Bathymetric Data Collection and Postprocessing

Bathymetric data (water depths and positions) were acquired using sonar technology and were collected using a high-resolution MBMS mounted on a boat (figs. 3A–C). A 24-foot (ft) flat-bottom cabin boat (fig. 3B) was used for the river survey, and a smaller 16-ft jon boat (fig. 3C) was used for the quarry survey because it could be launched and retrieved from the bank of the quarry more easily. The various components of the MBMS used for this project are described in more detail in reports about studies on the Missouri and Mississippi Rivers in Missouri (for example, see Huizinga, 2010, 2022; Huizinga and others, 2010). The survey methods used to obtain the bathymetric data for the proposed harbor were similar to these river studies and recent lake studies in Missouri (Richards and others, 2019; Huizinga and others, 2022), as were the methods used to ensure data quality. A brief description of the equipment and methods follows.

An MBMS is an integration of several individual components: the MBES, an INS, and a data-collection and data-processing computer. The INS provides position in three-dimensional space and measures the heave, pitch, roll, and heading of the vessel (and, thereby, the MBES) to accurately position the data received by the MBES. HYPACK/HYSWEEP software (HYPACK, Inc., 2021) installed on the data-collection and processing computer was used to prepare for the bathymetric survey and to collect the survey data. The MBES used was the Norbit iWBMSh, operated at a frequency of 400 kilohertz (kHz; fig. 3A). The iWBMSh is similar in operation to the MBES systems used in other recent bathymetry studies in Missouri. Optimum data usually are collected in a swath of less than 160 degrees (80 degrees on each side of nadir, or straight down below the MBES); nevertheless, the swath can be electronically rotated to either side of nadir, enabling data acquisition along sloping banks up to a depth just below the water surface.

Most of the bathymetric survey data within the river and quarry were collected with the swath range limited to 140 degrees, 70 degrees on each side of nadir. Along the banks in the river and quarry, however, the swath range was widened to 160 degrees and was electronically tilted to port or starboard as needed to cover a wider swath of the bottom and enhance acquisition of bathymetric data in the shallow areas near the banks. The electronically tilted swath generally extended from about 5 degrees above horizontal on the bankward side of the survey vessel to 45–65 degrees past nadir below the vessel.

The bathymetric data were collected along transect lines oriented longitudinally in the Arkansas River, and along roughly parallel lines across the quarry. The transect lines were spaced to create about 10- to 25-percent overlap of the survey swaths to ideally achieve complete coverage of the river and quarry bottom and minimize sonic shadows. Data along the shoreline were collected by navigating the boat parallel to the shore while overlapping the data collected in the main body of the river or quarry.

Real-time kinematic corrections for the INS during the surveys came from cellular communication with the Arkansas Department of Transportation Global Navigation Satellite System (GNSS) real-time network for the real-time navigation and tide solution; however, all the navigation information from the bathymetric surveys was postprocessed using the POSPac Mobile Mapping Suite (MMS) software (Applanix Corporation, 2021) to mitigate any effects of degraded positional accuracy of the vessel from a poor real-time solution. POSPac MMS provides tools to identify and compensate for sensor and environmental errors and computes an optimally blended navigation solution from the GNSS and inertial measurement unit (commonly known as IMU; fig. 3A) raw data from the INS. The position solution was further enhanced by collecting static GNSS data with a GNSS base receiver set up over a reference mark near the survey area, the coordinates for which were determined using techniques detailed in Rydlund and Densmore (2012). The blended navigation solution (called a “smoothed best estimate of trajectory” [SBET] file) generated by postprocessing the navigation data was applied to the bathymetric survey data.

The speed of sound in the water typically varies with time, location, and depth. Although sound velocity data are collected at the MBES head using the sound velocity probe (fig. 3A) throughout the survey to mitigate these variations near the water surface, potential changes in the speed of sound with depth need to be measured to accurately determine the depths acquired by the MBES. Sound velocity profiles, therefore, were measured with an AML Oceanographic Base X2 sound velocity probe at various times and locations throughout the quarry to apply to the depth data during postprocessing. A sound velocity profile taken on the river confirmed a uniform sound velocity in the river consistent with the sound velocity measured at the MBES head, as expected in the well-mixed flow of the river.

