Education Center | Photo Gallery

Data Advanced Data Search More Data Discovery & Access Easy-to-use Data Products Data Collections Data Software & Programming Tools Data Submission Data News Centers & Programs Antarctic Glaciological Data Center Arctic System Science Data (ARCSS) IARC Frozen Ground Data Center International Polar Year Data & Information Service (IPYDIS) NOAA at NSIDC NASA's Distributed Active Archive Center at NSIDC (NSIDC DAAC) U.S. Antarctic Data Coordination Center World Data Center for Glaciology, Boulder Science Scientists Research Projects Publications NSIDC Notes Glaciological Data Series Special Reports Posters & Presentations Annual Reports Education Resources News & Events Press Room NSIDC In the News Data News Events About Expertise Data & Services NSIDC Sponsors Contacts Jobs Related Organizations

SHEBA Ocean Turbulence Mast Data Archive

Summary

This data set includes a time series of temperature, conductivity, salinity, three-dimensional velocity, pressure, and magnetic heading at multiple levels in the boundary layer under the drifting ice floe. As part of the Surface Heat Budget of the Arctic Ocean (SHEBA) project, data were collected continuously for the approximately year-long experiment. However, the experiment was divided roughly in half when a severe deformation event occurred in March 1998. After this event, the ocean camp was shifted to a new site farther into the SHEBA floe.

In the first half of the experiment (October 1997 through March 1998), there were four turbulence instrument clusters (TIC) that included three mechanical rotor current meters, which sensed the mean flow velocity. These were oriented approximately at 4 m, 8 m, 12 m, and 16 m below the ice-ocean interface in the western Arctic Ocean. For the second half of the experiment (April through September 1998), only two TICs were deployed on the mast. The clusters measured mean quantities, turbulent momentum, and heat fluxes with reasonable approximations to salinity flux.

Data are distributed on a DVD.

Citing These Data

McPhee, Miles G. 2006. SHEBA ocean turbulence mast data archive. Boulder, Colorado USA: National Snow and Ice Data Center. Digital media.

Overview Table

Table of Contents

1. Contacts and Acknowledgments
2. Detailed Data Description
3. Data Access and Tools
4. Data Acquisition and Processing
5. References and Related Publications
6. Document Information

1. Contacts and Acknowledgments

Investigator

Miles McPhee
McPhee Research Company
450 Clover Spring Rd
Naches, WA 98937

Technical Contact

Acknowledgements

This research was funded by the National Science Foundation (NSF) Office of Polar Programs (OPP) grants 0084269 and 0084275 to M. G. McPhee.

2. Detailed Data Description

Format

Data are in MATLAB format (.mat and .m).

File and Directory Structure

The master directory, data_submission/, contains 3 subdirectories: conv_data_structures/, realizations/, and mfiles/.

conv_data_structures/

Each file contains an array of MATLAB data structures and corresponds in numbering to the original binary files as transmitted from the SeaBird deck box to the recording computer. The prefix A or B refers to the SBE 9+ underwater units used to collect the data. There were two SBE 9 instruments with slightly different characteristics and output data streams. Instrument A was used early in the project and then again beginning in late July 1998 when a bulkhead connector on Instrument B failed.

realizations/

The data files in this directory follow the Fmon.mat naming convention where mon is the first three letters of the month the data were collected (for example, jan).

Fapr.mat
Faug.mat
Fdec.mat
Ffeb.mat
Fjan.mat
Fjul.mat
Fjun.mat
Fmar.mat
Fmay.mat
Fnov.mat
Foct.mat
Fsep.mat

mfiles/

A_to_one_cluster.m
adjust_F.m
B_to_one_cluster.m
COEFSTRUCT.mat
COMP_COR.mat
convert_all_files.m
convertfile.m
flux_allmonths.m
get_coef_set.m
mfix_freqv.m
PCOR.mat
turb_realize.m

More information about the content of these files is provided in the Processing Steps section.

File Size

File sizes range from 1 KB to 161 MB.

Spatial Coverage

Southernmost Latitude: 74° N
Northernmost Latitude: 81° N
Westernmost Longitude: 170° W
Easternmost Longitude: 140° W

Spatial Resolution

Depending on velocity, samples were take at 1 Hz or 0.5 Hz intervals.

Temporal Coverage

Data were collected from 9 October 1997 through 27 September 1998.

Temporal Resolution

Depending on velocity, samples were take at 1 Hz or 0.5 Hz intervals.

