Scientific Investigations Report 2007–5012
U.S. GEOLOGICAL SURVEY
Scientific Investigations Report 2007–5012
Bathymetric data were collected over the navigable areas of Walker Lake from February through April, 2005 using the same equipment as Baskin (2005) in the Great Salt Lake. Depth and position data were collected using an automated system consisting of a single-beam echosounder (fathometer) coupled to a real-time, differentially-corrected global positioning system (GPS). Digital data were logged into a navigational computer, and a paper chart was used to backup digital data in case of computer malfunction. Depths were derived indirectly by measuring the time required for a sonar (SOund NAvigation and Ranging) signal to travel from the transmitter to the bottom of the lake and back to a receiver. Measured sound velocities were used to correct the basic time-of-travel data and provide a velocity-corrected distance between the transmitter and receiver. Distance from the water surface to the transceiver face was measured and added to the distance between the transducer and the lake bottom to derive the calculated depth of the lake. Paper charts were collected for all transects to cross-check digital readings and to more easily see bottom features during data collection. Temperature or salinity were measured during the survey but there were no gradients, indicating the lake was well mixed (U.S. Geological Survey, 2006). Thus, the sonar signal was not affected by thermoclines or picnoclines.
About 250 mi of transects were collected primarily on east-west transects spaced about 0.31 mi apart (fig. 3). North-south transects were spaced 0.62 mi apart. Transects also were done along the shore of most of the lake. Positional data were recorded at about a 1-second rate and depth data at about 10 points per second. Boat speed was limited to less than 10 mi/h to maximize the number of depth measurements gathered during the data-collection phase. For quality control purposes, data were collected when the lake surface had waves generally less than 1 ft. Because of limitations in the instrumentation and navigating in shallow water, data collection was limited to areas deeper than about 3 ft.
Water depth was measured with a survey-grade Reson, Inc., model 210 single-beam echosounder using a 2.7-degree beam width and 200 kilohertz (kHz) transducer. Manufacturer specifications for this echosounder indicate that the depth of operation is from 0.7 to 1,969 ft and that the depth measurement accuracy is 0.4 in. Speed of sound was set at a constant 5,250 ft/s during data collection. Speed-of-sound corrections were subsequently applied during data processing.
Sound-velocity data were collected with a direct “time-of-flight” sound-velocity sensor at two locations in Walker Lake at 3.28-ft vertical intervals. The sound-velocity sensor used during this study was accurate to ±0.8 ft/s in water with a 0.7 ft/s resolution. These data were transferred to sound-velocity correction files and used during data processing.
Latitude and longitude of depth measurements were obtained with a Trimble AG132 integrated Global Positioning System/Differential Global Positioning System (GPS/DGPS) coupled with an OmniSTAR Wide Area DGPS Solution. The integrated GPS/DGPS is capable of improving regular GPS accuracies to sub-meter accuracy by solving for atmospheric delays and weighting of distant base stations. Data were collected and processed while surveying and recorded with the unprocessed depth data while in the field. Position data were collected only during times of good satellite visibility and when OmniSTAR Wide Area DGPS Solution differential corrections were available. Differentially-corrected data were acquired for all data obtained during this study and positional accuracy is estimated to be 3.28 ft or less.
The PC-based Windows navigational and bathymetric mapping software, Hypack Max, was used to plan and manage the bathymetric surveys and edit and manage the data collected along each lake transect. This software was installed on a portable computer interfaced with the DGPS and survey-grade depth sounder for navigational use. Hypack Max was configured to display and track the boat position against a background of survey control lines and later used for processing of the bathymetric data.
Lidar (Light Detection and Ranging) is an active remote sensing technique that was used in this study to accurately measure land-surface altitude. Similar to the echosounder, Lidar determines distance by measuring the time for a pulse of photons transmitted from a laser to reflect off the land surface and return to a sensor (Wehr and Lohr, 1999). The National Aeronautics and Space Administration’s (NASA) Experimental Advanced Airborne Research Lidar (EAARL) was used during this study to map land-surface altitude of the valley floor and adjacent foothills from Hawthorne to Wabuska, Nev. EAARL is a raster-scanning, waveform-resolving, green-wavelength Lidar system designed to map near-shore bathymetry, topography, and vegetation structure simultaneously. The EAARL sensor records the time history of the reflected energy within a small area, called the footprint, for each laser pulse. The time history enables characterizing vegetation canopy structure in addition to “bare-earth” altitude under a variety of vegetation types.
