Bathymetry and Near-River Topography

The GPS data were edited using the HyPak 2008® single-beam editor resulting in 95,654 bathymetric points surveyed by boat with the echo sounder and 1,069 bathymetric and streambank points surveyed by walking. The combined surveys covered 4.1 km on the Yakima River and 0.6 km on the Naches River. The distribution of the walking-survey data points is shown in figures 6 and 7 and the distribution of the boat-survey data points is shown in figures 10 and 11.

River survey data are available on the Gap to Gap River Surveys project web page at (http://wa.water.usgs.gov/projects/yakimagap/). The data are available in an Excel spreadsheet and in ASCII text files.

Survey Accuracy

Accuracy of the surveyed points is a function of the accuracy of the control points, of the RTK GPS values relative to the control points, and of the echo soundings. These accuracies were evaluated from manufacturer’s specifications, check measurements on control points, and a unique validation test that compared repeat values of elevation from nearby points.

The horizontal and vertical positions for the river surveys were derived from RTK GPS locations relative to survey control stations set along a baseline (7,842 m) between fixed monuments—one near Yakima (designation GP39082-23) and two in Yakima (designations SARG and WW_Plant) (table 2). Survey checks on the SARG monument, a National Geodetic Survey monument with a B-order of horizontal accuracy and a second-order vertical accuracy, compared within 0.014 m in the horizontal dimension and within 0.038 m in the vertical dimension. No network adjustment was made to the coordinates of the control points.

The root mean square accuracy reported by Trimble for the vertical dimension for RTK GPS is ±2 cm plus 1 part per million (ppm) × baseline length. The longest distance between a survey point and a control point was about 5,600 m, or a baseline of 11,200 m. Use of the manufacturer’s guidelines would result in a root mean square (RMS) error of 0.0312 m for this study for the longest baseline or a 2-sigma error of 0.0624 m. The manufacturer’s reported RMS horizontal accuracy for RTK GPS is ±1 cm plus 1 ppm × baseline length (Trimble, 2006), which would equate to an RMS error of 0.0212 m or a 2-sigma error of 0.0424 m. The GPS rover data were checked against established control points to verify their accuracy. These checks showed differences ranging from 0.014 to 0.047 m in the horizontal and 0.036 to 0.078 m in the vertical dimension.

The horizontal and vertical error for each GPS data point was computed and reported by the Trimble® controller and stored with the positional information. Errors for the walking survey data represent the total error relative to the base station. The horizontal error ranged from 0.002 to 0.0413 m and averaged 0.007 m (RMS error = 0.015 m) relative to the base station. The vertical error ranged from 0.003 to 0.686 m and averaged 0.010 m (RMS error = 0.024 m).

The boat survey data included errors associated with the echo-sounder depths and errors associated with roll and pitch of the boat as well as errors with the GPS readings. A test was developed to estimate the total errors in the boat survey data. The test compared reported elevations for identical or nearby points that were surveyed twice at different times. The points that were surveyed twice included those located where survey lines crossed and those that were re-surveyed because the boat was still or moving slowly. The test compared all points within a distance of 0.2 m of each other, except for the first two points surveyed after a particular test point (table 3). These two sequentially surveyed points were not used to provide greater emphasis on comparisons of overlapping points from different survey tracklines. Many of the comparison points in the test included points on the same trackline that were re-surveyed from the slow-moving boat or from criss-crossing that occurred on the same trackline. The overlay footprint used to select points for comparison (radius of 0.2 m) was selected to minimize the topographic change in the channel bottom while remaining large enough to include a large sample of points. The reported errors from this test include the actual elevation differences in the streambed within the 0.2-m footprints.

Comparable riverine survey studies have reported similar topographic and bathymetric error ranges in the data, but several reports indicated a need for a high density of topographic data. Lane and others (1994) indicated that a minimum spacing between cross sections of less than 2 m is needed for volumetric (cut and fill) bed form estimates to be within 20 percent of the correct value. Marks and Bates (2000) determined that “flood hydraulics are directly affected by quite small changes in topography.” They compared a grid derived from cross-section surveys (low density of topographic/bathymetric points) with a LIDAR-derived grid (high density of points) and determined significant differences in the amount of simulated flood inundation area from a two dimensional hydraulic model. Hilldale and Raff (2007) analyzed the accuracy of using LIDAR bathymetry technology to provide the bathymetry information for a two-dimensional hydraulic model with the Yakima River as one of their test sites. They remarked that total station or RTK GPS surveys likely provide the best quality in shallow, slow water but are limited in deep or swift water and long reaches. They also remarked that single-beam echo sounding with RTK GPS (as used in this project) is common in long reaches of small rivers where multi-beam sounding is impractical. However, obtaining “high density, complete coverage” is difficult. Hilldale and Raff (2007) concluded that the precision of the LIDAR bathymetry elevations may be less than traditional methods but the density of data points gives it a “distinct advantage.” Barton and others (2004) used methods similar to those used in this project on the Kootenai River and reported that roving GPS checks on control points were “almost always” within 0.03 ft of the vertical coordinate and the echo sounder depths were within 0.1 ft during daily tests—similar to the results in this project. Eshleman and others (2006) used a differential GPS system mounted over a single-beam echo sounder on a personal watercraft in Capitol Lake, Washington, and reported subdecimeter accuracy but had problems accounting for salinity and aquatic growth.