Parameter or Variable

Parameter Description

Temperature, conductivity, salinity, three-dimensional velocity, pressure, and magnetic heading data were collected.

Sample Data Record

The sample data is found in the A900C.mat file. Data include but are not limited to temperature time series (T), conductivity (C), three-dimensional velocity (um, vm, wm), depth, and a compass reading (brg).

Quality Assessment

Some files with obvious problems or short durations are not included (for example, A001 to A006).

3. Data Access and Tools

Data Access

Data are distributed on a DVD. Order data.

Volume

The entire data set is 3 GB.

Software and Tools

Data files are viewable in MATLAB.

Related Data Collections

• SHEBA Reconnaissance Imagery

4. Data Acquisition and Processing

Data Acquisition Methods

The Ocean Turbulence Mast project during SHEBA maintained clusters of instruments designed to measure turbulent fluxes of momentum, heat, and salt at multiple levels in the boundary layer under the drifting ice floe. Each turbulence instrument cluster (TIC) included three mechanical rotor current meters (Smith partially ducted rotor design) arranged as an orthogonal triad, tilted with respect to horizontal so that all rotors sensed roughly equal parts of the mean flow velocity. The current meter triad was mounted near to Sea-Bird Electronics SBE03 temperature sensors and SBE04 conductivity sensors, except for one cluster that included a SBE07 microstructure open electrode conductivity sensor. The clusters were mounted on stainless steel rods that made up a rigid mast. Included on the mast was a highly modified SBE09 conductivity, temperature, and depth (CTD) unit equipped with a compass and pressure sensor. Frequency data from each cluster was collected by the SBE09 and multiplexed to a surface SBE11 CTD deck unit, interfaced to a computer, at a sampling rate of 6 per second. The clusters were spaced 4 m apart on the mast, which could be lowered as a unit to any level in the upper 100 m of the ocean. Typically, the uppermost cluster was situated about 4 m below the ice-ocean interface under the main SHEBA floe, which was about 2 m thick.

The experiment was divided roughly in half, with two different locations forced by a severe deformation event in March 1998. The original, Site 1, was situated about 300 m off the right-hand side of the Canadian Coast Guard Ship (CCGS) Des Groseilliers. After the breakup in March, the ocean camp was shifted to a new site farther into the SHEBA floe. In anticipation of summer biofouling requiring pulling the mast for cleaning and inevitable equipment loss (after spending months in salt water), the number of clusters was reduced from four to two for Site 2. The two clusters were still spaced 4 m apart, with the mast raised later in the summer so that the uppermost cluster was nominally 2 m below the ice-ocean interface. The reduction freed up a frequency channel for the micro-conductivity instrument, which was then added to Cluster 2 so that it had both types of conductivity sensors.

Windows XP was used for MATLAB archival preparation.

Processing Steps

The following steps were used to reduce the data to usable form.

Step 1: The binary files were decoded into frequencies for each channel (binary numbers for the compass data) as an intermediate step using a MATLAB function, fetch_raw_file.m. The original binary files and corresponding frequency files are not included here, but can be made available upon request.

Step 2 (conv_data_structures/): The function convert_file.m (all the MATLAB functions are in mfile/) converts the frequency and binary data into engineering units using the calibration values supplied by SeaBird Electronics, Inc. The function A_to_one_cluster.m and B_to_one_cluster.m perform the main work of converting the frequency data to engineering units. The result for each file is an array of data structures. The following is an example for file A900C:

>> load A900C
>> whos
Name   Size   Bytes       Class
B         2x1    7543844   struct array
Grand total is 942520 elements using 7543844 bytes
>> B(1)
hdr: [1x1 struct]
cluster: 1
T: [62820x1 double] temperature time series for cluster #1 (°C)
C: [62820x1 double] conductivity (S/m*10)

For Site 2, Cluster 2 this will be [n x 2] where the second data vector is micro-conductivity (µC).

um: [62820x1 double] um is nominally 45° down from horizontal (cm/s)
vm: [62820x1 double] vm is horizontal
wm: [62820x1 double] wm is nominally 45° up from horizontal

um, vm, wm form a right-handed coordinate system with the xm, zm plane oriented 45° clockwise from the mast direction (as indicated by the compass)

depth: [62820x1 double] depth from surface will depend on atmospheric pressure (m)
brg: [62820x1 double] compass reading, requires later adjustment for actual magnetic heading (°)
>> B(1).hdr
ans = filenumber: 900
comment: sheba900 19:55 ADT 09/18/98, shipwind: 126 @ 21kts
interface_flag: 1
scan_average: 6 number of samples averaged before recording.

