Timothy J. Leonard and David J. Coughlan
Abstract
The effectiveness of a method for mapping reservoir bottom surfaces with a combination of hydroacoustic equipment, global positioning system (GPS), and geographic information system (GIS) was investigated. A reservoir was mapped with this method and compared to bottom surface information collected in 1920. Apparent location and the thickness of sediments on the lake bottom were easily detectable. The method was also used in mapping a pond that had been mapped a year earlier with traditional surveying techniques. The difference of pond volume estimates calculated by both methods varied from 3 to 6% in multiple trials. Problems encountered were: the recording of multiple depths at the same point, inability to sample shallow areas, unavailability of high-resolution shorelines, and legal restrictions associated with professional surveying rules and regulations.
Introduction
Duke Power Company owns and manages eleven reservoirs on the Catawba River in North Carolina and South Carolina. Impounded between 1904 and 1963, they vary greatly in surface area ranging from 450 acres to 32,510 acres. Situated in a rapidly growing region, these reservoirs are managed for hydroelectric power, power plant cooling water, flood control, public drinking water, and recreation. These management goals require a thorough knowledge of the reservoirs' characteristics including their volumes. To date, the only such information available are surveys done before impoundment. However, sedimentation caused by development, poor agricultural practices, forestry, and other land uses has altered reservoir volumes.
Proper management of the lakes requires that current lake volumes and sedimentation rates be determined. Because of the lakes' sizes however, traditional surveying techniques would be too laborious and costly to be practical. With this in mind, scientists in Duke Power's environmental division proposed an alternate method of mapping the lake bottoms by using a hydroacoustics system with a GPS to collect depth measurements and locations and processing the data with GIS. In this way, bottom maps and volume estimates could be produced in a cost-effective manner. A proposal was made to Duke Power's research and development branch for funding and a project was initiated to determine the effectiveness of this method.
Objectives
The two main objectives of this study were to:
Methods and Materials
Surveys
Two water bodies were surveyed in this project. Lake Rhodhiss, a 3,515-acre Catawba River reservoir, was surveyed once in a series of trips during the period from October 1995 to January 1996. The 34-acre standby nuclear service water pond at Duke Power's McGuire Nuclear Station (MNS) was surveyed three times during June 1996.
Lake Rhodhiss was chosen because depth data from 1920 was available from the United States Geological Survey in an ArcInfo coverage. This older data would let us demonstrate the ability to determine sediment locations and thickness.
The service water pond was mapped to compare the study methods to traditional survey techniques used in 1994 on the same pond. It was completely sampled on three separate dates during June 1996 in order to evaluate the repeatability of our results.
Echosounder
A BioSonics, Inc. DT4000 digital transducer and echosounder were employed. The 200kHz 6° single-beam transducer, cable, and surface unit were calibrated to United States Navy Standards. A laptop computer running BioSonics, Inc. software controlled the equipment. A 36-mm diameter calibration sphere was lowered to known depths below the water surface to determine the accuracy of the echosounder at measuring depth.
Differential GPS
An OMNISTAR 6300A Receiver incorporating a Motorola 6-channel Oncore GPS receiver was used to obtain real-time differential GPS data with an accuracy of two meters or less. The longitude and latitude of each sample point was recorded with its depth.
Field Deployment
The digital transducer was mounted on an adjustable vertical support bracket attached to the side a boat. The bracket and transducer were lowered into the water and the depth of the transducer face below the water surface was measured and subsequently added to all depth measurements. The GPS and OMNISTAR antennas were mounted on poles above the level of all other objects in the boat. The OMNISTAR output was connected via serial cable to the laptop computer controlling the digital transducer. Sampling was performed along paths that were located at the judgment of the field technician conducting the survey. Shallow areas were surveyed as completely as possible without risking damage to the submerged transducer. Surveys were conducted on calm days of slight to no wind in order to minimize pitch and roll of the boat. Depths were recorded as distance below the transducer to the lake bottom.
Data Preparation
Upon return to the laboratory, data files were plotted and examined visually prior to analysis. The data were analyzed and distance to the bottom determined for each ping. Geographic coordinates of data were converted into decimal degrees with a precision of six decimal places. Depths were corrected for the submergence of the transducer and for the elevation of the lake surface below normal full pond. Text files containing a sequential identification number, longitude, latitude, and depth for each ping were prepared for transfer to the GIS.
GIS Processing - Lake Rhodhiss
The text files containing survey data were used to create an ArcInfo point coverage using the GENERATE command. Depths for each point were added as an attribute to the coverage. The point coverage was converted to North Carolina stateplane coordinates using feet as the mapping unit. Because of errors, a relatively small number of points were located outside of the lake shoreline. An existing coverage of the shoreline was used to delete these points from the coverage. The final coverage contained 29,101 points and covered 2,000 acres of the lake surface. Elevation in feet above mean sea level was calculated for each sample point from the point's depth.
The point coverage and a coverage of the full-pond shoreline were used to create a triangulated irregular network (tin) surface model of the lake floor. Flat triangles in the tin were identified and eliminated by one of two methods. If the flat triangle was caused by three sampled points with the same elevation, one of the points was chosen and deleted. If the flat triangle was caused by connecting three points from the shoreline coverage, as often happened in the ends of narrow shallow coves that could not be reached by boat, additional points were added along the center of the cove with estimated elevation values. Although not tested, the addition of these points was not thought to have a significant effect on volume calculations.
