Autonomous Lagrangian Float uses High-Precision SonTek Sonar
Dr. Eric D’Asaro and his group at the Applied Physics
Laboratory, University of Washington have been building water-following
"Lagrangian" floats to study circulation and mixing in the upper ocean and
coastal waters. The floats (see picture) combine a compressible hull, active
buoyancy control coupled to high accuracy CTDs and a folding cloth drogue to
accurately follow the motion of the water surrounding them. Alternatively, the
float can vertically profile. Onboard data logging and satellite communications
allow data to be reported in nearly real time and allow the float mission
changed by remote control.
A key goal for this instrument is to measure the water
velocity relative to the float. This is important both for determining how well
the float actually follows the water and for measuring small-scale shear, a key
quantity in understanding ocean mixing. The expected signals are usually small,
a few centimeters per second, but can occasionally be as large as 20 cm/s.
Velocity has to be measured as fast as once per second. The velocity sensors
have to be small, no larger than a few centimeters on the outside of the float,
and consume little power, so as to operate continuously on multi-month missions.
A 1.5-MHz SonTek pulse-pulse coherent sonar based on
the Argonaut family of instruments (see photo) was chosen for this job. Three transducers are
mounted on the top of the float each pointing at 45 degrees from the vertical
and separated by 120 degrees azimuthally. Every second, three pairs of pulses
are transmitted from each transducer. The return echoes from each pair are
compared to compute the water velocity in the direction of the beam averaged
over a volume extending about 50 cm from the transducer. This scheme
has the advantage of providing extremely high accuracy (a few mm/s) from a
single ping. However, it suffers from an ambiguity problem: the measurement of a
velocity V can result from actual velocities of V±
nVmax where Vmax is an ambiguity
velocity and n is an arbitrary integer. SonTek solved this
problem by having a different Vmax for each of the three pulse pairs. This allows the ambiguity to be resolved out to approximately 3 Vmax.
For our sonar, the values of
Vmax are 76.1, 124,
and 210 mm/s for pulses, 3, 2, and 1 respectively. The
graph on the right shows sample data
from one second of measurement. Ping 3 (blue) is measured at –51 mm/s. However,
because of the ambiguity, it could also be 127 mm/s or –25 mm/s. The blue
Gaussian curves show these three possibilities; the width of the curves shows
the uncertainty in each measurement. Pings 2 (red) and 1 (black) produce a
different set of possible values, but with larger errors. Only when all three
pings line up, is the velocity valid. In this example, this occurs for a
velocity of about 100 cm/s. We thus choose the 127 mm/s measurement from Ping 3
as the correct velocity.
A sample of ocean data processed in this way is shown
in the graph on
the left. Each dot shows the results of a single ping along one beam of the
sonar. Color coding identifies the ping. Most of the time, the three pings can
produce a good velocity as shown by the green curve. However, within the gray
box no good velocity is found. This is because electronic noise within our
instrument contaminated the measurements for a few seconds. The multiple-ping
system allows us to automatically quality control the data for such problems,
producing a very clean and accurate data set. Notice that velocity fluctuations
of only a few mm/s and lasting only a few seconds are easily resolved.
Data from all three beams is combined with compass data to
produce east, north, and vertical velocity. A typical 2-3 month float mission
produces over a hundred megabytes of 1-Hz sonar data with the expenditure of
only a few megajoules of energy. For example, the graph on the right shows the spectrum of
the clockwise and anticlockwise rotating components of velocity measured by a
sonar on a float drifting off Oregon. The strong dominance of clockwise rotating
motions near the inertial frequency, and the absence of a white noise floor, shows
that the instrument is working well and providing consistent accuracies of a few
mm/s.
*Photos and graphs courtesy of Applied Physics Lab,
University of Washington
Details about SonTek's Argonaut product family can be found at:
