- SPECIAL REPORTS
- THE MAGAZINE
In the early morning hours of Dec. 14, 2005, in the heart of the Missouri Ozarks, a 20-foot wall of water rushed down into the east fork of the Black River, flooding portions of Johnson's Shut-Ins State Park and causing substantial damage to vegetation and aquatic habitat. Caused by the catastrophic failure of the Taum Sauk Hydroelectric Plant Upper Reservoir, the flood surge swept away the state park superintendent's home and several vehicles from a nearby state highway into an adjacent field. Several injuries were reported. Fortunately, the event occurred during the park's off season and there were no fatalities.
The 1.5 billion gallon Upper Reservoir, owned and operated by St. Louis-based AmerenUE (the electric utility subsidiary of Ameren Corporation), was built in 1963 on the top of the 1,590-foot Proffit Mountain. Under normal conditions, water was pumped from the lower reservoir to the upper reservoir through a 7,000-foot shaft when power demand was low, then released through the same shaft by turning turbine generators during peak electrical demand.
The breach of the Upper Reservoir brought considerable attention and interest from state and federal agencies. The Missouri Department of Natural Resources collaborated with the U.S. Geological Survey Water Resources Division (USGS-WRD) to discuss first steps toward implementing remediation phases for Johnson's Shut-Ins State Park. The two agencies determined that immediate acquisition of Light Detection And Ranging (LiDAR) data would be valuable to provide robust analyses for models simulating the dam breach and flood frequency.
The USGS Mid-Continent Mapping Center in Rolla, Mo., was asked to identify a qualified company to provide LiDAR acquisition and processing to produce a highly detailed bare-earth digital elevation model (DEM). Under the Cartographic Services Contract (CSC-2), the USGS works with several companies and their partners to fulfill any mapping services the USGS may have. Within 24 hours following the dam failure, the USGS contacted Sanborn, a mapping company based in Colorado Springs, Colo., to deploy an aircraft and LiDAR sensor to the afflicted area. Sanborn mobilized immediately on Dec. 16, 2005, and successfully completed the LiDAR mission that evening.
Project Specs and DeliverySanborn's LiDAR acquisition specialist acquired the data for the DEM over a 31.4 square mile area using the company's Leica ALS50 Airborne laser scanner LiDAR system. The LiDAR sensor was calibrated by conducting a flight pass over a known ground surface prior to flying the mission. During final data processing, the calibration parameters were inserted into Terrasolid's (Finland) TerraScan version 6 post-processing software for processing, editing and classifying the LiDAR points. Surface models were produced using Terrasolid's TerraModeler software. Two airborne GPS base stations were tied to three additional NGS monuments, each to create a GPS survey network for the project. The network observations and adjustments were completed on the GRS80 ellipsoid.
The project required an average post spacing of 0.7 meters, meaning that the LiDAR elevation point is calculated by the sensor within every 0.7 m of area. The bare-earth DEM data was completed at a vertical accuracy of 15 cm (Root Mean Square, or RMS) and horizontal accuracy of 0.5 m (RMS). The LiDAR data was processed to obtain both the first and last return point data collected by the sensor, thereby enabling the analysis of both ground and canopy (above-ground) features. The last return data was further filtered to yield a surface representing the bare earth. Filtering was performed to remove 97 percent of all vegetation and 99 percent of all buildings. Sanborn delivered the completed bald-earth DEM within two weeks of the original data acquisition. Conventional terrain models would have required considerably more time and effort, including acquisition of aerial photography and stereocompilation mapping by experienced photogrammetrists.
Modeling Flood ProfilesAccording to Paul Rydlund, a hydrologist with the USGS-WRD, a detailed study is underway to define the flood profile from the upper reservoir to the dam of the lower reservoir, to determine peak discharge and to map the inundation area of the flood event. The study will also define spatially and volumetrically the movement of debris material. The LiDAR data serves as primary input data for much of this analysis. Flood profiles will be established and will include data from along the flow path downstream from the dam breach to the east fork of the Black River. The 100- and 500-year flood profiles will be established for several areas including from just above the upper reservoir drainage downstream to the lower reservoir. These profiles will be established using ground geometry obtained from the LiDAR-generated elevation data.