After completing the surveys, the acquired depth data were processed further using HYPACK/HYSWEEP software to apply sound velocity profiles and to remove data spikes and other spurious points in the MBES swath trace often caused by fish and submerged woody debris and other vegetation. The data were georeferenced using the navigation and position solution data from the SBET file from POSPac MMS and visualized in HYPACK/HYSWEEP as a triangulated irregular network (commonly known as a TIN) surface or a point cloud. The georeferenced bathymetric datasets were filtered and reduced to a 1.64-ft (0.5-meter [m]) data resolution and provided in the ASPRS laser (LAS) format (American Society for Photogrammetry and Remote Sensing, 2019) with associated metadata files as part of the related USGS data release for this project (Huizinga and others, 2021). The LAS format is a standardized binary format for storing three-dimensional point cloud data and point attributes along with header information and variable length records specific to the data. Data points are stored as a three-dimensional data cloud as a series of x (longitude), y (latitude), and z (elevation) points. Additional processing details are provided in the metadata of the associated data release (Huizinga and others, 2021).

Bathymetric Data-Collection Quality Assurance

For the MBMS, the principal quality-assurance measures were assessed in real time during the surveys. The MBMS operator continuously assessed the quality of the data collected during the survey by making observations of across-track swaths (such as convex, concave, or skewed bed returns in flat, smooth bottoms), noting data quality flags and alarms from the MBES and the INS and inspecting comparisons between adjacent overlapping swaths. In addition to the real-time quality-assurance assessments during the survey, beam angle checks and a suite of patch tests were completed before and after the surveys during the field survey season in 2021 to ensure quality data were acquired from the MBMS. These tests generally were completed in the deeper part of a surveyed lake and over a submerged feature such as an old channel or a submerged roadway.

A beam angle check is used to determine the accuracy of the depth readings obtained by the outer beams (greater than 25 degrees from nadir) of the MBES (U.S. Army Corps of Engineers, 2013), which may change with time because of inaccurate sound velocities, physical configuration changes, and water depth. A beam angle check is completed at least once each year during the survey season. The most contemporaneous beam angle check was completed at Lake of the Ozarks near Osage Beach, Missouri, on November 18, 2021, and the results were within the recommended performance standards used by the USACE for hydrographic surveys for all the representative angles less than 70 degrees (U.S. Army Corps of Engineers, 2013), permitting the use of the central 140 degrees of the sonar swath with confidence. Points acquired outside of the central 100–110 degrees of the sonar swath generally had overlap with adjacent swaths, which increases the quality of the survey in the overlapped areas because of increased spatial point density.

Patch tests are a series of dynamic calibration tests that are used to check for subtle variations in the orientation and timing of the MBES with respect to the INS and real-world coordinates. The lever arm offsets between the MBES and the INS are measured and verified for accuracy before each project. The patch tests are used to determine timing offsets caused by latency between the MBES and the INS and angular offsets to roll, pitch, and yaw caused by the alignment of the transducer head (Huizinga, 2022). These offsets have been observed to be essentially constant for a given survey, barring an event that causes the mount to change such as striking a floating or submerged object (see Huizinga, 2022). The offsets determined in the patch test are applied when processing the data collected during a survey. The most contemporaneous patch tests were completed at Lake of the Ozarks near Osage Beach, Mo., on November 18, 2021, and angular offsets were updated in the data-collection software as appropriate. For this project, no timing offset was measured, which is consistent with latency test results for this boat and similar equipment configurations used in other surveys (Huizinga, 2010, 2022; Huizinga and others, 2010). The angular offsets were consistent with those determined for this instrument configuration from other patch tests during the 2021 survey season based on quality-assurance data collected by USGS in 2021 (Richard Huizinga, USGS, written commun., July 2022).

Similar to the previous studies of bathymetry in Missouri (Huizinga, 2010, 2022; Richards and others, 2019; Huizinga and others, 2022), uncertainty in the multibeam surveys was estimated for each survey-grid cell in the surveyed area using the Combined Uncertainty and Bathymetry Estimator (CUBE) method (Calder and Mayer, 2003) as implemented in the MBMax processing package of the HYPACK/HYSWEEP software (HYPACK, Inc., 2021). The CUBE uncertainty is a measure of the variability of the individual points in the cell used to determine the CUBE-derived elevation for the cell. Most of the CUBE uncertainty values for the bathymetric surveys of the river and quarry (more than 95 percent) were less than 0.10 ft (fig. 4), which is within the specifications for a “Special Order” survey, the most stringent survey standard of the International Hydrographic Organization (International Hydrographic Organization, 2020). The median uncertainty value of the data was about 0.03 ft. The largest uncertainty value in these surveys was about 1.6 ft; however, uncertainty values of this magnitude typically are near moderate-relief features (steep banks and submerged ridges; fig. 4). The uncertainty values also were sometimes larger in the outermost beam extents of the MBES swath in the overlap with an adjacent swath, particularly when the swath was tilted for the survey lines along the banks (fig. 4).

Topographic Data Collection and Postprocessing

Topographic data (land elevations and positions) acquisition used lidar technology, and the data were collected as LPC data using a UAS with a YellowScan Vx20–100 lidar payload (fig. 5), which consists of the lidar scanner and an INS. The LPC data were collected as the UAS followed two sets of parallel transect lines, oriented perpendicular to each other (nominally north to south and east to west) on separate flights. The LPC was corrected using an SBET from a postprocessed kinematic (PPK) solution, which used a Trimble R8s base station set up over the same reference mark as was used in the quarry multibeam survey, and 10 additional ground control points (GCPs) consisting of Propeller AeroPoint smart targets which were PPK corrected to a nearby continuously operated reference station (commonly known as CORS).

The LPC data were collected with the swath range limited to 90 degrees, 45 degrees on each side of nadir, to reduce spot size elongation ambiguity in the measurements. The flight plan lines were designed to create about 50-percent overlap of the survey swaths to ensure complete coverage of the survey area and minimize optical shadows.

The points in the LPC data were classified with respect to the type of point using automatic and manual point classification tools in the Global Mapper software (Blue Marble Geographics, 2022). Points representing the last, strong return signal from the lidar scanner generally were classified as “ground” (code 2) according to the ASPRS standard lidar point classes (American Society for Photogrammetry and Remote Sensing, 2019), except when the point was identifiable as one of the other classifications based on other points around it, such as reflections from the water surface of the river or quarry, a building or other human-made object, high wires, or vegetation. Classification of nonground points was based on the intuitive expectation of the three-dimensional form of each object: a building or other human-made object has regular, straight, or angular edges; a tree or shrub looks like one in the point cloud; and the water surface is a set of points on a horizontal line above the bathymetric survey. Points above the last, strong return signal from the scanner that were not easily identifiable as one of the other types were set to the “created, never classified” (code 0) classification (American Society for Photogrammetry and Remote Sensing, 2019).

After classification of the points, the LPC data were colorized from a red-green-blue (RGB) orthoimage of the survey area collected the second survey day using a Ricoh GR camera mounted on the UAS. The georeferenced, colorized, and classified LPC data were output at maximum resolution, and the dataset is provided in the ASPRS LAS format (American Society for Photogrammetry and Remote Sensing, 2019) with associated metadata files as part of the related USGS data release for this project (Huizinga and others, 2021). Additional processing details are provided in the metadata of the associated data release (Huizinga and others, 2021).

Topographic Data-Collection Quality Assurance

Annual boresight calibration is completed on the UAS payload (similar to a patch test for the MBMS), which is a dynamic calibration test that is used to check for subtle variations in the orientation of the lidar scanner with respect to the INS and real-world coordinates. The boresight calibrations are used to estimate the angular offsets in the roll, pitch, and yaw between the lidar scanner and the INS. Furthermore, the lever arm offsets between the lidar scanner and the INS are measured and verified for accuracy before each project. The lever arms and angular offsets determined in the boresight calibration are applied during postprocessing. The lever arm and angular offset measurements were consistent with those determined for this instrument configuration from other tests during the 2021 survey season (Lance Brady, USGS National Uncrewed Systems Office, written commun., September 2021).

The horizontal accuracy assessment of the LPC included a comparison of the lidar data with the RGB orthoimage collected the second survey day. The orthoimage used Structure from Motion (commonly known as SfM) technology to create a georectified image that was corrected to 10 Propeller AeroPoint smart target GCPs with PPK-corrected locations with a Trimble R8s real-time kinematic GNSS base station. The RGB orthoimage is a second, independent dataset that was visually compared with the LPC. Visual comparison revealed generally good agreement with the locations of objects readily observable in each dataset, particularly the GCPs. The vertical accuracy assessment was completed using validation GCPs in nonvegetated areas. The average root mean square error for the nonvegetated GCPs was 0.10 ft (0.03 m). Grid spacing for this assessment was set to five times the point spacing from the GCP using a maximum number of 10 points and inverse distance weighting (Huizinga and others, 2021).