Holtschlag, David J.and John A. Koschik, U.S. Army Corps of Engineers, 2003, An Acoustic Doppler Current Profiler Survey of Flow Velocities in Detroit River, a Connecting Channel of the Great Lakes, Date Posted: August 12, 2003, US Geological Survey Open-File Report 03-219 [http://mi.water.usgs.gov/pubs/OF/OF03-219/index.php]

An Acoustic Doppler Current Profiler Survey of Flow Velocities in Detroit River, a Connecting Channel of the Great Lakes

Prepared in cooperation with the
MICHIGAN DEPARTMENT OF ENVIRONMENTAL QUALITY, DETROIT WATER AND SEWERAGE DEPARTMENT, and the AMERICAN WATER WORKS ASSOCIATION RESEARCH FOUNDATION

ABSTRACT

Acoustic Doppler current profilers (ADCP) were used to survey flow velocities in Detroit River from July 8-19, 2002, as part of a study to assess the susceptibility of public water intakes to contaminants on the St. Clair-Detroit River Waterway. More than 3.5 million point velocities were measured at 130 cross sections. Cross sections were generally spaced about 1,800 ft apart along the river from the head of Detroit River at the outlet of Lake St. Clair to the mouth of Detroit River on Lake Erie. Two transects were surveyed at each cross section, one in each direction across the river. Along each transect, velocity profiles were generally obtained 0.8-2.2 ft apart. At each velocity profile, average water velocity data were obtained at 1.64 ft intervals of depth. The raw position and velocity data from the ADCP field survey were adjusted for local magnetic anomalies using global positioning system (GPS) measurements at the end points of the transects. The adjusted velocity and ancillary data can be retrieved though the internet and extracted to column-oriented data files.

The Michigan Department of Environmental Quality (MDEQ) Source Water Assessment Program (SWAP), the Detroit Water and Sewerage Department (DWSD), and the American Water Works Association Research Foundation (AwwaRF) in cooperation with the U.S. Geological Survey (USGS), are assessing the vulnerability of public water intakes on the St. Clair-Detroit River Waterway to contamination. Intakes on this waterway provide a water supply to about six million residents of Michigan and Ontario. The assessments will identify likely sources of water to public intakes in Michigan and will provide a basis for preparing emergency responses to contaminant spills. This survey is similar to an earlier ADCP survey of St. Clair River (Holtschlag and Koschik, 2003), which is accessible at: http://mi.water.usgs.gov/pubs/OF/OF03-119/.

The primary purpose of this ADCP survey is to provide data to characterize local flow velocities within Detroit River. The velocity data described in this report will be used to improve the calibration of a hydrodynamic model of the waterway (Holtschlag and Koschik, 2001). Water velocities were measured at 130 cross sections from the head of Detroit River at the outlet of Lake St. Clair to the mouth of Detroit River on Lake Erie by USACE and USGS crews from July 8-19, 2002. This report documents the survey, provides necessary adjustments to the velocity and positional data to correct for local magnetic anomalies, reformats the data for analysis, and provides access to the velocity and ancillary data through the internet.

St. Clair River, Lake St. Clair, and Detroit River form a waterway that is part of the boundary between the United States and Canada (Figure 1). The waterway is a major navigational and recreational resource of the Great Lakes region that connects Lake Huron with Lake Erie. In the southern part of this waterway, Detroit River flows from the outlet in the southwestern part of Lake St. Clair to Lake Erie, a distance of about 32 miles. Throughout its length, water-surface elevations in Detroit River decrease about 3 ft as it discharges an average of 186,000 ft³/s from a drainage area of about 228,800 mi².

This report was developed by the U.S. Geological Survey (USGS) and the U.S. Army Corps of Engineers (USACE) in cooperation with Michigan Department of Environmental Quality, Source Water Assessment Program, directed by Bradley B. Brogren; the Detroit Water and Sewerage Department's Water Quality Division, managed by Pamela Turner; and the American Water Works Association Research Foundation's Christopher Rayburn, who directs the Research Management Department. Two crews were needed to survey Detroit River. Paul Taylor and Benjamin Harrison of the Detroit District operated the USACE boat and ADCP unit. Don James of the Grayling, Michigan Field Office operated the USGS boat. Marie Reynolds, of the USGS Michigan District, created the web pages that form this report. Andreanne Simard, of the USGS Michigan District reviewed the report.

