Bathymetry

Detailed comparisons of repeated bathymetric surveys are commonly inconclusive because the magnitudes of potential errors are equal to or greater than the actual changes of the seafloor morphology. The development of coastal sediment budgets and models for sediment transport and shoreline change require bathymetric surveys with vertical resolution and accuracy of 5 cm or better. Horizontal resolution and accuracy need to be at least 10 cm to quantify bedforms and bars. Sleds are probably the most accurate, widely used system for nearshore surveys, but their contact with the bottom limits their speed, spatial resolution, and ability to operate in many situations. Boat-based echo sounder surveys can achieve a higher spatial resolution and can operate where sleds cannot, but waves, tides, and other water-level fluctuations as well as boat dynamics and variations in the speed of sound in water can greatly limit their accuracy. Problems related to a survey sled's contact with the bottom cannot be overcome; therefore, echo sounder surveys must be improved.

Figure 1. The high-accuracy, high-resolution bathymetric surveying system (HARBSS) is designed to overcome the confounding effects of changing vessel draft, waves, and tides on depth soundings and to eliminate the need for measuring and modeling water level for a particular survey. The system combines Global Positioning System (GPS) receivers, an electronic motion sensor, a digital-gyro compass, a digital-analog echo sounder, a conductivity-temperature-depth probe (CTD), a computer, and custom software. The GPS antenna, compass, and motion sensor are aligned with the echo sounder's transducer. Using a bias-free phase solution from the GPS data (X,Y,Z accuracy of better than 1 cm), attitude information from the motion sensor, and heading information from the compass, the position and aim of the transducer is determined for each sounding. The CTD provides data to calculate the speed of sound. Using the above data, the sounding depths and horizontal locations of sounding points are corrected in X, Y and Z with respect to an Earth-centered ellipsoid.

Figure 3. HARBSS instrument mast mounted on 6 m aluminum tri-hull boat used for test survey in a lake. Mast is in a raised position for launching of the boat. During the test survey, the transducer was about 0.5 m below the water line.

Figure 4. Data from a test survey at Lake Travis, Texas. The transect obliquely crosses a drowned tributary creek valley and was surveyed five times. (a) Map view of track lines of the boat survey using HARBSS and wading survey using an electronic total station. Map view is highly exaggerated in the across-transect direction. Track lines are actually the positions of the sounding points on the bottom. (b) Profile view of transects. All profiles are corrected using HARBSS. Spikes in one survey to the left of -170 m are likely caused by water column returns, presumed to be fish. (c) Expanded view. Five corrected HARBSS profiles and one uncorrected profiles (dashed line) are shown. The uncorrected profile, which consists of raw depth measurements recorded by the echo sounder, was adjusted vertically to overlay the corrected profiles. Also shown are points measured using a conventional electronic total station (ETS) and a rod person who waded into the lake.

	HARBSS vs. HARBSS (cm)			HARBSS vs. ETS (cm)
	HARBSS vs. HARBSS (cm)			Absolute difference	ETS. HARBSS
Horizontal separation between points	<= 10.0	<= 50.0	<= 100.0	<= 100.0	<= 100.0
Mean horizontal separation	6.6	37.6	82.5	87.7	87.7
Mean vertical difference	5.2	5.3	6.5	4.9	87.7
Standard deviation vertical difference	3.7	4.7	5.7	3.7	6.0
Number of crossover points	72	534	767	50	50

We conducted a crossover analysis of the multiple transect data to determine the system's repeatability ( see above table). For this analysis, we avoided the spikes in the data by only considering data less than 170 m from the datum stake (see fig. 4). We extracted pairs of points obtained during separate passes along the transect that were closer than 10 cm, horizontally. Seventy-two pairs of points had a mean vertical difference of 5.2 cm, and the standard deviation of the vertical difference was 3.7 cm. If we include points within 50 and 100 cm of each other, the vertical standard deviation increases indicating horizontal repeatability of better than 10 cm.

To estimate accuracy, we conducted a crossover analysis between the multiple transects and the conventional ETS survey. Subtracting the heights of the ETS crossover points from the HARBSS points yields a mean height error of -1.2 cm. This is a measure of the bias of the HARBSS survey relative to the ETS survey. The bias is probably caused by either a tilting of the ETS survey rod or a small error (0.4 %) in the speed of sound setting of the echo sounder. The vertical difference between the HARBSS and ETS crossover points have a mean of 4.9 cm and a standard deviation of 3.7 cm. If we assume that the ETS survey represents the 'true' profile, the accuracy obtained during this survey is the same as our precision (repeatability) determined above.

We conclude that HARBSS can provide soundings that are within 5.2 cm (mean error) of their true elevations. Horizontal accuracy is estimated to be within 10 cm. This accuracy can be achieved from a small, open boat that is rolling, pitching, heaving, or listing. Error analysis indicates that we may be able to decrease the error by one half with better synchronization and interpolation of the various data streams and better incorporation of speed of sound data.