H51L-01
Using Remote-sensing to Survey Topography and Morphologic Change on Large Braided River Beds
Since 1999 we have made extensive use of a variety of remote-sensing technologies to survey bed topography over reaches of large braided gravel-bed rivers on the east coast of New Zealand's South Island. The motivations have been (i) to collect input and validation data for 2-d hydrodynamic models for quantifying in-stream physical habitat and for predicting flood levels and (ii) to survey spatially-distributed riverbed erosion and deposition in order to estimate bedload fluxes by the ‘morphological' method. Typical applications have been to river reaches 3-4 km long and 1 km wide, with grid cells from 1-5 m. We use different techniques to survey dry and wet areas of braided riverbed. For dry areas, we have used digital photogrammetry and infra-red airborne LiDAR. For wetted channels, we have generally used ortho-rectified colour imagery or multi-spectral scanning to map water depth, then we map bed topography by subtracting the water depth from a DEM of the water surface obtained from photogrammetry or LiDAR. The imagery is calibrated to water depth using field measurements on the day of imagery acquisition. Surveys are undertaken during low flows to maximise bed exposure. We use ground-based RTK-GPS and echo-sounding to collect calibration and validation data, and sometimes simply use these methods to survey the wetted areas. Orthoimagery at multiple river flows is used to validate 2-d model results. We have been able to achieve elevation accuracies at interpolated points of the order of 10-15 cm for dry areas. This accuracy typically degrades to 20-30 cm for wetted areas. Our experience has exposed a number of issues relating to survey accuracy and practicality at large river scales. These include: changing geoidal models between surveys; local systematic error with photogrammetric model mosaics; geospatial synchronisation of multi-platform data; time-synchronisation of LiDAR and imagery- collecting aeroplanes and suitable weather and river conditions; confusions in water depth mapping; and the critical importance of good data at key hydraulic controls for eco-hydrologic applications. We suggest that high resolution bathymetric LiDAR offers the best potential for future surveys in large river reaches. While the current bathymetry LiDAR systems do not appear to deliver a significantly better accuracy of submerged bed elevations than we have achieved with mixed-technology approaches for dry and wet areas, and their cost remains high, a one-stop package is hard to beat in terms of practicality and data synchronisation.
H51L-02
Modelling Braided River Morphodynamics With Terrestrial Laser Scanning
Advances in topographic survey and terrain modelling have enabled a revolution in the study in the fluvial morphodynamics in the last decade. Prior to the advent of electronic tacheometry in the 1990s, the analysis of channel dynamics was typically inferred from a combination of cross-section surveys and planform mapping. Distributed surveys acquired with GPS or EDMs enabled this analysis to be dimensionally extended and the pattern and magnitude morphodynamics elucidated in 3D; in particular through DTM differencing. Continuing developments in survey technology are now posed to reset this field once again. Now no longer confined to the laboratory, ruggedized laser scanners are capable of acquiring between 4-50,000 observations per second, at ranges exceeding 100 m. This latest development creates the potential for typical reach-scale (1-10 km) topographic datasets to rise in size by 7 orders of magnitude (hundreds to billions of points) in the coming years. Terrestrial Laser Scanning (TLS) offers a wealth of opportunities to better monitor fluvial systems; improving models of cut-and-fill, roughness and enhancing the prospect for ever more detailed parameterizations for fluid models. While this technology enables the creation of ‘virtual facsimiles' of landscapes, the demands of storing, processing and modelling geomorphological products from such data requires a wholesale reappraisal of our data management and modelling methods. Here we outline a field-to-product methodology for TLS of fluvial systems using data from two annual surveys of a 1 km reach of the River Feshie, Scotland. These surveys delivered 3D point cloud datasets, incorporating over 200 million xyz observations, with median spatial densities of over 1000 pts/m2. The surveys were fixed to a GPS-based control network, including over 200 coincident tie- points to register multiple setups to a global coordinate system (RMS errors 0.002-0.011 m). Modelling reach-scale geometries from such dense point clouds poses a non-trivial computational problem and required the development of a bespoke spatial filtering toolbox. This was designed to allow intelligent decimation of TLS data and extract multi-resolution and statistical data suitable for describing bar-scale morphologies over the entire reach, whilst retaining grain-scale information. We use this toolbox to explore the precision and reliability of a morphological sediment budget for the study reach, following a 10 year flood in November 2006. The results are benchmarked against a traditional survey/DTM methodology based on GPS data. http://www.aber.ac.uk/iges/staff/brasingtonjames.shtml
H51L-03
A critical evaluation of grid-by-number sediment sampling using laser scanner derived clast population statistics across a gravel bar
Water flow level in river channels is moderated by the interaction with the roughness of the surface over which it flows. The interaction is highly complex and remains poorly understood despite its economic and social importance in flood level forecasting. The empirical and semi-rational nature of approaches used to estimate hydraulic roughness makes them very difficult to apply and much of the hydraulic resistance has been attributed to grain roughness using various forms of the Colebrook-White equation where the grain diameter is modified by a multiplier to account for the non-uniform nature of gravel-bed surfaces. Fundamental to the accuracy of the particle size approaches is the sampling of river-bed gravels where sample size, operator bias, particle shape and surface heterogeneity can greatly affect the result. Despite these problems a standard surface sample of the intermediate axis of 100 clasts remains the accepted method for grain-size characterisation amongst scientists and engineers concerned with channel hydraulics. Surface roughness has also been measured using a random field of spatial elevation data. The success of this approach has been tempered by the lack of high-resolution topographic data covering all roughness scales, however, improved data-point resolution is now achievable using terrestrial laser scanning technology. The aim here is to reliably quantify the population grain-size distribution of a natural gravel surface using random field terrestrial laser scanner x,y,z data and by direct comparison to demonstrate the errors inherent in the conventional particle-size approach. Application of the random field approach, using a terrestrial laser scanner, across a gravel bar surface on the River South Tyne at Lambley, UK, generated an effective sample of 120,000 clasts yielding a D84 for use in the Colebrook White equation of 0.110m. Monte Carlo sampling within the 12000 measured clasts from the bar surface generated 560 simulated grid-by-number D84 estimates. Grain-size D84 values ranged from 0.100m to 0.195m with a median value of 0.130m. This represents an average 18% and a maximum 77% over-estimation of the grain-size value in the flow resistance equation. Such potential errors, inherent with the conventional grid-by-number sampling technique, impact significantly on flood level estimation options.
H51L-04
Comparison of Terrestrial LiDAR and Erosion Pin Networks for Bank Geometry Monitoring
Erosion pins and repeat surveys are valuable tools for measuring geomorphic change where airborne remote sensing platforms do not provide the required accuracy or are impractical. Recent advances in automated systems, e.g. the Photo-Electronic Erosion Pin system, permit the collection of high temporal resolution data, however, these systems do not address the inherently low spatial resolution of erosion pin networks and high spatial resolution digital terrain models (DTM) are time consuming to produce with standard surveying equipment. In contrast, Terrestrial LiDAR systems (TLS) enable the rapid generation of DTMs that routinely contain several thousand data points per m2 without disturbing the target area. We describe the use of TLS to monitor changes in bank geometry along an 800 m reach of the Rio Grande in Albuquerque, NM with comparison to data from an erosion pin network along the same reach. On 31 July 2006, shortly after the inception of our monitoring campaign, a thunderstorm produced flows out of the Calabacillas Arroyo that deposited over 10,000 m3 of sediment into the main stem of the Rio Grande within the study area. These deposits reduced the width of the Rio Grande by half and buried nearly all of the erosion pins downstream of the arroyo. Rio Grande and Calabacillas flows continued to rework these deposits since that time. This sediment pulse exceeded the measurement capacity of the erosion pin network and it only registered a large event, followed along part of the reach by a return to near pre-event conditions months latter. In contrast, the cm scale pre- and post-event DTMs produced by TLS document local changes in geometry and enable volumetric estimates of sediment gain and loss. These are minimum estimates, however, as the DTMs only include sediments exposed by low flows at the time of the scans. Furthermore, as with all resurvey techniques, the temporal resolution of the TLS time series is limited by the frequency with which the site could be reoccupied. Although additional work is needed these preliminary results demonstrate the utility of TLS for documenting geomorphic change with cm scale accuracy, in particular episodic large-magnitude events that may preclude direct measurement.