Acoustic Doppler current profiler (ADCP) measurements of flow (discharge of water) in rivers began in 1982 (Simpson, 2001). An ADCP uses sound waves to measure water velocity. In this survey, the ADCP transmitted sound waves at an ultrasonic¹ frequency of 600 kilohertz (KHz), where one kilohertz is one thousand waves or cycles per second. Water velocity is computed by use of the Doppler² principle, which relates the observed change (shift) in the acoustic frequency to the relative velocities between the source and the observer. Mathematically, this principle can be written as

is the relative velocity between the source and the observer, in feet per second, and

An ADCP measures water velocities by transmitting acoustic water pulses and bottom pulses (pings) as it moves across the river in a boat (Figure 2). The water pulse reflects off small particles of sediment and other material in the water, and returns an echo to a transducer on the ADCP. The shift between measured and transmitted frequencies is used to compute the relative water velocity between the particles and the ADCP. To determine water velocity components in three dimensions and error estimates, four transducers are arrayed on an ADCP at 90 degrees of horizontal separation and at a 20-degree offset from a downward orientation into the water (Figure 3). This configuration provides the data necessary to resolve the northing, easting, and vertical components of local flow velocities relative to the ADCP.

After transmitting the acoustic water pulse, the reflected pulse is subdivided into a number of discrete time intervals to describe the variation of local flow velocities with depth below the water surface. The number of time intervals is based on the expected maximum channel depth. Frequency shifts and corresponding velocity components are computed for each time interval. The collection of flow velocities for the set of time intervals from the returning signal is referred to as an ensemble. The ensemble describes the velocity profile at discrete intervals of depth (depth cells) in the water column. Depth cells (bins) are described by their heights and by the centroid of their depths below the water surface. In this survey, all bin heights were 1.64 ft (0.5 meters).

Velocity measurements near the tops of the uppermost bins in the water column were below the water surface by a depth that included the transducer draft and the blanking distance (Figure 4). The transducer drafts, the depths of the transducers below the water surface, for the USACE and the USGS survey vessels were 0.72 ft and 0.60 ft, respectively. The blanking distance allows time for oscillations in the transducer head that follow each ping to die out before the transducer can detect reflected pulses. The length of the die out interval and the speed of the acoustic signal in water determine the blanking distance. The blanking distance used in this survey was 0.82 ft (0.25 meters).

Velocity measurements near the channel bottom are limited by side lobe interference. Side lobes are secondary acoustic signals that are emitted 30-40 degrees off the main acoustic beam (Simpson, 2001). Reflections of the side lobe signals off the channel bottom interfere with the detection of the returning main acoustic signal. Velocity measurements near the shoreline were limited at some cross sections by shallow depths or navigational hazards, such as sunken pilings and littoral vegetation.

Although the primary purpose of this survey was to obtain point velocity data, flows (water discharges) also were computed to help compare measurements made by the USACE and USGS, and to enable comparisons of transect measurements made at the same cross section. In channel areas of measured velocity (Figure 4), measured flows were computed using the measured velocity and corresponding flow-area data. In addition, total flows were computed by combining measured flows with estimated velocities and flow areas in marginal cross section areas. Simpson (2001) describes methods for estimating flow velocities and areas used in ADCP measurements. In some areas, total flows are highly uncertain because aquatic growth limited the shoreline approach and long distances from the boat to the shoreline were estimated visually. In addition to the water pulse, a bottom-track pulse was used to measure the position and velocity of the boat with respect to the channel bottom. The bottom track (BT) is combined with the water velocities computed relative to the ADCP to determine absolute water velocities. Bottom tracking provides reference distances for computing flow areas and flow directions, and is essential when precision GPS measurements are unavailable. Bottom tracking, however, is reliant on a magnetic compass for directional information, which can be degraded by local magnetic anomalies. The accuracy of positions indicated by the bottom track can be degraded by a moving channel bed and by the pitch and roll of the ADCP during deployment. GPS measurements at the end points of all transects were used to reference bottom track measurements to Michigan SPCS 83 coordinates and to adjust the direction and scale the lengths measured using bottom tracking.