The EAARL has been operational since the summer of 2001, when it surveyed a coral reef tract in the northern Florida Keys (Wright and Brock, 2002; Brock and others, 2004). The EAARL sensor suite includes a high-resolution multi-spectral color infrared camera, two precision dual-frequency kinematic carrier-phase GPS receivers, and an integrated digital miniature altitude reference system which provides for sub-meter geo-referencing of each laser pulse. The EAARL is operated from a twin-engine Cessna 310 aircraft. Flights were based out of the Reno/Tahoe International Airport. A GPS base station was set up at the airport during the survey to obtain dual-phased differentially-corrected aircraft trajectories.
The EAARL laser samples up to 10,000 times per second from a short (1.2 nanoseconds) pulse of photons. The 1.2 nanosecond pulse is approximately 7 in. long in air and 5 in. long underwater. The system can measure distance with 0.8 to 2 in. accuracy depending on variations in the target reflectivity from pulse to pulse. The footprint of the pulse is 8 in. in diameter at a nominal flying altitude of 1,000 ft. The surveyed swath width, perpendicular to the line of flight, is 790 ft. Sample spacing along the swath is about 8 ft. Swaths are approximately 6.6 ft apart when the aircraft travels at 112 mi/h. The digitally recorded return signal for each transmitted laser pulse is a time history of the reflected photons within the laser footprint.
The EAARL flight survey around Walker Lake was collected from May 26 to June 2, 2005. The 31.3 hour survey acquired 42.5 gigabytes of data including 2.2 million small-footprint waveforms, and a total of 132,064 color infrared digital camera images. The total land area surveyed was 437 mi2.
High resolution imagery was collected in the Walker Lake area by EarthData International, LLC, on June 28, 2005. Aerial imagery was collected at an approximate altitude of 20,500 ft above mean terrain using a Leica ADS40 airborne digital system, which incorporates optics, electronics, data transfer, and storage. Natural color and color infrared ortho-images were produced at 3-ft pixel resolution over an area of 1,364 mi2.
Ortho-rectification processing of the imagery was performed using various programs in the ISTAR system. Using several tools that are part of the ISTAR workflow, a Digital Elevation Model (DEM) was correlated at a post spacing of 6.6 ft depending on terrain and land cover. The ISTAR correlation algorithm computes the latitude, longitude, and altitude for each DEM post utilizing every stereo angle that is available. A series of DEM files were created for each acquisition block. A mosaic was then created from the separate DEM files. The best vertical value for each posting was selected from all look angles compared against the aero-triangulation adjustment, which is incorporated into the mosaic. EarthData then edited the surface to the level required to support the ortho-photo production. A root mean square value was calculated based on the imagery utilized in the production of the tile by comparing the aero-triangulation latitude and longitude coordinates. This value represents an estimate of the accuracy of the horizontal coordinate measurements in the tile expressed in meters. A final data set of 13.1 ft, latitude, longitude, and altitude posting was distributed by EarthData.
Side-scan sonar is a specialized sonar system for mapping and detecting submerged objects. Side-scan sonar continuously transmits sound energy and analyzes the return signal, called the echo, that bounced off submerged objects and the lake bottom. The energy is transmitted directly under the sonar in the shape of a fan along the lake bottom on both sides of the sonar “towfish.” The strength of the echo primarily depends on the density, surface roughness, and angle of the reflecting surface. Objects that are dense, have a rough texture, and protrude from the bottom at a high angle have a strong echo that creates a bright image. Little or no echo behind the objects creates dark shadows. Most side-scan sonars produce clear images of objects on the lake bottom, however, they cannot provide depth information. The frequency of the side-scan sonar controls the spatial resolution of the returned image. Higher frequencies yield better resolution, but have less range. The spatial resolution or pixel size of the 600 kHz side-scan sonar is about 4 in. The 1,200 kHz side-scan sonar has a pixel size of about 1.6 in.