H51L-05
Composite Digital Terrain Models: Synthesizing Aerial and Terrestrial LiDAR with Conventional Survey Data to Monitor Sediment Transport Through the Sunol Dam Removal Site
In 2006, the San Francisco Public Utilities Commission removed Sunol dam, located on Alameda Creek, near San Francisco California. The primary goals of the project were to improve fish passage, restore a self- sustaining population of steelhead to the watershed, and eliminate an existing public safety hazard. Approximately 28,300 cubic meters of sand and gravel-sized sediment had accumulated upstream of the dam and was left in place to move downstream naturally over a period of several decades. To create a baseline for future monitoring of sediment transport through the dam area, a combination of Aerial LiDAR, Terrestrial LiDAR, and conventional survey data was compiled and synthesized to generate a three dimensional digital model of the study area both upstream and downstream of the damsite. The primary survey method for characterization of above ground topography was Terrestrial LiDAR, with an approximate point spacing of centimeters. In submerged areas conventional survey techniques were used to augment the Aerial and Terrestrial LiDAR data sets. We found this approach to be effective in developing a high accuracy-high detail sediment volume model from which sediment transport can be monitored and modeled.
H51L-06
High Resolution Mapping and Interpretation of Channel and Floodplain Topography With a Narrow-Beam Terrestrial-Aquatic Lidar
Basic description of channel and floodplain topography remains a fundamental challenge for modeling flow and sediment transport or even simply mapping habitat. Standard field wading and boat surveys of stream topography are limited by costs and logistics to relatively small sample reaches and floodplain maps are seldom well- integrated with channel bathymetry. We used the NASA Experimental Advanced Airborne Research Lidar (EAARL) to map channel and floodplain topography and investigate geomorphic controls on physical habitat in two diverse channels in the watershed of the Middle Fork Salmon River, Idaho. Bear Valley Creek is a small low-gradient gravel-bed stream flowing across an unconfined valley filled with glacial outwash materials. A hierarchy of nested geomorphic features is evident in this channel with the broadest fluvial domains a legacy of ~15,000 years of post-glacial valley evolution. Contemporary hydraulics operate on this broad template and control two smaller scales of pool-riffle morphology. Salmon spawning patterns closely reflect these nested physical domains, demonstrating how geomorphic history can influence modern distributions of aquatic habitat and organisms. In contrast, Big Creek is a higher-gradient stream predominately confined by steep side slopes in a deep valley. Here, the distribution of geomorphic domains and physical habitat is controlled by modern erosion processes and rock quality. Tributaries and valley walls contribute coarse debris, up to large boulders, to the channel, resulting in very rough and poorly organized bed topography. Tributary fans also function as local grade control with sediment deposition in lower-gradient reaches upstream of fans. A GIS toolkit is under development to extract at-a-station channel metrics from EAARL data, including for example, cross section and longitudinal profile characteristics. A new investigation has also begun to further investigate the quality of EAARL data. This study will explore the question of how well we must describe channel topography to adequately: i) map the spatial distribution of physical habitat for management purposes and in support of organism population growth models, and ii) define boundary conditions for flow and sediment transport predictions using the USGS model MD SWMS.