The ADCP data for Detroit River was obtained from July 8-19, 2002. The USACE crew surveyed 69 cross sections in upper Detroit River (upstream from Fighting Island) and Trenton Channel from July 8-11, 2002. The USGS crew surveyed 61 cross sections in lower Detroit River from July 15-19, 2002. The survey of lower Detroit River included Livingston Channel, Amherstburg Channel, and channels around Fighting Island, and Grosse Ile (Figure 5). Both USACE and USGS crews surveyed six cross sections in upper Detroit River, upstream from the Ambassador Bridge, to provide a basis for comparing possible differences between equipment and crews.

Flows in Detroit River during the ADCP survey were estimated by use of stage-fall-discharge relations (Koschik, USACE, written communi., April 11, 2003) and hourly water-level data on Detroit River from NOAA gaging stations (Figure 6). (Water level data from the NOAA gaging stations were retrieved from the internet at: http://co-ops.nos.noaa.gov/coastline.shtml? region=mi on April 24, 2003). The stage-fall-discharge relations are commonly used with average daily water level data to estimated daily and monthly flows. These relations do not account, however, for the dynamic changes in flow, which vary along the length of the river. The stage-fall-discharge relations described below approximate steady-state flow conditions. Detroit River flows were estimated using the average flow computed by use of equations (2) and (3):

Where, "Wyandotte", "Windmill Point", and "Gibraltar" in equations (2) and (3) refer to hourly water-level data, in meters, at the corresponding NOAA gauging stations. Flows, computed in cubic meters per second by use of equations (2) and (3), were converted to cubic feet per second for plotting purposes. Wind conditions on Detroit River, as indicated by NOAA weather station LSCM4 on Lake St. Clair, are shown in Figure 7. (Wind data were retrieved from the internet at: http://www.ndbc.noaa.gov/station_page.phtml?station=lscm4 on April 24, 2003).

ADCPs have a mechanism to adjust for the effects of minor pitch and roll. Horizontal water velocities, which are computed as the cosine of pitch and roll angles, are insensitive to angles of less than 5 degrees (Simpson, 2001). In contrast, vertical velocities are highly sensitive to small amounts of pitch and roll (Simpson, 2001).

The boat operated by the USACE carried a precision GPS (Global Positioning System) receiver that was integrated with the ADCP unit. This integration provided precise (sub-meter) geographic system coordinates (latitudes and longitudes) for each velocity ensemble. The boat operated by the USGS did not have an integrated GPS unit, but relied on a Garmin model GPS 76 handheld receiver for geographic coordinates measured at the first and last ensemble on each transect. The GPS data were combined with the bottom track data obtained by the ADCP to determine positions of the velocity profiles. GPS 76 receivers typically (95 percent of the time) provide geographic coordinates that are within 49 ft of their true position (Garmin Corporation, 2001). When a GPS 76 receiver is Wide Area Augmentation System (WAAS) enabled, picking up satellite and land-based differential corrections, the geographic coordinates are typically within 10 ft of their true positions (Garmin Corporation, 2001). The GPS 76 receiver used by the USGS during the Detroit River survey was WAAS enabled during most of the survey. The accuracy of the GPS receivers, however, was not field verified.

The ADCP survey included two transects of each of the 130 cross sections surveyed. Each transect traversed the river at approximately right angles to the average flow direction. The first traverse (transect) generally started from a point near the left bank and proceeded to a point near the right bank, in which left and right are referenced to looking downstream. The left bank was generally nearer the Canadian shoreline than the shoreline of the United States. A second traverse of the same cross section generally followed immediately after the first and was usually started from a point near right bank. Shallow water depths, aquatic organisms, or navigational hazards, such as sunken pilings, sometimes limited approaches to the riverbanks.

In this report, cross-section identifiers started with a two-letter code that helped identify the branch of Detroit River that was surveyed (Table 1). The next two numbers in the identifier were a sequence that increased in the downstream direction. In some cases, an "E" or "W" was appended after the sequence number to indicate whether the branch was east or west of a bifurcation formed by an island or dike. Alternatively, an "A" was sometimes appended after the sequence number if two cross sections were surveyed in the same general area. For transects that were surveyed by both the USACE and the USGS, ".USACE" or ".USGS" was appended to the cross section identifier. Individual transects of the cross section are identified by appending a "c" or a "u" to the cross section identifier to indicate whether the transect started near the Canadian shoreline (surveyed from left to right), or the United States shoreline (surveyed right to left), respectively.