During the week of September 12, 2005, a survey of south-central Walker Lake was done using side-scan sonar to acquire images of anomalies which appeared to be mounds that were located by the single-beam echosounder (fig. 3). The survey was done by the USGS Center for Coastal and Watershed Studies. The main survey area covered about 5 mi2 along 96 mi of surveyed lines. A survey grid was laid out with a swath width of 246 ft and line spacing of 328 ft. The swath width was later reduced in the field to 162 ft. A few additional swaths were done at selected coordinates on the east and north-center of the lake where additional anomalies were located.
The Marines Sonic Technology, Ltd. (MSTL) 600 kHz side-scan sonar towfish was the primary sonar used and towed 16 to 39 ft behind a boat and about 33 ft above lake bottom. A MSTL 1200 kHz side-scan system also was used to acquire more detailed images in a few areas. The MSTL Sea Scan PC side-scan system controlled the survey acquisition. The Canadian Systems International Wide Area Augmentation System-enabled DGPS was used to locate position. By August 2005, a thermocline had developed at a depth of about 43 ft (U.S. Geological Survey, 2006). About 80 percent of the data were acquired when the towfish was below the thermocline and were useful.
Quality assurance included removal of data that showed systematic roll or vertical boat movement. Relative positions of the DGPS antenna and echosounder transducer were fixed during the survey period to assure accurate horizontal control. Depth of the transducer below water was measured with a tape, recorded in the field, and used in processing of the data. Raw data were processed in Hypack Max and exported in a latitude, longitude, and depth format for importing into a geographic information system (GIS).
Initial processing of the data included the manual removal of spurious data points (outliers) such as single-point depths located substantially above or below the general lake-bottom trend (fig. 4A and B), zero depths, or data that showed roll or vertical boat movement. Outliers could have resulted for various reasons such as submerged debris, gas bubbles in the water column, or sudden changes in sound velocity. Sound-velocity profile data were then used to correct the raw depth data, resulting in a correction of ≤4 in. Depth from the lake surface to the transducer was added to the sound-velocity-corrected depths to calculate depth below the water surface.
Depth data were recorded at a high frequency [10 hertz (Hz)] because it was unknown how much detail was needed to accurately define the lake bottom. Differentially-corrected positional data were recorded at a frequency of 1 hertz. As a result, about two to five depth measurements were recorded for every meter of survey line depending on actual vessel speed. A total of 889,797 discrete depth measurements were initially collected along more than 250 mi of transects and processed during this survey. This was much more data than was needed because most of the bottom of Walker Lake is smooth. To reduce the number of data points and to remove minor bottom-surface variations, an average depth was calculated every 328 ft along each transect. The averaged depth below the water surface was used in defining the bathymetric surface of the lake (fig. 4B). A total of 4,072 averaged depth points were calculated and used in the final data analysis.
Depth and location data were imported into ArcGIS® for additional processing. Average daily lake-surface altitude, measured to 0.01 ft, was calculated from lake-stage measurements made every 15 minutes at USGS station 10288500, Walker Lake near Hawthorne, Nev. Corrected depth-to-lake-bottom values were subtracted from the average daily lake-surface altitude for the day the original data were collected and rounded to the nearest 0.1 ft to determine lake-bottom altitude along the transects. The lake boundary of June 28, 2005, identified from high resolution imagery, was digitized about every 328 ft. Selected locations of near-shore transects also were digitized. The lake-bottom altitude, altitude of the lake boundary, and altitude of selected near-shore points were used in the ArcGIS® kriging module to interpolate lake-bottom altitudes between data points. The ordinary kriging method was used with a spherical semi-variogram model, variable search radius, and a minimum of twelve data points to interpolate values for 16-ft cells.
Lidar data were processed using NASA’s Airborne Lidar Processing System (ALPS). ALPS is written in Yorick and Ytk programming languages running in a Linux environment. Custom processing modules handle raw position and waveform data and organize the data in buffered 1.24 × 1.24 mi index tiles. Bare-earth elevation values were computed by determining the time of the last return within the individual small-footprint waveforms. The Random Consensus Filter algorithm was used to separate the vegetation clutter or anomalous elevations from the bare-earth returns (Nayegandhi and others, 2006). Data were then converted from units of time to points of latitude, longitude, and altitude.
