the Nation program and provide a capability for on-going, comprehensive mapping to monitor elevation changes associated with natural hazards, human activity, and ecosystem, glacier, ice sheet and snow cover response to climate change.
G51B-0433
Survey-scale airborne lidar error analysis from parallel swath comparison
The determination of robust system calibration parameters and reliable survey error statistics is still an area of active research in airborne laser altimetry. We show how these twin challenges are in part related to long-period GPS noise due to multipathing and tropospheric effects. Systematic differences in absolute GPS positioning over periods of tens of minutes to hours mean that a major component of airborne survey error is both time and space dependent and cannot be determined by local measurements alone. Furthermore, it is standard practice to compare intersecting swaths at one or two locations to determine calibration parameters for a given flight. Correlation between vertical translations and other geolocation biases means that calibration parameters determined without explicitly accounting for GPS elevation error might inadvertently correct a portion of this error and thus be incorrectly estimated. Angular errors would then be apparent in footprint geolocation at other points in the survey. We diagnose error in a recent airborne laser altimeter survey by comparing 5 parallel and partially overlapping swaths from a continuous flight over a section of the Southern San Andreas fault (from the 2005 "B4" Survey). Correlated elevation errors between swaths range over +/- 10 cm, which may explain the presence of linear "corduroy" patterns in DEMs created from point cloud data. Although this survey segment was calibrated to remove biases in aircraft attitude and laser scan angle, the pattern of the residual misfit throughout much of the flight (e.g. across-track "ramps" in elevation) suggests that the estimated calibration parameters are not optimally determined. We also show the effect of long-period GPS noise on the mean bias between swaths and report on an attempt to independently estimate and remove this error.
G51B-0434
New Visualization Techniques to Analyze Ultra-High Resolution Three- and Four-Dimensional Airborne and Tripod LiDAR Point-Cloud Data
In the context of the UC~Davis W. M. Keck Center for Active Visualization in the Earth Sciences (KeckCAVES, http://www.keckcaves.org), we are developing an immersive visualization application to display and interact with very large (billions of points) three- and four-dimensional point-position datasets, such that point groups from repeated airborne and ground based Light Detection And Ranging (LiDAR) surveys can be selected, measured, and analyzed for quality control and land surface change detection. One of the difficulties of analyzing dense 3D and 4D point-cloud data is that there are few software packages that can display and analyze the data at full resolution and in the natural 3D perspective in which it was collected. We developed an octree-based, multiresolution, point-set data representation that allows very large point cloud datasets to be displayed at the frame rates required to create immersion (between 60 Hz and 120 Hz). Data inside an observer's region of interest is shown in full detail, whereas data outside the field of view or far away from the observer is shown at reduced resolution to provide context. Users can navigate LiDAR data sets and accurately select related point groups in two or more point sets by sweeping space using 3D input devices provided by immersive display environments such as CAVEs. Users can then guide the software in deriving positional information from point groups to compute displacements between surveys, or to extract survey measurements. This software runs on UNIX-like operating systems and can be used on laptop or desktop computers, 3D display systems such as Geowalls, and in fully immersive environments such as CAVEs. It is available for download from http://www.keckcaves.org. Examples of the wide range of applications of the software for airborne and Tripod LiDAR (T-LiDAR) include: 1)~visualization of airborne LiDAR data from the southern San Andreas Fault; 2)~quality control assessment of ground based T-LiDAR from the March 2006 Ka~Loko Dam breach on Kauai; and 3)~4D T-LiDAR time-series analysis from the June 2005 Blue Bird Canyon landslide in Laguna Beach, southern California. http://idav.ucdavis.edu/~okreylos/ResDev/LiDAR
G51B-0435
Efficient, Off-Grid LiDAR Scanning of Remote Field Sites
As terrestrial LiDAR scanning systems become increasingly available, strategies for executing efficient field surveys in settings without access to the power grid are increasingly needed. To evaluate scan methods and develop an off-grid power system, we used a tripod-mounted laser scanner to create high resolution (≤40 mm point spacing) topographic maps for use in neotectonic studies of active faulting in arid, high elevation settings. We required 1-2 cm internal precision within point clouds spanning field sites that were ~300 x 300 m. Main components of our survey system included a Trimble GX DR200+ terrestrial laser scanner, a Leica TCR407power total station, a ruggedized laptop (2 GB RAM, 2.33 GHz dual-processor, and an Intel GMA 950 graphics card), batteries, and a portable photovoltaic array. Our first goal was to develop an efficient field-survey workflow. We started each survey project by using the total station for 1-2 days to locate an average of 8 ground control locations per site and to measure key geomorphic features within the project area. We then used the laser scanner to capture overlapping scans of the site, which required an average of six, 5-hour scanning sessions and an average of ten station setups. At each station, the scanner located itself on a particular point by measuring the relative positions of an average of four backsights, each of which is a ~17 x 17cm reflective target mounted on a tripod over the ground control point. To locate the scanner at a particular station prior to scanning, we experimented with both setting up over known points as measured using the total station, and resectioning, by positioning the scanner over an unmeasured location and backsighting on previously scanned points. We found that resectioning provided the smallest errors in scan registration. We then framed and queued a series of scans from each station that optimized point density and minimized data repetition. We also increased the accuracy of the scanner location by adding backsight measurements between scans. During scanning, incoming data were displayed in real-time by the scanner software, allowing the user to check scan area, shadowing, and resolution by interactively visualizing the project point cloud individually or in the context of previously scanned point clouds. Because individual scans must be stitched to build the total point cloud, we are currently testing different scan registration techniques to better quantify which minimizes registration errors in the final point cloud. A second goal of our study was to develop a low-cost, off-grid method for powering the survey equipment. To this end, we used two sets of three, 65Ah, 12V sealed lead acid batteries, which we charged using two 55W and one 25W, 12V photovoltaic arrays. We found that key elements for maximizing efficiency included real-time data visualization for planning future scans, use of polygons to delimit scan area as tightly as possible, distance-limited scanning to minimize unnecessary measurements, target tear-down and set-up synchronous with scanning. Using these strategies, we completed two survey projects, each of which covered a field site of approximately 300 x 300 m with ~31 million data points at an average point spacing of ~37 mm. Our experience demonstrates the feasibility of executing terrestrial LiDAR scanning with an average of ~5.2 million points per scan day in remote, off-grid field areas.
