2000 Joint Conference on Water Resources Engineering and Water Resources Planning & Management

July 30 - August 2, 2000 — Minneapolis, Minnesota

Analyzing ADV Data Using WinADV

Tony L. Wahl (1)

(1) Hydraulic Engineer, U.S. Bureau of Reclamation, Water Resources Research Laboratory,
P.O Box 25007, Mail Code D-8560, Denver, CO 80225-0007; Phone: 303-445-2155;
Fax: (303) 445-6324; email: [email protected]

Abstract
Since their introduction in 1993, acoustic Doppler velocity meters (ADV’s) have quickly become
valuable tools for laboratory and field investigations of flow in rivers, canals, reservoirs, the
oceans, and around hydraulic structures and in laboratory scale models. The post-processing and
analysis of ADV data present unique requirements compared to traditional current-metering
equipment, due to the types of data obtained, the analyses that are possible, and the need for
filtering of the data to ensure that technical limitations of ADV’s do not adversely affect the
quality of the results. One useful software tool for this purpose is the WinADV program, written
by the author for use in the Bureau of Reclamation’s Water Resources Research Laboratory, and
made available to the public via the Internet since 1996. The program is designed for post-
processing of real time data files (*.adv files) recorded by SonTek and Nortek ADV’s, and offers
powerful filtering and processing options that make it valuable for field and laboratory studies.
The software has been regularly upgraded to maintain compatibility with new file formats, and
also now has limited capability to analyze data collected by SonTek’s Argonaut series of
monostatic Doppler current meters. This paper describes the software and illustrates the
application of some of its advanced features to typical hydraulic measurement situations.

Introduction
Acoustic Doppler velocity meters (ADV’s) in a variety of configurations are widely used for
laboratory and field measurement of 2- and 3-dimensional water velocities. The instruments are
relatively rugged, easy to operate, and can be readily mounted and maneuvered within the flow
field. Although the probe is inserted into the flow, the sensing volume is several centimeters
away from all physical parts of the probe, so the presence of the probe generally does not distort
the measurement. Velocities can be measured in any body of water containing suitable acoustic
scatterers, or scattering particles may be added to the flow. Kraus et al. (1994) described the
technical features and capabilities of ADV’s.
To some degree, ADV’s can fulfill the roles of a range of other flowmeters, including propeller-
type current meters, hot-film probes, electromagnetic current meters, and laser-doppler
velocimeters, depending on the methods used to collect and process the data. The primary data
stream provided by an ADV is a time series of velocity vector components. The sample
reporting rate can be varied from 0.1 Hz to 25 Hz on the instruments that operate at an acoustic
frequency of 10 MHz, and sampling rates of up to 200 Hz are possible with some of the newer
probes. Some averaging always takes place at the time of data collection, since the probes
collect data internally at sampling rates higher than the nominal upper limit and average it before
reporting and storing the data at the chosen sampling rate. Data processing needs can include
additional averaging over time periods greater than 10 seconds, combining velocity components
into resultant vectors, and computing basic statistics needed for analyses of turbulence. At times
there is also a need for more sophisticated analyses. For example, power spectrum plots and

1
 2000 Joint Conference on Water Resources Engineering and Water Resources Planning & Management
July 30 - August 2, 2000 — Minneapolis, Minnesota

