Open-File Report 2011–1226
Bathymetry and Acoustic Backscatter—Elwha River Delta, Washington
The general processing work flow for converting raw bathymetric soundings to a digital elevation model (DEM), also called grids, is shown in fig. 4. Critical aspects of the processing procedure are discussed below.
Vessel Position and Attitude
The R/V Parke Snavely was equipped with a CodaOctopus F180 inertial motion unit (IMU) for the duration of the survey. The F180 was running F190 firmware and received real-time kinematic (RTK) corrections directly from the base station. The RTK GPS data (2-cm error ellipse) are combined with the inertial motion measurements directly within the F190 hardware so that high-precision position and attitude corrections were fed in real time to the sonar acquisition equipment. The WGS84 (G1150) Epoch 2010.1548 3-dimensional reference frame was used for all data aquisition.
Sound Velocity Measurements
Sound velocity measurements were collected continuously with an Applied Micro Systems Micro SV deployed on the transducer frame for real-time sound velocity adjustments at the transducer-water interface. The Micro SV is accurate to +/- 0.03 m/s. In addition, sound velocity profiles (SVP) were collected with an Applied Micro Systems, SvPlus 3472. This instrument provided time-of-flight sound-velocity measurements by using invar rods with a sound-velocity accuracy of ±0.06 m/s, pressure measured by a semiconductor bridge strain gauge to an accuracy of 0.15 percent (500 dbar, full scale) and temperature was measured by thermistor with an accuracy of 0.05 degrees Celsius (Applied Microsystems Ltd., 2005).
Sonar Sounding Processing
GPS data and measurements of vessel motion were combined in the F180 hardware to produce a high-precision vessel attitude packet. This packet was transmitted to the Swath Processor acquisition software in real time and combined with instantaneous sound velocity measurements at the transducer head before each ping. As many as 20 pings per second were transmitted, with each ping consisting of 2,048 samples per side (port and starboard). The returned samples were projected to the seafloor using a ray-tracing algorithm working with the previously measured sound-velocity profiles in SEA Swath Processor (version 3.05.18.04). A series of statistical filters were applied to the raw samples that isolated the seafloor returns from other spurious targets in the water column. Finally, the processed data is stored line-by-line in both raw (.sxr) and processed (.sxp) trackline files. Processed (.sxp) files were further processed with sxpegn (build 151) by David Finlayson (USGS) to remove erroneous data from the files and produce valid gain-normalized amplitude data for processing backscatter data.
Digital Elevation Model Production
The digital elevation model (DEM) produced in this work was derived solely from the bathymetric data collected by the USGS during field activity S-6-10-PS. CARIS HIPS and SIPS (version 7.0.2 Service Pack 2) bathymetry processing software was used to clean and bin the raw bathymetry. Processed .sxp files were imported to CARIS, and field sheets were created within CARIS to encompass the 4 survey areas: (1) the Elwha River Delta (primary area of interest), (2) the Rayonier Pier nearshore, (3) a section of Ediz Hook inside the Harbor, and (4) a section of Ediz Hook outside the harbor.
Survey lines were filtered to remove adjacent line data from nadir gaps. A CARIS Swath Angle BASE (bathymetry with associated statistical error) surface was created at 1-m resolution and the subset editor was used to clean artifacts and other unwanted soundings. The binned data were exported as an ASCII table along with calculations of the bin standard deviation (of all soundings within the 1 m x 1 m cell spacing) and the sounding density.
To convert the data from ITRF2000 (also known as WGS84 G1150) ellipsoid heights to orthometric heights on NAD83 (CORS96)/NAVD88, the CARIS gridded data was exported as an ASCII table. The data was then transformed from the ITRF2000 ellipsoid to the NAD83 (CORS96) ellipsoid using a 14-point Helmert transformation using the parameterization described in Soler and Snay (2004) with the command line tool CS2CS in the Proj4 library (http://trac.osgeo.org/proj/). The parameters were calculated for an Epoch date of 2010.1548 as shown in table 3. The conversion from WGS84 (G1150)/ITRF2000 to NAD83 (CORS96) shifts the data about 0.90 m to the east, 0.80 m to the south, and 0.30 m in elevation.
At the base station, the NAD83 ellipsoid height was -11.04 m and the orthometric height based on Geoid09 (National Geodetic Survey, 2009) was 9.16 m, representing a vertical offset of 20.2 m between the NAD83 ellipsoid heights and NAVD88 altitudes. This static offset was applied to the each of the datasets to convert the NAD83 (CORS96) ellipsoid elevations into NAVD88 altitudes. Finally, the ASCII data were gridded in Surfer (Version 10.3) at 1 m resolution using the IDW algorithm with a 3 m search radius and a smoothing parameter set to 0.25. This process filled small gaps in the surface and provided some minor smoothing to reduce the statistical noise inherent in interferometric bathymetry measurements. This resulting surface was converted to ESRI ASCII grid format (appendix describes the ESRI ASCII grid format).
If you have trouble reading these tables, please contact the WCMG team.
Backscatter Image Production
The raw 16-bit backscatter data recorded simultaneously with bathymetry by the SWATHplus was georeferenced and gain-normalized by the program SXPEGN, software written by the USGS to enhance the backscatter of the SWATHplus system. The program normalizes for time-varying signal loss and beam directivity differences. The resulting normalized amplitude values are rescaled to 16 bit, exported as point files and gridded into GeoJPEGS using GRID Processor Software supplied by SEA.
Estimates of Bathymetric Uncertainty
For relatively flat parts of the seafloor, the standard deviation of sounding elevations within a specified cell spacing is a good measure of the precision of the sonar measurements; this is not true in areas where the seafloor is naturally variable or steep; in these areas the standard deviation reflects natural variation rather than sonar measurement uncertainty. Each survey area discussed below includes two plots of the within-cell standard deviations. The first plot is a simple histogram of the standard deviations in units of meters. The second is a plan-view map of the standard deviations across the selected survey area. These absolute numbers are a good indication of the distribution of uncertainty in bathymetric soundings; however, bathymetric errors naturally increase with range from the system (that is with depth), so a measure of survey reliability that accounts for water depth as a factor is also important. The International Hydrographic Organization (IHO) standards described next provide such a metric using depth as a proxy for distance.
The IHO defines several survey orders based on a combination of coverage, depth, and accuracy (International Hydrographic Organization, 2008). The IHO defines maximum allowable total vertical uncertainty (TVU) using a depth-dependent formula with two additional variables as defined below (from International Hydrographic Organization, 2008):
This table is for formatting purposes only.
a represents that part of the uncertainty that does not vary with depth, b represents that part of the uncertainty that varies with depth, and d is the depth.
IHO standards require a 95-percent confidence level, defined as 1.96 x the standard deviation of sounding uncertainty. IHO survey orders use the following values as minimums for their survey orders (note that descriptions are generalized and order 1 and 1a are combined because they have the same values for minimum TVU):
If we assume that the elevation uncertainty in bathymetric soundings is normally distributed, we can estimate the total vertical uncertainty (TVU) in an elevation cell at the 95-percent level using:
This table is for formatting purposes only.
where σ(x,y) is the standard deviation of a cell located at coordinate (x, y). By comparing the calculated values from equation 2 against the standardized TVU thresholds represented by equation 1, we have an objective measure of the quality of the bathymetric surface.
Each survey area described in the following section has a plot with the calculated TVU values (equation 2) for a representative subsample of the survey as a function of water depth. Plots include lines indicating the IHO thresholds for special-order, first-order, and second-order surveys derived from equation 1.
For more information, contact the PCMSC team.