G51B-0436
Satellite and Instrument Influences on ICESat Waveforms
The White Sands Space Harbor (WSSH) has served as the principal ground calibration site throughout the Ice, Cloud and land Elevation Satellite (ICESat) mission. The Center for Space Research (CSR) at the University of Texas at Austin continues to conduct various experiments designed to validate the timing, geolocation and geometric characteristics of individual laser footprints on the surface. In addition, two airborne lidar surveys of the calibration site and surrounding area were conducted during the mission, first in 2003 and again in 2007. Chosen for its limited surface roughness and topographic flatness, this area has been targeted 3-4 times in each of the 12 ICESat mapping campaigns to date, yielding a significant altimetry data set. The derived surface elevations are compared with those from the airborne lidar surveys, as well as those obtained by the Shuttle Radar Topography Mission (SRTM). Furthermore, the Geoscience Laser Altimetry System (GLAS) onboard ICESat records a digitized waveform for each laser pulse returned from the surface. The two methods currently used to fit such signals in ICESat data processing are examined and compared for the WSSH waveforms. The first fits up to two distinct Gaussians and provides a surface elevation at the location of the maximum peak. The second fits up to six overlapping Gaussians and provides a surface elevation at the centroid of the pulse. Observed differences in the reported elevations are discussed in terms of the satellite's off-nadir targeting geometry, the laser energy, and the skewness of the returned waveforms.
G51B-0437
Building Discrete Fracture Networks From Ground Based LiDAR Data
Simulations which employ Discrete Fracture Networks (DFN's) offer one of the most accurate methods to model the flow of fluids within fractured reservoirs/aquifers. Geometries and spacings of fractures in many DFN's are often stochastically generated using statistical distributions gathered from one- and two-dimensional data. Ground based LiDAR surveys of large fractured outcrops may improve the construction of DFN's by providing highly accurate three-dimensional (3D) measurements of fracture spacing and geometry. We have constructed a 3D DFN based upon the cross-cutting/abutting relationships of fracture sets observed in outcrop and a semi- automated method which uses differential geometry to extract fracture planes from a three-dimensional point cloud.
G51B-0438
LiDAR Acquisition for the GeoEarthScope Community
LiDAR acquisition is a key component of the GeoEarthScope Initiative. LiDAR provides data with a broad range of applicability to many of the EarthScope goals. A working group was convened to identify primary targets for data acquisition, rank these targets, and propose a data acquisition scheme to effectively acquire these data within the GeoEarthScope funding time frame. The Regional Targets are: a.) Northern California – including the San Andreas Fault north of Parkfield, and other major strands of the San Andreas Fault system; b.) Southern California – including the Garlock Fault, Eastern California Shear zone south of the Garlock, and the Elsinore Fault; c.) Eastern California, Walker Lane, and Basin and Range fault systems – including faults of the Eastern California Shear Zone north of the Garlock Fault; d.) Intermountain Seismic Belt – including the Wasatch Fault, Teton Fault, and Yellowstone Park area; e.) Alaska – including the Castle Mountain and Denali Faults; f.) Cascadia – including the Little Salmon fault zone in southern Cascadia, the Calawah Fault in the Washington forearc, and imagery in the Yakima Fold belt termination. The Working Group recognized that available funding would likely preclude obtaining data from all high priority sites The Working Group wrestled with several important issues that affect the data acquisition plan and the prioritization of sites. The WG tried to develop a plan that honored the primary EarthScope goals, recognizing the limited funding available. In particular, in order to maximize the coverage obtained and serve the broadest community, the WG elected to utilize relatively narrow swath widths (typically 1 km, widened to 2+km in key regions), which allowed more line-kilometers of data to be obtained. The unavoidable consequences of this choice are that areas away from the main fault strands will be unsampled. Data acquisition is underway. The Northern California acquisition (supplemented by financial support from state and federal agencies such as the PUC and USGS) was completed in April 2007, with processed data available through UNAVCO data management tools (including the GEON LiDAR workflow project). Additional data acquisition for the remaining targets is ongoing with all data acquisition to be completed by September 2008.