USACE and USGS crews measured about 2.2 velocity profiles per second. The USACE crew operated their boat at an average speed of 4.3 ft/s, which resulted in an average distance between velocity profiles of about 1.8 ft. USGS operated their boat at an average speed of 2.5 ft/s, which provided a velocity profile at about 1.3 ft spacing across transects (Figure 9). Appendix A provides a summary of transect measurement data.

During the ADCP field surveys, onboard data processing provided real time information on the velocity data being acquired. The processing uses WinRiver software, which was developed by RD Instruments (2001a), in cooperation with the USGS, as a part of a cooperative research and development agreement. In particular, the stick ship's track (Figure 10) indicates the depth averaged flow velocities for each velocity profile at their relative east and north positions indicated by the ADCP's bottom track. The easting, northing, vertical, and error velocities also are plotted with depth for each velocity profile (Figure 11).

In addition, cross sectional plots show the speed (Figure 12) and direction of flow for all depth cells in the transect. Cross sectional plots show the maximum depth of measured velocities, as the lighter line at the lower limit of the measured depth cells, and the channel bottom, as the darker line at the bottom of the plot. These data are monitored continuously to ensure the integrity of the velocity survey. Missed ensembles are displayed as white vertical streaks in the profiles. Many data acquisition characteristics of the ADCP were specified by commands in the configuration files (RD Instruments, 2001b). Typical ADCP configuration files for USACE and USGS measurements in the July 2002 ADCP survey of Detroit River are shown in Appendix B.

In both USACE and USGS surveys, discrepancies between GPS and bottom track coordinates at the first and last ensembles were used to scale and rotate bottom track measurements of position and to adjust the azimuths of horizontal velocities. The following paragraphs detail the mathematical procedures involved in applying these adjustments. The precise GPS data are thought to provide a more accurate indication of the absolute position of the profiles than positions based on bottom track measurements, even after these adjustments. The relative positions of the profiles and the horizontal rotation of the velocity vectors, however, are more appropriately based on the bottom track information. Finally, post processing was used to reformat the ADCP data to facilitate its subsequent use.

Let

and

represent the easting coordinates in Michigan SPCS 83 (zone 2113) International feet at the beginning and ending ensembles derived from a GPS measurement, and let

and

be the corresponding northing coordinates. Also, let

and

be the change in beginning and ending eastings and northings indicated by GPS. Then, the distance between the endpoints of the transect is calculated as

Similarly, for

representing the easting and northing of the first ensemble on the transect indicated by the bottom track (BT) and

and

representing the easting and northing of the last ensemble on the transect, the corresponding distance between end points indicated by the BT is

The resulting scale factor, which represents the ratio of the length between the first and last ensembles measured by the GPS and ADCP bottom track respectively, is computed as

This ratio is multiplied by distances indicated by the bottom track to obtain distances consistent with GPS measurements.

The azimuth of the transect, as defined by the end points of the transect from GPS measurements is

where the azimuth is the angle computed by the arctangent (arctan) function from the specified changes in northing and easting. Here, the results of the arctangent function are positive when measured clockwise from zero degrees north (along the positive y-axis);

(360 degrees) was added to negative values of the AzmGPS so that all reported azimuths would be positive. Similarly, the azimuth of the transect defined by BT measurements is

The corresponding angle between transects as defined by GPS and bottom track measurements is

Application of the scale and angle adjustments to positional data is described in the following paragraphs. For each velocity profile indexed by n, for

, the adjusted easting and northing was computed as This equation contains two rows of equations. See long description

[D] (10)

where sin and cos are the sine and cosine functions, respectively. The adjusted easting and northing velocity components,

and

, were computed from the measured velocity magnitudes and azimuths,

and

for

. Appendix C provides a summary of measurement times and adjustments applied to individual transects.

Both USACE and USGS crews surveyed six cross-sections, labeled DR19 to DR24 (Figure 13), on Detroit River near the Fort Wayne gage (Figure 5) to provide a basis for comparing flows and measurement characteristics during the July 2002 survey. In addition, the second ADCP survey of the reach was considered appropriate because the USACE crew reported difficult measurement conditions in the reach caused by high wind and wave conditions (Paul Taylor, USACE, oral commun., July 2002). Differences between measurement conditions during the USACE and USGS surveys, however, may have degraded the comparability of velocity data.

USACE surveyed the six cross sections in Detroit River by use of ADCP on July 10, 2002 from about 7 to 11 a.m. During this period, winds, as indicated by data on Lake St. Clair (Figure 7), averaged 20.3 miles per hour from the northeast (38.5 degrees east of north), with gusts up to 27 miles per hour. These fresh winds were associated with an increase in water levels of about 1 ft near the mouth of Detroit River at the NOAA gage at Gibraltar (Figure 6). Water levels near the head of Detroit River at the NOAA gage Windmill Point decreased only about 0.2 ft during this period. Rising water levels near the mouth of Detroit River and decreasing water-level fall within Detroit River were associated with generally decreasing total flows, which were estimated to average 167,000 ft³/s, during the USACE measurements of the reach. The USGS measured flows in the selected reach on July 19, 2002 from about 8 a.m. to 12 p.m. During this period, winds averaged 7.9 miles per hour from the northeast (29.5 degrees east of north). These gentle winds were associated with nearly constant water levels and total flows, which averaged about 188,000 ft³/s.

The measured flow component of total flow observed by USACE was consistent with the generally decreasing estimated flows in Detroit River during the morning of July 10, 2002 (Figure 14). The measured flow component in the three more upstream cross sections (DR19 to DR24) showed a fairly consistent decrease in magnitude, given that the right-to-left transects (starting on the United States side of the river) were measured before the left-to-right transects (starting on the Canadian side). The measured flow component stabilized at a lower value in the three more downstream transects at about 10 a.m. The measured flow components observed by the USGS crew reflected the more stable water-level conditions and higher estimated flows in Detroit River during the morning of July 19, 2002. Average point velocity magnitudes measured by the USACE tended to lower than those measured by USGS, which is consistent with the flow data (Figure 15). Average velocity azimuths measured by USACE and USGS showed greater consistency than velocity magnitudes (Figure 16).

Both velocity magnitude and azimuth data showed less variability in the USACE data than in the USGS data. In particular, the heights of the box plots, as indicated by the differences between the interquartile ranges (the differences between the 0.75 and 0.25 quartiles), were consistently greater in the USGS data than in USACE data (Figures 15 and 16). In contrast, greater variability in magnitude and azimuth data would have been anticipated during conditions of higher winds and waves, which may have decreased the stability of the ADCP platform. The pitch (front to back movement) and roll (side to side movement) of the ADCP unit was measured for each velocity ensemble. The pitch distribution of the USACE measurements (Figure 17) indicates an average pitch of about -1.8 degrees, with the 0.25 and 0.75 quartiles typically varying from -1.3 to -2.5 degrees. In contrast, the pitch distribution of USGS measurements showed considerably more range, although both positive and negative pitch angles were common. Similarly, the distribution of roll measurements (Figure 18) showed greater ranges in USGS data than in the USACE data. Neither USACE nor USGS data showed consistent departures from a roll of zero degrees. The greater variability in the pitch and roll characteristics of the ADCP unit operated by the USGS may have contributed to greater variability in the velocity measurements obtained by USGS than those obtained by the USACE. Because both USACE and USGS measurements generally had pitch and roll values less than 5 degrees, horizontal velocities are considered reliable (Simpson, 2001). The accuracy of vertical velocity components is highly sensitive to pitch and roll (Simpson, 2001). Thus, the utility of reported vertical velocity data are of uncertain utility.

Consistency of the measured flow component for transects surveyed in opposite directions along the cross section provides a measure of the accuracy of the flow measurement. Both USACE and USGS data show consistent measured flow components in transects surveyed from left-to-right (starting nearer the Canadian shoreline) with flow measured in transects surveyed from right-to-left (starting nearer the shoreline of the United States) (Figure 19).

