Survey vessel - Multibeam echo sounder data was collected aboard the Canadian Hydrographic Survey vessel Frederick G. Creed, an aluminum SWATH (Small Waterplane Area Twin Hull) ship that surveys at speeds up to 16 knots. The survey, in cooperation with the National Oceanic and Atmospheric Administration (NOAA), was conducted 15 hours a day in April, 1995 from Provincetown, Massachusetts. The ship has two submerged torpedo-shaped hulls that support the main deck via two thin struts. This design reduces wave motion on the ship and resistance to the ship's forward motion through the water, thus making it a very stable platform at high survey speeds. Stability of the ship is also enhanced by the computer-controlled action of four stabilizer fins located fore and aft on the inboard side of the hulls. The stabilizers control the pitch and roll of the vessel and allow adjustment of the heel and trim of the ship while under way. During the survey, the ship's position was determined with an accuracy of 10 m or better using a Magnavox 4200 geographic positioning system (GPS) receiver in conjunction with differential GPS corrections transmitted by U.S. Coast Guard radio beacons. At the start of each day, casts were made to record the sound velocity profile of the water column in the area to be surveyed. This information is used by the data acquisition system to correct for the refraction of the transmitted multibeam sound signal as it travels through the water to and from the sea floor. Velocity information was updated during the day whenever the survey moved from one area to another.

Multibeam data acquisition -The data presented here in the topographic and sidescan sonar backscatter images were collected utilizing a Simrad Subsea EM 1000 Multibeam Echo Sounder that is permanently installed in the hull of the Creed. The EM 1000 has 128 ceramic elements, operates at a frequency of 95 kHz, and will survey in water depths ranging from 5 to 1000 m. The raw data are logged on a Sun SPARCstation 10 using Simrad's Mermaid software and displayed on personal computers (PC's). When operating in water depths between 5 and 200 m, the EM 1000 produces 60 aimed beams spaced at intervals of 2.5 degrees, thus giving a total beam width of 150 degrees (75 degrees to each side of the vessel) which insonifies a strip of sea floor measuring in width approximately 7.5 times the water depth. Signal processing methods that employ a combination of amplitude and phase detection are used to determine (for the patch of sea bed under each formed beam) the distance between the hull transducer and the sea bed. Values related to the strength of the returning signal for each beam are recorded and used to construct a composite seabed backscatter (reflectivity) image across the entire strip of insonified sea floor. Sounding rates depend on water depth but vary from 2 to 4 per second in water depths less than 100 meters. Horizontal spatial resolution at these sounding rates is on the order of 10% of the water depth at 16 knots; however, vertical resolution is approximtely 1% or better in the same depth range. The disposal site region is located in water depths of 50 to 90 meters. Thus, the average horizontal resolution averages 5 to 9 m, and the vertical resolution 50 to 90 cm. Both the bathymetric and the sidescan-sonar backscatter data are displayed in real time on the Sun workstation using software designed and written by the Ocean Mapping Group, University of New Brunswick. This display allows data gaps to be identified during the survey and also gives a measure of the data quality. An Applied Analytics POS/MV motion sensor located in the hull near the EM 1000 transducer detects changes in pitch, roll, and heave of the vessel. The motion information is recorded concurrently with the acquired multibeam signal and both are logged in a single file on the Sun workstation and made available to other workstations for further processing.

Data and image processing - After the echo sounder data was logged onto the hard disk of the Sun workstation, a suite of processing software developed by the Ocean Mapping Group was used to correct for artifacts and errors that may have been introduced during data collection. This software also enhanced the corrected data by resolving beam pattern and aspect ratio distortions and by imposing a linear contrast stretch before it generated bathymetric and sidescan sonar image mosaicks in a Mercator projection. All data processing described here is initiated using onboard Silicon Graphics workstations as soon as each acquisition file is closed by the Simrad Mermaid workstation (usually at the end of each survey line). As tidal information is not available in real time, tidal corrections are merged into the sounding data files at the end of the day for subsequent grid file generation. Tidal information is essential for mapping the depth information over the course of the entire survey as all depths are referenced to mean lower low water. The processing and editing steps on board the ship are as follows:

a. Demultiplex, or unravel, the acquired Simrad signal to generate separate files containing navigation, depth soundings, sidescan sonar backscatter values, and sound velocity information.

b. Automatically reject bad data. For the multibeam soundings, reject data outside expected depth ranges (operator's decision based on nautical chart data); for navigation data, reject fixes with poor GPS statistics.

c. Edit the navigation data on-screen to remove undesirable points, including turns at the ends of survey lines.

d. Edit the multibeam soundings on-screen to remove individual anomalous soundings.

e. Merge tidal information and the corrected navigation back into the data files. Tidal information from the Boston tide gauge was obtained via modem from computers maintained by the Ocean and Lake Levels Division of NOAA in Silver Spring, Maryland. The tidal database is updated by NOAA approximately every six hours.

f. Initiate UNIX c-shell script files that map the bathymetric soundings from each processed data file onto a Mercator grid with node spacings and scale selected by the operator.

g. Concurrently, UNIX c-shell script files map the extracted sidescan sonar backscatter values onto a digital mosaic in the Mercator projection at a scale selected by the user.

h. Generate sun-illuminated bathymetric raster files using the mapped grid node information to depict the depth information in a shaded relief view.

A Mercator projection allows individual map areas to be joined edge to edge when creating a composite image. The sun-illuminated image shown here uses a sun elevation angle of 45 degrees from an azimuth of 315 degrees; and a vertical exaggeration of four times to emphasize seafloor features.

Data archival - Raw data logged by the Simrad Mermaid workstation is backed up each day on Exabyte 8 mm tapes and then transcribed to CD-ROM's after the end of the survey for more permanent and stable storage. The map products and processed data files are also saved onto 8 mm tapes; they were transferred to magneto-optical read-write media for easier access in the event post cruise processing was required.

Topographic contours - Contouring of processed bathymetric data was accomplished with Arc/Info geographic information system software (Environmental Systems Research Institution, version 7.03). Processed data was formatted using the Arc/Info "point generate" routine into a point file containing 3,251,355 records comprised of latitude, longitude, and water depth in a geographic projection. The point file was transformed to a Mercator projection with the longitude of the central meridian at 70^o 34' 43.5"W, the latitude of true scale at 42^o N, a false easting of 4000 m, and a false northing of 0 m. The "point grid" routine was used to create a grid from the point file and to assign depth values to individual grid cells. The cell size of the output grid was chosen carefully to minimize both data loss and a banding artifact that occurs when point spacing is rectangular instead of square. As point spacing is rectangular here, it was necessary to select an output cell dimension that was larger than the input point spacing. A final cell size of 5 m was chosen. Topographic contours at 5-meter intervals were generated using the "lattice contour" routine which converts a grid (or lattice) to contours of equal depth value. At this stage, the resulting contour lines contained artifacts in the form of high-frequency distortions that were expressed only in areas of very smooth sea floor. These artifacts are due to problems encountered during data acquisition at nadir (beneath the vessel's keel) and to refraction effects at the outermost edge of the beam pattern when survey lines are spaced too far apart. The high-frequency distortions of contour lines were smoothed by using a user-defined low frequency "focal median" filter routine on the grid created by "point grid". For each cell location in a grid, this filter finds the median cell value within a specified neighborhood of cells and assigns it to the corresponding cell location on the output grid. Square focal median filters varying in size from 5 x 5 cells, 11 x 11, 15 x 15, to 21 x 21 were tried and the last was selected as the best solution. The filtered grid was converted to contours with the "lattice contour" routine. Finally, the resulting contours were edited manually with "arcedit" to remove small artifiacts that were not removed by the focal median filter in several very smooth areas of the sea bed.

Back to Introduction and Table of Contents