Ice Accelerometer Data
Collected During Leg-II of the Polarbjorn Drift
October - November 1988
Coordinated Eastern Arctic Experiment (CEAREX)
Seelye Martin and Robert Drucker
School of Oceanography WB-10
University of Washington
Seattle, WA 98195 USA
1. Description of Experiment
In October-December of 1988, we participated in the winter drift phase
of the Coordinated Eastern Arctic Experiment (CEAREX) aboard the ship
POLARBJORN, which was frozen into the sea ice north and east of Svalbard.
The ship was moored between two multi-year ice floes, a kilometer-scale floe
(Alpha) and a 100 meter-scale floe (Beta). These floes in turn were
surrounded by primarily multi-year ice floes. During the drift, we deployed
three ice accelerometer buoys at the sites which we call Alpha-1, Alpha-2,
and Beta, where the buoys are located on their named floes. The purpose of
these buoys was to measure the ice accelerations associated with deformation
and breakup. We also measured the relative compass heading of each buoy.
The details of the buoy deployment and of their locations relative to the
ship and the floes are given in "Observations of ice floe collisions during
Leg-II of the POLARBJORN drift" (Martin, S. and R. Drucker, 1991, Journal of
Geophysical Research, 96(C6), p.10567-10580.)
Each buoy contained a compass and a set of three orthogonally-mounted
accelerometers, with two accelerometers in the horizontal plane and one in
the vertical. For each buoy, we will refer to the accelerometers as X, Y
and Z, where Z is vertical, and X and Y lie in the horizontal plane, but
have no particular orientation. At the beginning of each hour, each buoy
transmitted 840 seconds of acceleration data, followed by 60 seconds of
calibration and compass data. At the ship, the data were received,
digitally sampled at 10 Hz, and recorded. The compass heading was sampled
at hourly intervals and transmitted with the calibration data. For the
present experiment, the standard deviation of the quiescent acceleration
was about 0.5 mm s**-2, except for buoy Alpha-2, channel Y, which was
noisier than the others and had a standard deviation of about 1 mm s**-2.
The bandpass of the accelerometers was 0.04 - 5 Hz or 0.2 - 25 s.
In our conduct of the experiment, we arrived at the ship on 11 October
(Day 285) and deployed the three buoys by 22 October (Day 296). The hourly
transmitting and recording process, which was interrupted at approximately
weekly intervals to download data from the computer, continued until
25 November (Day 330). Between Day 296 and 330, the ship and buoys drifted
southwest in the broad passage between Svalbard and Franz Josef Land toward
Kvitoya Island, a small island east of Svalbard lying between the Arctic
Ocean and the Barents Sea. "CEAREX drift experiment", (Pritchard, R., and
twenty-eight others, 1990, Eos, Transactions of the American Geophysical
Union, 71(40), p.1115-1118) shows the drift track of the ship for this
period. Before Day 310, the ice drifted as a solid body, and we observed
no activity from our buoys. Following Day 310, we began to observe
accelerations in the ice cover, which occurred during discrete periods
between Days 310 and 329. On Day 327, floes Alpha and Beta broke up, and
the ice experienced its largest accelerations and rotations. This
continued until Day 329, when the buoys were apparently destroyed by
compressive forces. Because of the importance of these last twenty days
of the experiment, the paper and the attached records concentrate on the
ice behavior between Days 310 and 330.
2. Data Set
Our study showed that the buoy activity consisted of a noise floor and at
least four periods of enhanced activity. Most of the observed peaks occur in
the following four periods:
I. Days 312 - 314;
II. Days 317 - 321;
III. Days 322 - 324;
IV. Days 327 - 329.
We found that accelerations during Period I resulted from shearing of the
adjacent ice; those during Period II from a mixture of shearing and ridging,
and those during Period III from ridging. The largest period of activity
occurred during Period IV, at which time the floes broke up and experienced
large accelerations. Our measurements stopped at Day 329, hour 10, which was
our last coherent signal before the buoys were apparently pushed over and
destroyed by ice forces.
This data set contains 76 sampled hours of data from our experiment.
These hours occur between Days 308 and 330, and include most of the hours
from the four periods during which we observed significant accelerations
above the noise level. In particular, the data set contains all of the
hours which we analyzed in our paper. The other hours are included for
completeness and to provide several hours in which nothing happened as a
control.
3. Instrument Design
The design of the instruments is discussed in Martin and Becker, 1987,
1988 (Martin, S., and P. Becker: High frequency ice floe collisions in the
Greenland Sea during MIZEX-84. J. Geophys. Res., 92, 7071-7084, 1987, and
Martin, S., and P. Becker: Ice floe collisions and their relation to ice
deformation in the Bering Sea during February 1983. J. Geophys. Res., 93,
1303-1315, 1988.)
4. Data Organization and Format
4.1. Ice acceleration data
The ice acceleration data are organized into hourly files. Each file is
named according to its collection day and hour in the format `Addd.hh', where
`ddd' is the day of the year and `hh' is the hour in Universal Time (UT).
Data collection occurred for 15 minutes each hour beginning at 6 minutes
past the hour. Thus, for example, the file A328.01 was collected from
01:06 UT to 01:21 UT on day 328 (23 November 1988).
Each file consists of nine time series corresponding to the nine channels
(three buoys; three channels per buoy). There are 9000 samples per channel,
corresponding to a sampling rate of 10 samples per second for 900 seconds.
Approximately the first 8400 samples are acceleration data; the last minute
of each collection period consists of calibration signals and compass
readings. This period is easily distinguished visually from the data
period by its large pulse-like excursions.
The files are written as ASCII flat files of columnar number tables in
the directory \ACCEL on the CD-ROM "CEAREX-1". There are nine columns of
space-delimited numbers corresponding to the nine time series, and 9000 rows
corresponding to the 9000 samples per channel. The order of the channels
is as follows:
BETA X / Y / Z 1ALPHA X / Y / Z 2ALPHA X / Y / Z
The sample numbers are written as 5-digit decimal integers in a
7-character field. This provides for 1 sign character (minus sign or
space) and a minimum of one additional space between entries.
The decimal integers represent accelerations in units of 0.01 mm
s**-2. That is, the data may be read as mm s**-2 by dividing by 100. The
dynamic range is thus -999.99 to 999.99 mm s**-2. Sample values in excess
of this range are set to the limits (+/- 99999), but any such data should
be considered spurious.
There are no headers. This facilitates machine reading of the data and
simplifies importation into spreadsheets, graphics software, or other
software. HOWEVER, IT IS ESSENTIAL THAT THE FILE NAME NOT BE LOST AS IT
CONTAINS THE ONLY REFERENCE TO THE IDENTITY OF THE FILE. The filenames
are in the format "A308_09.ACC" where "308_09" represents "308.09", the
beginning year-day of the data. The decimal point in "308.09" had to
be changed to an underscore ( _ ) because of the constraint that filenames
on a CD-ROM must have a 3-character extension, in combination with the
data constraint that the filenames must remain meaningful.
4.2. Compass Data
In addition to the acceleration data, a single ASCII file `COMPASS.DAT'
contains compass readings for selected hours between days 300 and 330. The
compass readings were manually derived from the information in the 60-second
calibration cycle at the end of each hourly sample. Because of the laborious
nature of extracting them, compass readings are given only as needed; i.e.,
hourly readings are only provided during periods of rapid rotation. This
occurred primarily after day 327, when floes Alpha and Beta broke up into
small floes which rotated relative to one another.
The compass readings are in units of degrees with arbitrary orientation.
Because these numbers were derived manually, their accuracy is probably on
the order of +/- 5 to 10 degrees. Missing values are flagged as 999 in the
compass file.
5. Processing and Data Conversion
The acceleration data presented here were minimally processed in order
to allow the greatest flexibility of use. The only alteration of our raw
digital data was to remove means, filter out certain large errors due to
radio noise, and reformat binary data into ASCII. A brief description of
these processing steps follows.
The buoy acceleration signals were FM-telemetered to a ship-board
receiver on three audio-frequency FM subcarriers, individually
discriminated, digitally sampled, and stored in binary files in a 2-byte
format. The sampling resolution was approximately 6 bits per mm s**-2 and
varied slightly per channel. Calibration was provided by pre-deployment
measurement and by the 1-minute calibration cycle following each collection
period.
For the purposes of this CD-ROM it was decided that ASCII format would be
desirable because of its greater machine independence. To convert to ASCII,
we first read the binary data into memory arrays and computed and removed the
mean of each time series. This was justified because the buoy instruments
were AC-coupled and any remaining mean in the signal would be due to offsets
in the discriminators.
After removing the means, we passed the signals through a software
despiker which identified large single-point errors due probably to radio
noise. For this purpose we used a threshold of 5 mm s**-2 above or below
adjacent samples. Samples lying outside the thresholds were replaced by
the mean of their immediate neighbors. These are not flagged.
The 2-byte samples represented signed integer data with a range of -32768
to +32767. These were multiplied by gain constants that had been established
for each channel. These constants varied slightly by channel but were
approximately 0.16 mm s**-2 per bit. Thus, although the resolution of the
5-digit data representation is 0.01 mm s**-2, its true resolution is only
about 0.16 mm s**-2 due to the analog to digital conversion (quantization).
The resulting data in units of acceleration have a possible range of about
+/- 5000 mm s**-2. However, we did not experience any actual accelerations
above a few hundred mm s**-2. For this CD-ROM we therefore limited the
range to +/- 999.99 mm s**-2 in order to reduce data size; we then
multiplied each sample by 100 and rounded to integers to provide a 5-digit
integer with no decimal point. Thus the samples on this CD-ROM can be put
directly into acceleration units of mm s**-2 by dividing by 100.
6. References
Martin, S. and R. Drucker. (1991) Observations of ice floe collisions
during Leg-II of the POLARBJORN drift. Journal of Geophysical Research,
96(C6), p. 10567-10580.
Martin, S. and P. Becker (1988) Ice floe collisions and their relation
to ice deformation in the Bering Sea during February 1983. Journal of
Geophysical Research, 93(C2), p. 1303-1315.
Martin, S. and P. Becker (1987) High frequency ice floe collisions
in the Greenland Sea during MIXEX-84. Journal of Geophysical Research,
92(C7), p. 7071-7084.
7. Contact Information
We welcome any feedback, comments, or questions about this data set.
The hours included on this CD-ROM are a sampling of a much larger data set
and include hours that we have found to contain interesting or representative
data. We plan to archive the full data set on magnetic tape in the near
future. Any correspondence can be addressed to Dr. Seelye Martin or Robert
Drucker at:
School of Oceanography WB-10
University of Washington
Seattle, Washington 98195
telephone: (206) 543-6438
e-mail: robert@isbjorn.ocean.washington.edu.
8. Acknowledgments
We gratefully acknowledge the support of the Office of Naval Research
under contract N00014-87-K-0160. We thank the Captain and crew of the
Norwegian ship POLARBJORN for their help during the drift cruise, and Robert
Pritchard for his service as chief scientist, and Alan Gill and Jay Ardai for
their help with deployment.
November 1990 (rev. 8/91)