The data obtained during the ADCP survey were initially stored in a binary file written by the WinRiver software. This software also can output an ASCII (American Standard Code for Information Interchange) formatted data file, the contents of which are described in detail by the help file in the WinRiver software. This ASCII format includes three sets of lines that describe the ADCP measurement (Table 2). The first set of lines describes comments and properties common to the entire transect. The second set of lines describes summary properties of a particular ensemble (velocity profile), which include position, timing, and number of depth cells. The third set of lines describes velocity characteristics of individual bins (depths cells). The number of depth cells specified in the configuration file fixes the number of lines in the third set. The number "-32768" in the third set of lines indicates that no velocity data were obtained at that depth cell, often because the depth interval exceeded the maximum channel depth at the location of the ensemble. The second and third set of lines is repeated for each ensemble. A detailed description of the ASCII output format is provided in the help file of the WinRiver application and a brief description is provided in Table 2.

The ADCP position and velocity data were modified as discussed in the preceding sections and reformatted ASCII files were created to facilitate the analysis of flow velocities in Detroit River. The new format is a regular rectangular array of (column oriented) data, similar to that seen in a spreadsheet, with each valid velocity measurement at a depth cell contained in an individual row. The column contents of each row are described in Table 3. The new format includes all valid data from a cross section by combining data from both the left-to-right and the right-to-left transects in a single file. The column "L2R" identifies the direction the transect was surveyed. Together, the ADCP data files provide point velocity data at more than 3.5 million locations in Detroit River.

The adjusted ADCP data obtained during this survey can be retrieved through the internet by use of Figure 20. To facilitate online storage and retrieval, the ASCII files were compressed using the WinZip utility. Although the WinZip utility is needed to uncompress the retrieved files, an evaluation version of the utility is freely available at the following URL: http://www.winzip.com/. A vicinity map showing transect locations can be obtained by left clicking on the red rectangles of the index map. Clicking on the transects shown on the vicinity maps will initiate the retrieval of the ADCP data for the corresponding cross section. The file names of the cross sections are indicated on the vicinity maps. The extension of the file names upon retrieval is "zip." Once downloaded, the files can be decompressed with the WinZip utility, and will have the same file name, but with the extension "3d."

Crews from the U.S. Army Corps of Engineers (USACE) and the U.S. Geological Survey (USGS) used acoustic Doppler current profilers (ADCP) to measure flow velocities in Detroit River. The survey was conducted to provide a basis for describing flow patterns in Detroit River and for improving the calibration of a hydrodynamic model of the waterway. The model is being developed to assess the susceptibility of public water intakes to contamination.

The surveyed reach extends from the head of Detroit River at the outlet of Lake St. Clair near Detroit, Michigan, downstream to the mouth of Detroit River on Lake Erie, a distance of about 32 miles. The survey was conducted from July 8-19, 2002. Both crews used RD Instruments' Workhorse Rio Grande ADCP units operating at 600 kHz, and were deployed from moving boats. Velocities were measured at 130 cross sections separated by an average distance of 1,800 ft. At each cross section, two transects were measured, one from left-to-right across the river (looking downstream) and a second running right-to-left. For each transect, ensembles describing flow velocities with depth were generally measured 0.8 to 2.2 ft apart as the boats traversed the river. Each ensemble contains a profile of flow velocities at 1.64 ft (0.5 m) intervals of depth, starting less than 3 ft below the water surface and extending nearly to the bottom of the river. More than 3.5 million point velocities were measured during the survey. The ADCP position and velocity data have been adjusted for the effects of the local magnetic declination and differences between the channel positions indicated by the ADCP bottom track measurements and global positioning system (GPS) measurements at the ends of the transects. In addition, the data has been reformatted to facilitate analysis. Flow data obtained by USACE and USGS crews is believed to be comparable because of similarities in equipment and techniques. Flows measured in transects surveyed from left to right are comparable with flows in transects that were surveyed from right to left. This report provides a mechanism for downloading the adjusted and reformatted ADCP data through the internet.

¹ Ultrasonic refers to sound frequencies greater than 24 KHz, which is the upper limit of frequencies that a human can hear.

² Christian Johann Doppler discovered the principle of physics that relates the change in frequency between a source and an observer to their relative velocity.