histograms can reveal periodicity and unusual distributions of velocity in the time series, and
model scaling computations and adjustments for relative probe motion are commonly needed in
laboratory studies to allow quick interpretation of data collected from scale models.
ADV’s have several unique data processing requirements due to their method of operation and
some of the inherent limitations of the acoustic Doppler measurement technique. Unlike many
other measurement methods that exploit a physical property of water to obtain a velocity
measurement (e.g., momentum to turn a propeller, mass flow through a magnetic field), ADV’s
actually measure the velocity of the scattering particles in the flow. As a result, the quality of the
measurement is dependent on the presence of scatterers and their behavior within the sampling
volume. To ensure that ADV measurements provide an accurate representation of the flow
velocity, one should evaluate two additional parameters provided in the ADV file, the signal-to-
noise ratio (SNR) and the correlation (COR). Filtering the data using one or both of these
parameters can improve the quality of the measurement.
The implementation of the acoustic Doppler measurement technique in an ADV probe is
accomplished by measuring the phase change produced in the acoustic signal reflected off of
scattering particles in the flow. The magnitude of the phase difference is proportional to the flow
velocity. The phase difference can be positive or negative, allowing ADV’s to measure both
positive and negative velocities. The limits of the measurement range are the velocities
corresponding to phase differences of ±180°. The probe can be configured during a specific
measurement session to establish the user’s choice of measurement ranges. Because positive
phase angles greater than 180° cannot be distinguished from negative phase angles between
–180° and 0°, if the phase angle is greater than 180° it will be seen as a negative phase angle
between –180° and 0°, and will be incorrectly interpreted as a negative velocity. Similarly,
phase angles less than –180° (e.g., –181°) are interpreted as positive phase angles and positive
velocities. As a result, when the flow velocity approaches the limits of the measurement range, it
is possible for the probe to report velocities that alternate from large positive to large negative
values and vice-versa. This behavior is known as aliasing, and erroneous velocities reported by
the probe when operating in this range are called velocity ambiguities. Identifying and
eliminating velocity ambiguities is critical to obtaining accurate results when measuring flow
velocity with an ADV.
The variety of data products that can be derived from an ADV time series, the need for filtering
and quality control checking of the data, and the need for scaling and adjustment of the data to
account for model scale and measurement techniques makes customized data processing
software a valuable asset when analyzing ADV data. To fulfill these needs, the WinADV
software program was developed for use at the Bureau of Reclamation’s Water Resources
Research Laboratory in Denver, Colorado. Since its development in 1996, the program has been
available to the public for download from the Internet, and has been periodically updated to
maintain compatibility with newer ADV data formats and provide new and improved features.

The WinADV Program

The WinADV program is written in Visual Basic and compiled in a 32-bit version for Windows
95/98/NT/2000 computers and a 16-bit version for Windows 3.1 computers. The program
includes an integrated online help system and is distributed with an electronic copy of a printable
user’s manual. The initial purpose for developing the program, and still its primary function, is
to provide a means for quickly displaying the velocity time series in a format similar to that

2
 2000 Joint Conference on Water Resources Engineering and Water Resources Planning & Management
July 30 - August 2, 2000 — Minneapolis, Minnesota

provided by ADV data collection software (i.e., the adv.exe or adf.exe programs available from
SonTek and Nortek, or the CollectADV program available from Nortek).
WinADV is able to load the *.adv files created by the current and past versions of the real time
data acquisition software programs provided by SonTek and Nortek. The program can load files
containing data collected from just one probe, or from multiple probes connected to a single data
acquisition computer (up to 16 using the most recently manufactured ADV’s and data acquisition
software). The program cannot load the *.adr files created by remotely deployed ADV’s. These
files contain data from multiple bursts of ADV operation occurring at variable intervals, rather
than the single, simple time series contained in a *.adv file. WinADV is also incompatible with
the *.lfw file format used by the first generation of ADV instruments known as the ADVLab-1.
A recent addition to the WinADV program is the ability to load data files created by the SonTek
Argonaut series of monostatic Doppler current meters. These instruments produce a data stream
similar to that of the ADV, i.e., a single time series of velocity components and associated signal
strength and measurement quality parameters.
Three primary views of the ADV data are available: time series, histogram, and power spectrum.
Each chart is provided on a separate tab of the main WinADV screen (fig. 1). The time series
chart can display velocity components and magnitudes, signal amplitude, signal-to-noise ratio,
and correlation score. For probes equipped with other integrated sensors, such as compass-tilt
sensors, pressure transducers, or temperature measurement devices, these parameters may also be
shown. In addition, flags embedded within the file to mark points of particular interest noted
during data collection can be displayed on the same chart used to show the SNR and COR time
series. The user may choose to display either the average or minimum values of the SNR and
COR parameters, since multiple values are recorded within the ADV file (one for each acoustic
beam). Several formatting options are available for the display of the time series graphs (e.g.,
line chart, symbols-only, bar chart, etc.)
The histogram view shows a histogram of the collected data, again with options to display one or
more velocity components, signal
amplitudes, SNR or COR values,
or data collected from integrated
sensors. The chart can be shown
in the format of a probability
density function or a cumulative
distribution function. The
histogram, like the time series
and power spectrum charts, can
be saved to a file, copied to the
system clipboard, or printed.
Figure 2 shows a histogram chart
created by WinADV.
The power spectrum chart can
display power spectra for the
same parameters as those
available on the time series and
histogram charts. Power Figure 1. — Main screen of WinADV showing velocity
spectrum charts may be valuable time series, SNR and COR values, and recorded flags.

