H52E-01
Airborne LIDAR Data Processing and Analysis Tools
Airborne LIDAR technology allows accurate and inexpensive measurements of topography, vegetation canopy heights, and buildings over large areas. In order to provide researchers high quality data, NSF has created the National Center for Airborne Laser Mapping (NCALM) to collect, archive, and distribute the LIDAR data. However, the LIDAR systems collect voluminous irregularly-spaced, three-dimensional point measurements of ground and non-ground objects scanned by the laser beneath the aircraft. To advance the use of the technology and data, NCALM is developing public domain algorithms for ground and non-ground measurement classification and tools for data retrieval and transformation. We present the main functions of the ALDPAT (Airborne LIDAR Data Processing and Analysis Tools) developed by NCALM. While Geographic Information Systems (GIS) provide a useful platform for storing, analyzing, and visualizing most spatial data, the shear volume of raw LIDAR data makes most commercial GIS packages impractical. Instead, we have developed a suite of applications in ALDPAT which combine self developed C++ programs with the APIs of commercial remote sensing and GIS software. Tasks performed by these applications include: 1) transforming data into specified horizontal coordinate systems and vertical datums; 2) merging and sorting data into manageable sized tiles, typically 4 square kilometers in dimension; 3) filtering point data to separate measurements for the ground from those for non-ground objects; 4) interpolating the irregularly spaced elevations onto a regularly spaced grid to allow raster based analysis; and 5) converting the gridded data into standard GIS import formats. The ALDPAT 1.0 is available through http://lidar.ihrc.fiu.edu/.
H52E-02
Integration of Airborne Laser Scanning Altimetry Data in Alpine Geomorphological and Hazard Studies
A digital terrain and surface model derived from an airborne laser scanning (ALS) altimetry dataset was used in the Austrian Alps for the preparation, improvement and the evaluation of a digital geomorphological hazard map. The geomorphology in the study area consists of a wide variety of landforms, which include glacial landforms such as cirques, hanging valleys, and moraine deposits, of pre- and postglacial mass movement landforms and processes, such as deep seated slope failures, rock fall, debris flows and solifluction. The area includes naked and covered gypsum karst, collapse dolines and fluvial landforms and deposits such as river terraces, incisions, alluvial fans and gullies. A detailed symbol based paper geomorphological map served as a basis for the digitalization of basic morphogenetic landform and process units. These units were assigned a `geomorphological unit type`, `hazard type` and `activity` code in the attribute table, according to a morphogenetic classification scheme. Selected zonal statistical attributes - mean height, aspect and slope angle - were calculated in a GIS using the vector based morphogenetic landform and process units and the underlying 1m resolution laser altimetry raster dataset. This statistical information was added to the attribute table of the `geomorphological hazard map`. Interpretation of the zonal statistical information shows that indicative topographic signatures exist for the various geomorphological and hazard units in this region of the Alps. Based on this experience a further step is made towards semi-automated geomorphological hazard classification of segmented laser altimetry data using expert knowledge rules. The first results indicate a classification accuracy of 50-70 percent for most landform associations. Areas affected by slide processes resulted in less accurate classification, probably because of their polygenetic history in this area. It is concluded that the use of lidar data improves visual recognition of landslides, especially in combination with high resolution 3D air-photo information. The zonal statistical information helps to more realistically delineate, identify and classify landscape units and promotes consistent data extraction. This leads to efficient and time saving (semi-automated) classification procedures in alpine areas. For future developments it seems promising to further integrate lidar in automated landscape classification tools and in hazard mapping strategies. If high resolution satellite borne laser data becomes available on a regular base a further step towards monitoring of landscape evolution in geomorphological hazard studies can be made.
H52E-03
Automatic Delineation of Drainage Basins From Contour Elevation Data Using Skeleton Construction Techniques
A new method for automatic delineation of drainage basins from contour elevation data is presented. As a preprocessing step, contour line vertices are used to construct Delaunay and Voronoi diagrams along with other useful structures known in computational geometry as the crust and the skeleton or medial axis transform. Using the skeleton of contour lines, a recursive algorithm is then developed to solve critical topographic structures such as ridges, saddles, and peaks in a fully-automated and accurate manner. Numerical experiments based on high- accuracy contour elevation data of real terrains (generated from LiDAR surveys) show that the proposed method is able to process automatically any topographic structure and to produce results that are comparable to those that can be interpreted visually from contour lines. The gain in accuracy over state-of-the-art solutions is generally found to be significant and to increase as the contour interval increases. Finally, it is shown how the proposed method can be easily applied to construct accurate flow nets in a fully-automated manner. Skeleton construction techniques allow the morphological information implicitly present in contour elevation data to be explicitly revealed and appropriately processed by a computer program, and therefore appear useful means for improving the accuracy with which physiographic features of drainage basins are determined. The proposed method can be used to advance the construction of flow nets and contour-based digital elevation models (as outlined in this study) and to test the reliability of algorithms for the analysis of more efficient and straightforward, gridded or triangulated, elevation data (as shown in a companion study).
H52E-04
Relict Landscape Response to Knickpoint Migration on the Roan Plateau, Western Colorado, Explored Through ALSM Data Analysis
The unprecedented spatial resolution of Digital Elevation Models (DEMs) derived from Airborne Laser Swath Mapping (ALSM) makes them ideal for detecting subtle morphologic features. We explore to what extent information about knickpoint migration is communicated upstream by analyzing an ASLM-derived DEM in a transient landscape. We target the Roan Plateau in western Colorado, a landscape developed in flat-lying Eocene shales, in which multiple upstream-migrating waterfalls triggered by base level fall have incised dramatic canyons in their wake. The waterfalls separate low-gradient, bedrock-floored reaches above the waterfalls from steep, boulder-choked canyons below. Similarly, a sheer canyon rim separates a smooth, relict landscape above the canyon walls from the steep cliffs and talus-mantled slopes below. Waterfall and canyon rim elevations correlate well with the outcrop of a resistant oil-shale layer. We use a 1-meter DEM to develop two simple metrics that detect channel and hillslope response of the upper landscape to knickpoint migration. The first metric is stream profile analysis of upper plateau tributaries. In a steady-state stream channel, slope should decrease with distance downstream, as drainage area and the associated water discharge increase. Departures from this trend can be attributed to either lithologic variation, or transient oversteepening that may be associated with the presence of the waterfall. Oversteepening of the channel can incite hillslope response, and give rise to a box canyon upstream of the waterfall. We document the slope of channels as they approach the free overfall, and the presence and lengths of box canyons upstream of the waterfall lip. The second metric is the curvature of plateau hilltops. In a steady state landscape, hilltops should be roughly parabolic in cross-section, reflecting a balance between a uniform rate of regolith production and diffusive transport at all points along the hillslope. Departure from this parabolic form can reflect lithologic variation, transient hillslope response to increased rates of stream incision, or transport processes that are not linearly dependent on slope. Isolation of roughly the upper 100 m of hilltops and evaluation of the curvature of these crests provides constraint on the ratio of weathering rate to transport efficiency in the landscape most likely to be in steady state. Changes in crest curvature with distance from the canyon rim can be used to document transient hillslope response of the upper plateau surface in areas with uniform lithology. Paired with field observations and mapped bedrock contacts, analysis of an ASLM-derived DEM allows us to evaluate the extent to which the upper plateau channels and hillslopes have responded to knickpoint migration and the carving of canyons downstream. Morphologic evidence for significant upper plateau response to this incision event would suggest that the timescale for landscape adjustment to base level fall may be shorter than that required to propagate a knickpoint upstream.
H52E-05
Spectral Analysis for Characterizing Landslide-Prone Terrain Using High Resolution Topographic Data
Analyses of surface features in landslide-prone terrain can provide insight into spatial and temporal patterns of mass movement as well as landslide mechanics. Traditional analyses involving topographic maps, air photos, and field observations are often subjective and limited by time constraints, vegetative cover, and low-resolution topographic data. Various statistical measures of surface roughness developed and automated in recent years have improved landslide detection and characterization, but an objective, comprehensive technique remains elusive. Here, we apply a two-dimensional discrete Fourier transform (DFT) to 1m resolution LiDAR data from the Eel River catchment in northern California to analyze the spatial extent and internal structure of large earthflows prevalent throughout this region. The 2D DFT provides information about the spatial frequency, amplitude, periodicity, and orientation of topographic features, such as headscarps, lateral levees, and compressional folds, through a range of spatial scales. To highlight patterns in the meter-scale roughness typical of earthflows, we implement a moving window algorithm that computes the DFT periodogram for a window of specified width at each point in the DEM. By filtering the resulting spectral power matrices according to spatial frequency and/or orientation, we can identify several key features of earthflows. Parallel levees and gullies of ~10m width typically characterize the boundaries of an earthflow's main body, whereas sharp scarps of similar scale typify the boundary at the head of the flow. The spectral power of elements in periodograms oriented perpendicular to these boundaries tends to be orders of magnitude greater than in other portions of the terrain. Blocky, hummocky terrain within the earthflow has a unique spectral signature when the DFT periodogram is filtered to include spatial frequencies of comparable scale. In addition, we use the DFT in our moving window algorithm to generate one-dimensional power spectra for ~30000m2 patches of terrain representative of active and dormant earthflows and unfailed terrain. The upper envelopes of these spectra are lowest for the unfailed terrain, up to an order of magnitude higher for the dormant earthflows, and up to two orders of magnitude higher for the active earthflows. This indicates that once an active earthflow becomes dormant, erosional processes systematically subdue its roughest meter-scale features through time. Use of the 2D DFT in a moving window algorithm will enable us to map spatial and temporal patterns of instability in landslide-prone terrain with improved efficiency, objectivity, and precision.
H52E-06
Beyond Cross Sections; LiDAR in Support of Sprague River geomorphology Studies, Klamath Basin, Oregon
LiDAR terrain data was collected in November 2004 for 750 square kilometers of the Sprague River valley, Oregon, by Watershed Sciences, Inc., under contract with the Klamath Tribes. This coverage, obtained to support multiple ecologic analysis and restoration activities in the Klamath Basin, encompasses about 90 km of valley- bottom corridor for the main Sprague River as well as downstream alluvial sections of principal tributaries, including 15 km of the Sycan River, 15 km of the South Fork Sprague River, and 10 km of the North Fork Sprague River. Acquisition conditions were leaf-off at normal fall low flow. Assessment of the vertical divergence between 967 surveyed points and the laser points gave a RMSE of 0.052 m with a standard deviation of 0.051 m. The resulting bare-earth 1-m grid has been used for geomorphic mapping, paleo- and historical-channel mapping, qualitative and quantitative analysis of floodplain morphology, and assessment of channel incision over various timescales.4 Using the high resolution topography in combination with floodplain stratigraphy, we delineated the late Holocene (post 7.7 ka Mazama eruption) active floodplain. The morphology of this floodplain reveals processes locally important in forming the Sprague and Sycan river floodplains, including avulsion, meander abandonment and filling, lateral bar building, deposition by crevasses, and lateral migration. Mapping present and paleo-channel positions determined from historic photos and maps onto the LiDAR digital terrain model, coupled with augering to determine channel gravel depths, has allowed local assessment of long term channel incision and aggradation over the last several thousand years. Reaches of possible historic incision have also been delineated by comparing floodplain elevation to channel elevation in an essentially continuous manner along the valley corridor. This analysis reveals multiple reaches where the 1940 channel, as depicted on aerial photographs, is apparently higher than the 2004 channel, and these incised reaches generally correspond to reaches of lower sinuosity (locally imposed by channelization) and rates of channel migration. The continuous and high-resolution elevation data provided by LiDAR terrain models provides a means for continuous mapping of floodplain and channel features at scales relevant to geomorphic and ecologic processes on many river systems, thus providing a means to obtain significantly more information on state and process than can typically be accomplished by the sparse sampling of river systems afforded by a cross-section based approach.
H52E-07
The Effectiveness of Airborne LiDAR Data in the Recognition of Channel-bed Morphology
High-resolution topographic data have the potential to differentiate the main morphological features of landscape. New survey techniques that have been introduced in the last few years may significantly contribute to this objective. One of these techniques is the Airborne Laser Swath Mapping (ALSM), also known as Light Detection And Ranging (LiDAR). A valuable characteristic of this technology, which marks advantages over the traditional topographic survey techniques, is the capability to derive a high resolution Digital Terrain Model (DTM) from the last pulse LiDAR data by filtering the vegetation points. In this work, we tested the capability of airborne LiDAR- derived data in the analysis of channel-bed morphology of a small alpine stream. For the purpose of this study, 0.5 m and 1 m resolution Digital Terrain Models (DTMs) were derived from the bare ground LiDAR data. The analysis was carried out both at 1-D scale, i.e. along the longitudinal channel profile, and at 2-D scale, taking into account the whole extent of the channel bed. The 1-D approach analyzed the residuals of elevations orthogonal to the regression line drawn along the channel profile and the standard deviation of local slope. The 2-D analysis was based on a roughness index, consisting on the local variability of the elevation, defined as the standard deviation of residual topography. The study area is the Rio Cordon basin, a headwater catchment located in the Dolomites, a mountain region of the Eastern Italian Alps. We analyzed the main channel, which is characterized by an alternation of different morphologies. The results suggested a good capability of LiDAR data in the recognition of channel-bed morphology giving the potential to distinguish the riffle-pool and step-pool reaches.
H52E-08
High resolution, spatial distribution of unit stream power using ALSM-derived stream networks
Unit stream power is a measure of energy expenditure, as the channel does the work of transporting sediment and modifying channel morphology. The distribution of energy through the channel network creates a distribution of channel morphologic types by taking disorganized hillslope inputs of wood and sediment and creating channel structure, organization, and regularly spaced bedforms. Unit stream power has been used as a proxy for fluvial process, to establish causality between channel form and process. However, spatial distributions of unit stream power have been constrained by the coarseness of the available topographic data. ALSM data was used to create high resolution spatial distributions of unit stream power on a reach of headwater mountain stream. Spatial distribution of unit stream power at fine scales was used to detect spatial organization in a reach of Spring Park Creek, Tenderfoot Creek Experimental Forest, Montana. ALSM-derived digital elevation models of varying resolution were used to delineate stream networks. Channel slope derived from stream networks, discharge data from an in-situ gaging station, and field surveys of channel width were used to calculate unit stream power distributions at 0.5, 1.0, and 5.0 meter resolution. As a secondary method of calculating unit stream power, a HEC-RAS hydraulic model of the reach was created by cutting cross-sections through the ALSM-derived topography and inserting thalweg elevations to define channel inverts. This model was calibrated by stage- discharge data and field-derived roughness values. The HEC-RAS-calculated values of unit stream power were used to create spatial distributions of unit stream power. Spatial distributions of unit stream power derived by both methods were analyzed by spectral analysis for periodic behavior. The relative dominance of alluvial versus colluvial forcing constrains the degree of organization.