The ArcInfo coverage of 1920 bottom contours of the lake was obtained and edited in order to overlap it with the 1995 data. As the 1920 map had no reference points that could be reliably located by present landmarks other than the dam, the contours were rubber-sheeted using the historic and present shorelines to create a best fit within the current lakeshore. These 1920 contours and shoreline were used to create another tin.
The volumes of the 1920 tin and the 1995 tin surfaces were calculated using the ArcInfo VOLUME command with a z_factor of -1. Without the negative z_factor, ArcInfo would have calculated a volume below the lake bottom to the base value instead of within the reservoir basin. TINCONTOUR was used to create contour lines at 10-foot intervals for comparison with the 1920 contours as a check of the similarities of the data.
Both tins were converted to overlapping ArcInfo grids with equal resolution (Figure 1.). Identically located cells that contained real values in only one grid were converted to NODATA values. A depth of sediment grid was created by subtracting the grid of 1920 data from the grid of 1995 data (Figure 2.). Positive values indicated depths and locations of sediments. Negative values indicated areas of possible scouring of the lake bottom or errors.
GIS Processing - McGuire Standby Nuclear Service Water Pond
The files from each sampling date contained the locations and depth for each ping and were imported into a database table. Many of the records contained identical longitude and latitude coordinates but often different depths. This was the result of collecting data at locations that were separated by distances less than the recorded precision of the GPS data. The data were summarized to determine the average depth and the range of depths recorded at each coordinate location. Locations with a depth range greater than 0.1 m were deleted (Table 1.). The summarized data were used to create three ArcInfo point coverages. As with the Lake Rhodhiss data, some points were located outside of the pond because of errors and were deleted from the coverages. The average depth was used to calculate the elevation in feet above mean sea level for each point.
The three point coverages were used to create three separate tin models of the pond floor. A shoreline coverage was created from engineering drawings and was used in the tin creation to establish the full-pond elevation. The volume of each of the three tins was calculated in the same manner as the reservoir.
Results
The results of this study were very encouraging. Although the horizontal position of the 1920 data could not be considered very accurate, comparisons of the 1920 and 1995 Lake Rhodhiss data demonstrated the study method's ability to locate and quantify sedimentation in reservoirs. However, more investigation is needed to determine how accurately sedimentation can be measured.
Actual time spent collecting hydroacoustics data in the field was quite short and required little manpower. Only one technician was required to perform the field sampling. The collection of over 32,000 data points extending over 2,000 acres of Lake Rhodhiss took only 19.6 hours of actual sampling time. The longest sampling time on the service water pond was three hours when 10,322 points were collected. This is in marked contrast to the traditional survey of the pond which took a crew of six people forty hours to collect less than 600 sample points.
Data analysis also took less time with the project method than with traditional surveying. Analyzing and converting acoustic data and GIS processing, analysis, and map production for the pond for one sampling trip was performed in about six hours. Traditional surveying methods required approximately 40 hours to process the field data from the pond. Overall, the labor required to survey the pond with the study methods took 274 hours less than traditional survey techniques, a 98% savings.
Pond volume estimates calculated with the project methodology compared favorably with those from the earlier survey with differences of only 6.2 and 3.9 percent (Table 2.). Such differences are quite possible less than the margin of error of either technique.
Discussion
This study indicated that the combination of hydroacoustics, GPS, and GIS are capable of producing bottom maps comparable in accuracy and quality to traditional surveying with a fraction of the labor. The advantages of this labor savings are two-fold. First, the low cost makes it practical to map lakes that would never be mapped due to the expense of traditional surveys. Second, lake surveys can be afforded on a regular basis to evaluate sedimentation rates.
However, certain issues and problems were recognized during this study. Sampling needs to be performed when the reservoir is as close to full pond elevation as possible. If not, much of the shallow areas will be missed. Sampling must be performed on calm days to minimize pitch and roll of the boat which would cause erroneous depth measurements. Real-time display of the sampling track would be helpful in ensuring adequate and efficient sampling and eliminating sampling overlap. Suitable shoreline data must still be obtained from other sources and may be lacking in some instances. Sedimentation investigations will be hampered if the baseline data is of poor quality. State laws governing surveying practices may limit the application of these methods. According to the statutes in North Carolina, only registered land surveyors can perform this type of work. While there is an exclusion for public utilities performing work for in-house use which allows non-registered Duke Power personnel to survey company owned lakes, other entities might still require surveyors to perform bathymetry.
Still, even in light of these issues, the results of this study were promising enough that funding has been approved for bathymetric mapping of the remaining ten Catawba River reservoirs.
Timothy J. Leonard
Duke Power Company
13339 Hagers Ferry Rd
Huntersville, NC 28078
(704) 875-5247
(704) 875-5038 fax
tjleonar@dpcmail.dukepower.com
David J. Coughlan
Duke Power Company
13339 Hagers Ferry Rd
Huntersville, NC 28078
(704) 875-5236
(704) 875-5038 fax
djcoughl@dpcmail.dukepower.com