3
 2000 Joint Conference on Water Resources Engineering and Water Resources Planning & Management
July 30 - August 2, 2000 — Minneapolis, Minnesota

in studies involving waves or other periodic tlwmet - Histogram - Velocities (cm/s)

flow oscillations, and can also be used to

detect unintended periodic probe motion V-x

that may have occurred during data

collection. The user chooses the size of
sampling window to be used for calculating V-y

the power spectrum; the number of samples

used must be an integer power of 2, i.e., 2n.
If there are sufficient samples available, V-z

WinADV will compute power spectra for -40 -30 -20 -10 0 10 20 30
Velocity (cm/s)
adjacent blocks of 2n samples and average
the spectra together before plotting. Power
Figure 2. — Histogram chart created with
spectra are computed without the use of any
WinADV and saved in a Windows metafile.
windowing function; the objective in
providing the power spectrum chart in WinADV is only to supply a means for easily identifying
periodicity in the data. WinADV does not provide any more sophisticated frequency domain
analysis options, since details of the computational techniques vary depending on the application.
In addition to displaying the measured data in an ADV file, WinADV also provides the
capability to view and print the probe and measurement configuration used to record the data.
This information is similar to that provided by the GETCTL utility program provided by SonTek
and Nortek. Information provided includes probe types, serial numbers, velocity range settings,
and measured distances to the sampling volume and nearby boundaries.
WinADV provides powerful data processing options which will be discussed in more detail later
in this paper. A quick summary view of the computed statistics for a single ADV file can be
obtained from the main screen of WinADV by clicking on the Stats Summary button. Statistics
for the file are displayed in a scrollable grid, just as they would be processed using the data
processing options described later in this paper.
WinADV can load single data files for review, filtering, and processing, or the program can be
used to quickly review or process multiple data files contained within a single subdirectory. The
View or Process Multiple Files command leads the user to a file selection dialog box in which
multiple files may be selected. If the View button is clicked from this page, WinADV will load
the first data file and add Next and Previous buttons to the display to allow the user to browse
forward and backward through the list of selected files. If the Process button is clicked,
WinADV will process each of the selected files in sequence, using the same set of processing
options on each file. Details regarding the processing options are discussed later in this paper.

Program Options
Data handling options are available in four categories:
• Sampling options to define the range or ranges of samples to be analyzed
• Filtering of data to exclude poor-quality or erroneous data
• Traversing options to define the motion of a moving probe
• Scaling to convert data to prototype scale and make simple coordinate transformations
Options settings can be saved to a *.opt file for later use with other ADV files.

4
 2000 Joint Conference on Water Resources Engineering and Water Resources Planning & Management
July 30 - August 2, 2000 — Minneapolis, Minnesota