H51L-07
Identifying Subtidal Coastal Environments Using Airborne Lidar Bathymetry (ALB)
Airborne lidar bathymetry (ALB) survey planning differs from one coastal zone to another because environmental factors affect the success of the lidar. Environmental factors are so dominate in the lidar success that the same survey configuration at different times may produce different results. A comparison of results between two different ALB systems (Tenix LADS and Optech SHOALS) in the Portsmouth Harbor, NH and offshore Gerrish Island, ME showed a striking correlation of the lack of bottom detection in shallow waters (3-25 m). This lack of bottom detection is independent of the tide status, the date of data collection, and the direction of the survey flight. Multibeam echosounder measurements (Simrad EM3002) were used as reference measurements. In both the Portsmouth Harbor and Gerrish Island surveys, the lack of bottom detection by the lidar was independent of the bathymetry. Because the water-column environmental factors are directly related to the water depth, these results show that the success of the laser measurements observed here are also independent to the optical properties of the water. A comparison of the laser bottom detection to the seafloor slope shows a close correlation. Steep- sloped features such as bedrock outcrops off Gerrish Island result in a lack of bottom detection by the lidar. The multibeam echosounder backscatter in the Portsmouth Harbor and Gerrish Island surveys shows the acoustic properties of the seafloor also have a high correlation between the backscatter intensity to the areas lacking bottom detection by the lidar surveys. Ground-truth underwater video imagery in the Portsmouth Harbor area show that the seafloor in areas of successful bottom detection by the ALB are composed of sands, whereas the seafloor in areas that produced a lack of bottom detection are composed of pebbles and rock outcrops. To date, the only environmental factor that is considered in ALB survey planning is the water column (diffuse attenuation coefficient, Kd). The observations presented here show that in water depths deeper than 3 m, the surficial characteristics of the seafloor becomes a dominant environmental factor that affects the success of bottom detection with ALB.
H51L-08
High-Resolution Characterization of Intertidal Geomorphology by TLS
Observational fluvial geomorphology has greatly benefited in the last decades from the wide availability of digital terrain data obtained by orthophotos and by means of accurate airborne laser scanner data (LiDAR). On the contrary, the spatially-distributed study of the geomorphology of intertidal areas, such as tidal flats and marshes, remains problematic owing to the small relief characterizing such environments, often of the order of a few tens of centimetres, i.e. comparable to the accuracy of state-of-the-art LiDAR data. Here we present the results of Terrestrial Laser Scanner (TLS) acquisitions performed within a tidal marsh in the Venice lagoon. The survey was performed using a Leica HDS 3000 TLS, characterized by a large Field of View (360 deg H x 270 deg V), a low beam divergence (< 6 mm at 50 m) and a nominal accuracy of 6 mm at 50 m. The acquisition was performed at low tide to avoid interferences due to water on the marsh surface and, to minimize shadowing effects due to the tilting of the laser beam (especially in the channel network), the scanner was mounted on a custom-built tripod 3 m above the marsh surface. The area of the marsh, about 100m x 150m, was fully surveyed by just 2 scans. A total amount of about 3 million points was acquired, with an average measurement density of 200 points/m2. In order to reconstruct the geometry of the marsh, the two scans were co-registered using 8 reflective targets as matching points. Such targets were placed within the area of interest and surveyed with high accuracy (2 mm), while their position in the Italian national grid was determined with a double-frequency GPS receiver, in order to georeference the point clouds within an absolute framework. Post-processing of the very high resolution data obtained shows that the laser returns coming from the low vegetation present (about 0.5-1.0 m high) can be satisfactorily separated from those coming from the marsh surface, allowing the construction of a DSM and a DTM. This is important e.g. in eco-geomorphic studies of intertidal environments, where conventional LiDAR technologies cannot easily separate first and last laser returns (because of the low vegetation height) and thus provide models of the surface as well as of the terrain. Furthermore, the DTM is shown to provide unprecedented characterizations of marsh morphology, e.g. regarding the cross-sectional properties of small-scale tidal creeks (widths of the order of 10 cm), previously observable only through conventional topographic surveys, thus not allowing a fully spatially-distributed description of their morphology.