A DEM was created using ArcGIS® software. The latitude, longitude, and altitude points were used to produce a 6.6-ft cell size raster surface. The inverse distance weighted algorithm was used to create the surface and required two points to determine the cell value. If 2 points were not within 16 ft of the cell, no datum value was determined. The resulting surface still had spurious cell values. A 5 × 5 cell median filter was then run over the surface. Each 1.24 × 1.24 mi tile was processed and then mosaicked together.
The bathymetry and DEM data were merged together to create a single map showing land-surface altitude contours delineating areas that are currently or were submerged by Walker Lake. The Lidar DEM primarily covered the valley floor and did not include measurements in the lake. The DEM from high resolution imagery covered the valley floor and mountains and incorporated a much larger area than the Lidar data. Paleohydrologic research suggests the lake-surface altitude of Walker Lake has not exceeded 4,120 ft throughout the Holocene (Adams, 2007). Contours were mapped up to 4,120 ft because of the uncertainty in the historic lake-surface altitude and to aid future paleoclimatic studies.
The original horizontal datum for the bathymetry data was WGS 84 and the vertical datum for lake-surface altitude was NGVD 29. The original horizontal and vertical datum for the Lidar and high resolution imagery DEMs was WGS 84. The horizontal datum was set to NAD 83 and vertical datum was set to NGVD 29 for all data sets. Using National Geodetic Survey software and comparison to known lake-surface elevations, the Lidar DEM was adjusted by 79.7 ft. Land-surface altitudes determined from Lidar were re-sampled every 16 ft from the 6.6-ft cell size. Problems were encountered with acquisition of Lidar data on May 26, 2005, that could not be reduced and consequently were not used. The flight on this day passed over the Wassuk Range on the western side of Walker Lake. An inverse distance weighting algorithm was used to create a surface of 16-ft cells from the high resolution imagery DEM, which was used where Lidar data was missing.
The three data sets of 16-ft cells then were merged together. Bathymetry data were used up to the lake boundary. Lidar data were used for the rest of the study area, except for the Wassuk Range. The high resolution imagery DEM was used to fill in remaining areas with no data. This produced a final DEM of 1,364 mi2.
Errors were found in the merged data mostly along the north, south, and eastern shores. Errors occur where Lidar data indicate land-surface altitudes that are less than the lake-surface altitude. These errors could be due to moist soils along the shore affecting the return of photons from the Lidar. Errors up to 16 ft were found in 3 percent of the area within 328 ft from the shore. Errors were corrected by calculating the mean of 10 pixels that have valid land-surface altitudes and are adjacent to the pixel with invalid data, then replacing the invalid data with the mean value. The altitude of the shoreline was raised by 0.328 ft to ensure the mean value was greater than the lake-surface altitude.
Surface area and storage volume were calculated at 0.5-ft intervals for lake-surface altitudes of 3,851.5 ft, the minimum averaged depth, to 4,120 ft using the 3-D surface-analysis software in ArcGIS® 9.0 (Environmental Systems Research Institute, Inc., 2004). For each interval, the raster grid was examined to determine the surface area and storage volume of each cell contained within the limits of that particular altitude. The sum of these cells is the surface area and storage volume at that lake-surface altitude (appendix A).
Estimating the error in surface area and storage volume will help evaluate the uncertainty in the water budget. Errors in surface-area and storage-volume were estimated by assuming the error in altitude of the merged data set is ±0.5 ft. Error was then calculated as the difference between two consecutive surface-area and storage-volume values in appendix A. Errors in surface area range from 60 to 660 acres; most errors are between 100 and 200 acres (fig. 5A). Percent error in surface area ranges from 0.1 to 54 percent and generally is less than 1 percent for lake-surface altitudes ≥3,880 ft. Errors in storage volume range from 50 to 44,000 acre-feet and increase with lake-surface altitude (fig. 5B). Percent error in storage-volume ranges from 0.3 to 75 percent and generally is less than 1 percent for lake-surface altitudes ≥3,930 ft. Error in surface area contributes little to the uncertainty in estimating evaporation from the lake. Percent errors are small, but a small error in storage volume on a percentage basis is a large volume of water. Estimating changes in storage volume may be the largest uncertainty in the water budget.
U.S.
Department of the Interior | U.S. Geological
Survey
Persistent URL: https://pubs.water.usgs.gov/sir20075012
Page Contact Information: Publications Team
Page Last Modified: Thursday, 01-Dec-2016 19:42:28 EST