Partitioning Data within a Single ADV File Using Sampling Options - Although the real time data
acquisition software used to operate an ADV probe only allows the storage of a single time series
of data within each file, there is a limited capability to mark multiple segments of data within a
file so that the file may be processed in pieces. This is sometimes preferable to recording
individual files for each sequence of data. This is most readily accomplished by marking flags in
the data file using the adv.exe or adf.exe programs. Each time the F7 key is marked while
recording data, the flag counter in the ADV file is incremented by one. The flag marks can then
serve as indicators of the start and/or end of sampling windows to be analyzed. Through its
Sampling Options screen, WinADV offers nine methods for subdividing the data file into
sampling windows for viewing and analysis. When the data are processed by WinADV (to
compute mean velocities, turbulence quantities, etc.), each sampling window is processed
separately. The sampling window choices are:
• All Samples - The entire file is used.
• All Samples (delimited by flags) - A separate set of summary statistics is computed for each
range of samples between consecutive flag markers (e.g., 1-2, 2-3, 3-4, 4-5, …).
• Odd/Even Flags - Each odd-numbered flag (1, 3, 5, 7, …) marks the start of a sampling
window, and each even flag marks the end of the sampling window.
• Even/Odd Flags - Opposite of above; this may be useful if you get off-sequence while
marking flags during data collection.
• Use Pause and Restart Marks – Works with recent versions of the adv.exe and adf.exe data
collection programs that offer a burst mode of data collection in which the user can pause the
recording of data to the file and then restart recording at a later time. The event marker (flag)
is incremented by ten at each restart.
• N Samples After Each Flag - Each flag mark is considered to mark the start of a sampling
window containing N samples (N specified by the user).
• T Seconds After Each Flag - Similar to above, except that the length of the sampling
window may be specified in seconds.
• Range of Sample Numbers – The user can define the range of samples to be analyzed using
the sample number identifier. The first sample in the file is considered sample number 1.
• Range of Sample Times - This is similar to the range of sample numbers, except that the
range is specified using times in seconds rather than sample numbers.
Once sampling windows are established, the user will note changes in the appearance of the time
series, histogram, and power spectrum charts. The time series chart will show only the data
contained within the defined sampling windows, while the histogram and power spectrum charts
will operate on only a single sampling window at one time. The user can browse forward and
back through the available sampling windows.
A powerful feature for controlling and adjusting sampling windows in ADV files is to use the
sampling options based on flag marks in combination with the flag editor provided in WinADV.
The flag editor allows the user to create a custom set of flag marks for a data file and save them
separately in an ASCII file for later recall (WinADV does not alter the original file). Using the
flag editor and the flag-based sampling windows, multiple sampling windows could be
established in a single ADV file, even if no flag marks were made when the data were recorded.

5
 2000 Joint Conference on Water Resources Engineering and Water Resources Planning & Management
July 30 - August 2, 2000 — Minneapolis, Minnesota

Data Filtering Options - Data filtering is an important aspect of the analysis of ADV data. The
signal-to-noise ratio and correlation scores are good indicators of some of the possible problems
encountered while using an ADV, and WinADV offers filters based on their values. Filters are
also available to address problems that cannot be detected by analysis of the SNR and COR
values. WinADV’s filters can be applied separately or in combination.
The SNR parameter indicates the relative density of acoustic scatterers in the flow and the
resulting strength of the signal received compared to the noise level of the instrument. ADV
manufacturers recommend an SNR value of at least 5 when measuring average flow velocities,
and 15 or higher when measuring instantaneous velocities and turbulence quantities. If possible,
problems with low SNR should be corrected during data collection. Although it is not
specifically recommended by ADV manufacturers, filtering based on the SNR value may be
useful in some situations, such as when only a short segment of a data file is affected. WinADV
offers a filter based on the SNR value, with a user-selected cutoff level.
The correlation parameter, COR, is an indicator of the relative consistency of the behavior of the
scatterers in the sampling volume during the sampling period. ADV’s collect data at a higher
sampling rate than the sample reporting period, and the COR parameter indicates the consistency
of the multiple measurements that take place within each sampling period. The value varies
from 0 to 100, and ADV manufacturers have typically recommended filtering to remove any
samples with correlation scores below 70. WinADV offers such a filter, with the option to adjust
the cutoff value. In the author’s experience, samples with correlation values much less than 70
can sometimes still provide good data, especially when the signal to noise ratio is high and the
flow is relatively turbulent (a condition that seems to lead to generally lower correlation scores).
The problem of overranging of ADV probes was mentioned earlier. When the flow velocity is
too high relative to the velocity range setting of the instrument, aliasing of the data occurs,
creating velocity ambiguities. Figure 3 shows velocity ambiguities recorded in the flow in a
laboratory flume. The velocity of this flow was correctly measured during the first 23 seconds of
the measurement period shown. When the discharge was increased slightly, the probe
overranged and the reported velocities changed sign. Near the end of the period shown, the
velocity was reduced and positive velocity spikes began to occur as the flow reentered the
measurable range. It is critical to note that the SNR and COR values do not give a clear
indication that there is a problem, and would not be useful for filtering. Clearly, other filtering
strategies are needed if this type of problem is to be eliminated from the analyzed data.
WinADV’s velocity cutoff filter can be helpful when dealing with data files containing velocity
ambiguities. The user may specify a cutoff value for one of the three velocity components and
whether to keep data above or below the cutoff. This filter is successful in data files in which the
overranging is consistent and produces clear and dramatic changes in the reported velocity, as
shown in figure 3. This filter is less successful when overranging is more severe (i.e., velocity is
well above the value that first causes overranging), or when the flow is very turbulent and
overranging occurs differently from sample to sample (i.e., different acoustic beam overranging
each time).
A second filtering method that is sometimes successful with data files containing velocity
ambiguities is the spike detection filter. This filter was developed at the New Zealand Institute
for Water and Atmospheric Research (NIWA) by Nikora and Goring (personal communication).
It is based on the concept that there should be a physical upper limit to the change in flow