The USGS-WRD is utilizing RiverCAD software (Boss International, Madison, Wis.) to create one-dimensional models that will gauge the magnitude of the flood event. Two-foot contours were generated from the LiDAR data and input as primary source data to support development of cross sections and slope calculations. According to Rydlund, "the fundamental hydraulic principal in computing discharge (cubic feet per second) of water involves a slope-area methodology. The model will compute many different hydraulic parameters at each cross section including water-surface elevation, velocity and area. The models will also simulate the dam breach, and flood inundation maps will be produced for both the dam break event and flood frequency discharges (2-, 5-, 10-, 25-, 50-, 100- and 500-year flood events)."
A substantial amount of debris was deposited both upstream and downstream from where the floodwaters entered into the east fork of the Black River and into the lower reservoir. Using sources such as the available pre-flood survey data and LiDAR data obtained after the flood, the area and volume of erosion and deposition are being defined and computed. Results and conclusions of the analysis will be provided in a USGS Scientific Investigations Report, documenting the flood profile, peak discharge and debris movement from the flood. The LiDAR data will also be made available and include appropriate metadata.
Investigating the CauseThe actual cause of the dam breach is still under investigation. A recent report released from AmerenUE provides the results of a study to examine the cause of the dam failure. The study cited "stability failure" as the root cause of the breach, stating that the reservoirs' failure began shortly after overtopping the dam wall. According to the report, overtopping resulted when water level instrumentation failed to shut off the pumps, thereby overflowing the reservoir banks. Backup water level protection probes were not properly maintained as designed and did not shut the pumps down when the primary system failed. Due in part to a large quantity of fine-grained material used in the dam construction, water was unable to flow through the dam, resulting in catastrophic failure of the dam wall. In a regulatory filing submitted by Ameren Corporation, the damages and liabilities caused by the reservoir breach will cost between $53 and $73 million.
As demonstrated in the remediation efforts following the Taum Sauk reservoir dam failure, LiDAR technology has proven itself as a highly accurate and rapid response mapping tool for flood analysis and hydrologic modeling. Data collected using this technology will provide scientists and engineers with the specifics they need to investigate the cause of the Taum Sauk reservoir breach. In time, the Johnson's Shut-Ins State Park should be fully accessible and ready to serve residents and visitors of the beautiful Missouri Ozarks area.
Sidebar: LiDAR PracticumDuring a LiDAR mission, a LiDAR sensor relies on a scanning mirror to generate a swath of light pulses. To determine elevation, the sensor measures the amount of time it takes for a light pulse to travel from the sensor to the ground (or an above-ground feature) and back. Variables including flying height, swath angle, scanner rate and aircraft velocity determine the point density for a given project; these factors are customized according to project needs. A geodetic grade GPS receiver measures the aircraft position (e.g. latitude and longitude) every second and an Inertial Measurement Unit (IMU) computes the aircraft's trajectory about 200 times per second.
The most modern LiDAR sensors are capable of acquiring up to 100 kHz (100,000 pulses per second) resulting in massive data sets. However, depending on the application, this point density may not be required to produce a highly detailed terrain model. Since LiDAR is an "active" sensor, data can be acquired day or night. Missions are routinely flown at night when air turbulence is diminished and air traffic is light.
The most common LiDAR deliverable is a bare-earth digital elevation model (DEM), where all above-ground features (vegetation, buildings, etc.) have been removed, depicting only the terrain. The density of the elevation points varies depending on the terrain relief and amount of vegetation, but most projects specify a minimal point spacing to ensure adequate coverage. Dense vegetation can reduce the ability of the light pulses to penetrate the ground, but this can be offset by increasing the point density.
The vertical accuracy of each LiDAR point is precise, typically 15 cm root mean square error (RMSE) or better while horizontal accuracy can exceed .50 m. A LiDAR intensity image that reveals the intensity of the light pulse back to the sensor can also be produced. These images can be viewed stereoscopically for use in delineating terrain breaks and drainage features, thereby increasing the accuracy of the bald-earth DEM and derived contours.
LiDAR data has become widely accepted and used for floodplain mapping studies and contour generation, and commonly serves as the input terrain model for orthorectification of aerial imagery. Other LiDAR applications include forest canopy studies, 3D modeling of urban areas, telecommunications, transportation engineering and volumetric analysis.