This note gives a brief description of the shipboard ADCP
and its processing from the Ron Brown cruise RB-01-08 in October
of 2001. Please note that this initial level of processing is meant
to provide an overview of the data suitable for a data report.
The time intensive visual editing of individual profiles for
bottom interference, CTD interference or other "glitches" has
NOT been performed, which implies that the dataset in its current
form should not be used for scientific analysis.
Processing was done with the CODAS software package developed
by Dr. Eric Firing's group at the University of Hawaii.
1) Scanning and loading
The raw data consist of 24 pingdata files numbered 000
to 023. They
contain five-minute ensembles. A comparison between ADCP and GPS
time indicates that the ADCP clock was within 2 seconds of GPS time,
which suggests that the userexit program UE4.EXE was used.
The raw data cover the time interval from 2001/10/09 17:05:34
(decimal day 281.712199) to 2001/10/25 10:46:38 (decimal day 297.449051).
Bottom tracking was turned on for the first two hours of the cruise.
All data were "loaded" into a CODAS database.
]2) Ashtech heading correction
The userbuffer contained heading differences apparently
derived from a
comparison of ship's heading and heading from a GPS antenna array
(also called "Ashtech heading" after a popular manufacturer of such arrays).
For a number of reasons, the Ashtech heading is currently not used to
replace gyro heading, but a comparison of the two provides a correction
to gyro oscillations such as the so-called Schuler oscillations, which are
strongly damped gyro oscillations with time periods of roughly 80 minutes
that may be excited by rapid ship's speed changes (e.g., station arrival and
departure). The data (first, second, third set) indicated relatively low gyro compass
oscillations here compared to other installations I have seen (e.g., the whoi ships),
which was at least partly due to the infrequent ship's accelerations
during this cruise.
Each profile was rotated by the Ashtech-gyro heading difference,
converting from gyro to Ashtech heading. An unknown constant heading
offset due to a misalignment of the Ashtech antenna array relative to the
ship's forward axis (of the order of 3 degrees or so based on the Ashtech plots)
will be adjusted during the transducer misalignment correction, i.e.,
the ADCP calibration.
There are two general types of ADCP calibrations: water
track and bottom track
calibrations. The former compares ship's accelerations including station arrival
and departures and ship's turns in the ADCP and the GPS record. In shallow
water (which for a 150KHz transducer means roughly 45 to 500 meters),
bottom track calibration compares the ship velocity relative to the bottom
as measured by the ADCP with the GPS-based ship's track. The water track
method is typically noisy, and requires many data points to provide meaningful
results - such as the frequent CTD stations from a typical WOCE cruise.
Bottom track calibrations also become more accurate with the length of
available ADCP bottom tracking; the roughly two hours from the beginning
of this cruise provide a very marginal set of transducer alignment and ADCP
amplitude. The calibration indicates, however, that the ADCP was basically
well aligned in that the amplitude is close to 1.00 and the alignment mirrors
the apparent offset of the Ashtech antenna array.
The constant ADCP calibration was determined from bottom
track to be
1.0028 for ADCP amplitude and -2.912 for transducer misalignment (relative
to the Ashtech antenna array). A "rotation" of that magnitude was applied
to the data base. The bottom track calibration was repeated after
the rotation was applied to check that amplitude and phase were 1 and 0,
respectively, which they were.
Under CODAS, the conversion of velocity profiles relative
to the ship to
estimates of absolute water velocity involves the intermediate calculation of
an absolute reference layer velocity. For the arguments behind it and
for details, see one of the ADCP data reports referenced under the SeaSoar
web page, e.g., under Japan/East Sea.
One of the advantages of this intermediate step is to
identify outlier GPS fixes
and other problems. Since we assume that the ocean velocity varies slowly,
any extreme velocity spikes would be due to outlier fixes.
Though rare these days, the initial, unedited reference layer velocities included
three cases pointing towards "bad" fixes, on decimal day 292 and 293. In each
case, the problem was traced back to an obviously erroneous longitude:
-20.1224692 6 1 1 16.95
% 292.3622743 -85.0916575 -20.1225892 6 1 1 17.05
292.3657350 -85.1675717 -20.1216333 6 1 1 23.35
-20.1234108 6 1 1 25.35
% 293.1643576 -85.0916767 -20.1234183 6 1 1 22.45
293.1678183 -85.1666350 -20.1234850 6 1 1 12.35
-20.1226108 6 1 1 16.45
% 293.4629688 -85.0916608 -20.1228950 6 1 1 16.15
293.4664294 -85.1666208 -20.1228025 6 1 2 16.55
( the % sign indicates that the respective line is commented out.)
In addition, the very first profiles from leaving port
had obvious problems. It
is likely that these were caused by bottom interference, which has not been
edited out yet from this dataset. For this initial report, the first 3 GPS fixes were
commented out from the rb0108.ags file. This implies that these profiles are
available only as shear relative to the ship in the database.
The navigation routines were then re-run, and the corrected
Maps show average ADCP vectors within a 1/4 degree by
1/4 degree lon/lat grid
for a series of standard vertical averages:
Layer: 21m to 25m
Layer: 25m to 75m
Layer: 75m to 125m
Layer: 125m to 175m
Layer: 175m to 225m
Layer: 225m to 275m
Layer: 275m to 325m
Layer: 325m to 375m
Although the above links are to jpeg files of relatively
the original postscript files "layer#.ps" are available on the cd in the
directory "vector", together with the input data (rb0108.vec) organized
as lon, lat, u(level1), v(level1), u(level2), v(level2), ...
5) Data in matlab format
A file in matlab format called "rb0108.mat" located in
"edit" subdirectory contains the complete database. It was
generated by combining the CODAS database and the estimate of
ship speed (contained in file nav/rb0108.sm). Its contents are
(output of matlab command "whos"):
Name Size Bytes Class
400 double array
txy 4508x3 108192 double array
u 4508x50 1803200 double array
v 4508x50 1803200 double array
Thus there are 4508 profiles with 50 vertical bins each.
The variable txy contains time, longitude, and latitude;
time is in "decimal days", with noon of January 1 corresponding
to decimal day 0.5 (not 1.5 as in Julian days).
6) Color contour plot of on-station data.
The ship spend roughly 6 days on station, presumably for
mooring work. These data were plotted as a time series for
zonal and meridional velocity components. The respective
postscript (and larger jpeg) files are located in the "edit"
subdirectoty, as is the simple script "p_uv_pcolor.m that
was used to generate the plot from the matlab-format data.