6
 2000 Joint Conference on Water Resources Engineering and Water Resources Planning & Management
July 30 - August 2, 2000 — Minneapolis, Minnesota

velocity (i.e., the acceleration) that can occur in a flow, and any measurements that indicate
higher accelerations should be excluded from the analysis. This filter successfully removes
velocity ambiguities, but only if they are of short duration. If overranging causes velocity
ambiguities to persist for several sample periods, then they will not all be detected as spikes.
The spike detection filter also successfully removes other erroneous data points whose origin
may have been low signal strength or inconsistent scatterer behavior (i.e., low correlation). To
make the filter dimensionless, the acceleration threshold is specified in g’s.
To reduce the chance of velocity ambiguities tainting the data collected with an ADV, it is best
to record data at the highest normal sampling rate (i.e., 25 Hz for 10 MHz ADV’s), whenever
there is the possibility of encountering velocities that might overrange the probe. At lower
sampling rates (e.g., 0.1 Hz), velocity ambiguities may occur for just a portion of the sampling
period, and the erroneous data will be averaged in with valid data before the sample is reported
to the user and recorded in the ADV file. This makes it more difficult to visually identify and
automatically filter out ambiguities.
Velocity ambiguities sometimes occur in two situations where users do not expect them. First,
large velocity components directed toward or away from the transmitting transducer of an ADV
can cause overranging even when the total velocity magnitude is well below the nominal range
setting of the probe. This is due to the geometric arrangement of the transmitting and receiving
acoustic transducers, which causes the velocity range normal to the transmitting transducer face
to be about one third of the range in the plane parallel to the transducer face. The stated nominal
range setting is most closely associated with flow in the plane parallel to the transmitting
transducer face. Second, the range setting of an ADV probe can sometimes change during the
boundary-adjustment phase of configuring the probe at the start of each measurement. This
occurs when the probe detects nearby acoustic boundaries that might cause interference with
operation of the probe. To overcome this interference, the lag times of the acoustic signals
transmitted by the ADV are reduced, causing a reduction of the velocity range that may lead to
overranging of the probe. The fact that the velocity range was reduced is not noted in the
information provided by the GETCTL utility. However, WinADV can read the lag time settings
of the probe from the header of the
ADV file and compare them to the alias - Velocities (ft/s)
1.0
standard lag time settings for the 0.8
0.6 V-x
chosen velocity range. If they do not 0.4

match, WinADV computes a new 0.2

0.0
10 20 30 40 50 60 70
nominal velocity range and displays -0.2
-0.4 V-y

the chosen range and the actual range -0.6

Time (seconds)
in the upper middle portion of the
main WinADV screen, just below the
Probes button. If unexpected alias - SNR/COR/Flags

overranging is occurring, the user 100

80

should check to see if the velocity 60

40

range was reduced. 20

0
0 10 20 30 40 50 60 70
Time (seconds)
The final filtering option offered in
WinADV is the filtering of samples
marked with a communications error Figure 3. — Example of velocity ambiguities and
flag. This is only an issue with data aliasing caused by overranging of an ADV probe.

7
 2000 Joint Conference on Water Resources Engineering and Water Resources Planning & Management
July 30 - August 2, 2000 — Minneapolis, Minnesota

files collected from field-type ADV probes over a serial interface, using the adf.exe program.
Momentary loss of communications over the serial cable can lead to corruption of individual
samples, which are then marked in the ADV data file with flag 999. If the user chooses to filter
out communications errors, WinADV gives the option of replacing the filtered samples with
interpolated data. This allows the computation of power spectra, which require a continuous
record. The interpolation feature is only available with respect to filtering of communications
errors, not for filtering by the other methods discussed above.
Features for Scale Modeling – The user can select the geometric scale of a Froude-scaled model,
and WinADV will calculate the corresponding velocity scale ratio from the relation λ V=λ L0.5,
where λ V and λ L are the velocity and length scale ratios, respectively. WinADV uses a separate
scaling ratio for each component direction. This allows the use of a negative scale factor to
reverse the sign of one or more of the measured velocities, a feature that is valuable when the
ADV probe can only be oriented at 180° from the desired direction.
The maximize coverage of a model area while minimizing data collection effort, probes are often
installed on traversing systems. To properly interpret these data, the motion of the probe must be
added to the velocities that are measured relative to the probe. The traversing options in
WinADV allow the user to specify the velocity of the probe in each component direction (using
the same coordinate system as the measured velocities), and optionally to plot the velocities
versus the probe position, rather than the sample time. In addition, a traversing period can be
specified if the probe is making multiple passes (i.e., reversing direction and returning to the
starting point), a situation described as a bouncing traverse. The traversing period entered into
WinADV is the travel time from one end of the traversing system to the other (not the time for a
complete lap). During the return-travel period, WinADV reverses the signs of the probe
velocities. Figure 4 shows an example of data collected with a traversing system.
Data Processing Options - The processing module of WinADV operates on an entire ADV file,
or on the individual sampling windows defined in it by the user. Data files are exported in
ASCII files compatible with most spreadsheets. Processing can be carried out on a single file or
on multiple files contained in a single subdirectory. Output options include histogram, power
spectrum, and time series data export, as well as calculation and export of summary statistics.

a01lrd34 - Scaled Velocities

1.0
0.8
V-x
0.6
0.4
0.2
0.0 V-y
1 2 3 4
-0.2
Probe Position

Figure 4. — Data collected from a 2-D side-looking ADV traversing a 1:12 scale fish screen
model. One pass was made up and down the length of the screen at a speed of 0.76 cm/s.
Velocities are in meters/sec and probe position in meters at prototype scale. V-x (the larger of
the two displayed velocity components) is parallel to the screen face, and V-y is normal to the
screen face. This figure was created in WinADV.

8
 2000 Joint Conference on Water Resources Engineering and Water Resources Planning & Management
July 30 - August 2, 2000 — Minneapolis, Minnesota

Histogram and power spectrum data export are relatively straightforward, with the program
simply producing the data needed to recreate the histogram and power spectrum plots shown in
WinADV. Time series data export produces a time series file that has been modified by the
sampling and/or filtering operations available in WinADV. Unfiltered and filtered time series
data export options are available, with the data written to ASCII files with the name advfile.vu
and advfile.vf, where vu and vf indicate unfiltered and filtered velocities, respectively, and advfile
is the base name of the ADV data file. In addition to the velocity data, signal-to-noise ratios,
correlations, and signal amplitude data are exported.
Summary statistics for the filtered and unfiltered time series data are exported into files with the
default names of Filtered.Sum and UnFilter.Sum. These files accumulate data from multiple
processing sessions, with the results from the most recently processed files added as new rows at
the bottom of the file. A partial list of the computed statistics includes:
• Mean, standard deviation, skewness and kurtosis of velocity components
• Covariances of velocity components
• Span of 95% confidence intervals on velocity components (i.e., difference between upper and
lower endpoints of the confidence interval)
• Mean velocity magnitudes, correlations, signal-to-noise ratios, and signal amplitudes
• Mean values of external sensor data stored in the file, such as pressures, temperatures, etc.
Details of the computational formulas for all of WinADV’s statistical parameters are given in the
online help system and printable electronic user’s manual distributed with WinADV.

Automated Data Processing with WinADV

WinADV offers several tools for simplifying the analysis and processing of ADV data. The
processing of multiple files in a single subdirectory can automate data processing when all of the
files are to be processed identically. Saving options settings in *.opt files is helpful when
different processing options are needed due to variations in probe setup, such as having to install
the probe at different orientations, etc. Truly automated data processing can be achieved through
the use of saved options files and the WinADV command-line interface. The command-line
interface allows the specification of an ADV file to be loaded on program startup, along with an
associated options file and instructions to carry out unattended processing of the data. Complete
syntax for the command-line interface is given in the help file and user’s manual.
An example application of automated processing is the use of WinADV to process the data
collected from the Bureau of Reclamation’s model studies for the fish screen structures at
Durango Pumping Plant. Eight different orientations of a traversing table were used, with
varying directions of probe motion and varying probe positions relative to the prototype fish
screen coordinate system. In addition, data were collected at three different elevations in the
model, and filtering was used to eliminate data with low correlation values. Collection of the
data was automated by starting and stopping the ADV probe using the same signals that
controlled the traversing system. The eight data collection configurations were designated LLD,
LLU, LRD, LRU, RLD, RLU, RRD, and RRU, and these designations were used in the file
names of the ADV files (e.g., B04LLD31.ADV indicates test flow condition B04, with data on the
left-hand (L) screen at the left-downstream (LD) traversing table position, with the probe at

9
 2000 Joint Conference on Water Resources Engineering and Water Resources Planning & Management
July 30 - August 2, 2000 — Minneapolis, Minnesota

prototype elevation 31. Only four options files were necessary for processing these data, LD.opt,
LU.opt, RD.opt, and RU.opt. A single data file could be processed with the DOS command
winadv32 B04LLD31.ADV LD.opt –pfsumfiltered.sum –pvf
which instructs WinADV to load the ADV file and options file, compute summary statistics on
the filtered data (-pfsum) and save them in filtered.sum, and export the time series data (-pvf).
To further automate this process, the DOS for command can be used:
for %f in (LD LU RD RU) do winadv32 B04L%f31.ADV %f.opt –pfsumfiltered.sum –pvf
This command would process all data files collected at prototype elevation 31. Incorporating
similar commands into DOS batch files can further automate data processing. Figure 5 shows an
example of data collected and processed using this technique.

Conclusions
The WinADV program is a valuable tool for filtering, analyzing, and processing data collected
from acoustic Doppler velocity meters. The program can be used to analyze ADV files recorded
using the real time data acquisition programs provided by ADV manufacturers. The software is
available for 32-bit and 16-bit Windows-based computers and can be downloaded by the public
from the Bureau of Reclamation’s Internet site at http://www.usbr.gov/wrrl/twahl/winadv.html.

References
Kraus, N.C., A. Lohrmann, and R. Cabrera (1994). “New Acoustic Meter for Measuring 3D
Laboratory Flows.” Journal of Hydraulic Engineering, Vol. 120, No. 3.
Nikora, V.I., and D.G. Goring. “Turbulence structure in a gravel-bed flow with weak bed-load.”
Draft manuscript submitted to Journal of Geophysical Research.

Acknowledgments Test A05 - Approach and Sweeping Velocities at Elevation 6434.0

1007 ft3 /s river flow, 500 ft3 /s pumped flow (median pumping)
3
P1-P4 running at 125 ft /s each, P5 is not running
Recent improvements to 4
Compare to A01, A02, and A03, in which P2-P5 were running
4

WinADV have been made under

a cooperative agreement between
3 3
the U.S. Bureau of Reclamation
and SonTek, Inc.
Velocity, ft/s

Sweeping velocity
2 2

Disclaimer
References to commercial 1 1

products in this paper are for

informational purposes only and Approach velocity
0 0
do not constitute endorsement by
the author or the U.S. Bureau of 0 10 20 30 40 40 30 20 10 0 0 10 20 30 40 40 30 20 10 0

Distance from Upstream End of Screen, ft

Reclamation.
Figure 5. — Example of velocity data collected using a
traversing system and processed by WinADV. The data
were collected using eight separate traverses of portions of
the model fish screen.

10