The SBE9s samples at a rate of 6 per second, so scan_average: 6 gives the one second average samples.

no_clusters: 3

For Site 2, there were only two clusters spaced 4 m apart, but Cluster 2 included both a standard SBE04 conductivity sensor and a SBE07 microstructure conductivity instrument. Cluster 3 substitutes the micro-conductivity (µC) for conductivity (C), otherwise it is the same as Cluster 2. For Site 1, October 1997 through March 1998, Cluster 1 was equipped with the micro-conductivity sensor only.

record_size: 73
no_records: 62893
Tsn: [7x1 double] serial numbers for SBE calibrations
Csn: [7x1 double] serial numbers for SBE calibrations
bearing: [7x1 double] N/A
depth: [7x1 double] depth offset for each cluster
start_time: [262.1682 1998 1.0043] decimal year day (noon Universal Time (UT) on 1 Jan=1.5), year, seconds per sample based on the computer clock start and end times (use 1 in this example)

These files comprise the basic turbulence mast data set. Some files with obvious problems or short durations are not included (e.g. A001 to A006).

Step 3 (realizations/): The usual method for estimating turbulent fluxes assumes the "frozen field hypothesis" and calculates the covariance of the velocity components (and scalar deviatory fields with the vertical velocity). This involves several steps in analysis of the data from the "raw" calibrated files. The basic unit in this phase is a flow "realization" for a duration of 15 minutes. The data files from Step 2 are divided into 15 minute segments (900 samples for scan_average=6 and 450 samples for scan_average=12). The routine turb_realize.m divides the data file for each cluster (e.g. A900C.mat, Cluster 1) into 15 minute segments. Then for each segment:

• Rotates the velocity from the measurement reference frame into a streamline reference frame so the mean flow is aligned with the u direction, w is vertical and v is the cross-stream component so as to form a right-handed system. It is assumed that the vertical velocity over the 15 minute period is zero, so that <w> and <v> are zero; <u> is the mean current speed. See the subfunction u_streamline. In any mechanical current measuring system, there is a threshold velocity below which the rotors stop turning. For each velocity component, the routine determines the number of samples for which the velocity is nonzero and forms a ratio with the total number of samples, r{u,v,w}. It then determines a quality flag as follows:
  flg=0:ru < 0.9 | rw < 0.9 no valid vertical fluxes
  flg=1:ru > 0.9 & rw > 0.9 & rv > 0.9 three-dimensional currents considered valid
  flg=2:ru > 0.9 & rw > 0.9 & rv < 0.9 in this case the w is ok, but terms with v are not, so for example, <w'T'> would be valid. Any segments with flg=0 are dropped from further consideration.
• Early in the project (October through early November 1997), data from the upper two clusters on the ocean turbulence mast were occasionally susceptible to electronic noise in the temperature (T) and conductivity (C) frequencies, apparently generated by the neighboring ocean profiling winch. The routine uses a histogram-based algorithm to remove this noise (subfunction remove_noise3); however, it is not always successful, and caution should be exercised in interpreting fluxes derived from these data.
• Calculates average values of temperature (T), salinity (S), speed, current heading (magnetic), depth, and θ, which is the angle through which the measurement frame is rotated to bring the vertical velocity to zero. This should be within a few degrees of 45, and serves a later quality check.
• Calculates variance and covariance quantities: <u'u'>, <v'v'>, <w'w'>, <u'w'>, <v'w'>, <T'T'>, <w'T'>, <S'S'>, and <w'S'>. It should be noted that S' quantities are somewhat suspect, since the ducted design of the SBE04 conductivity sensors tend to cause significant flow disturbance at higher frequencies. The result seems to be to underestimate the actual covariance with w by about 25%. The micro-conductivity instrument (SBE07) has a better high frequency response, but the records tend to be spiky with sudden significant shifts. It may be necessary to visually inspect every time series from the micro-conductivity instrument before accepting the covariance statistics as valid flux estimates. This has not been done here.
• Produce a n x 5 matrix of five data vectors: T, S, u, v, w, where n is the number of samples in the 15-minute period. The data are cross-referenced to the original file name and the beginning and ending indices in the .hdr structure.

The routine adjust_F.m operates on the structure produced by turb_realize.m to correct several known data defects, then it adjusts the compass reading to magnetic heading, and finally corrects the depth (pressure) reading for changes in atmospheric pressure. Note that the temperature and conductivity values are based on the pre- and post-cruise calibration values provided by SBE. With the slowly changing properties of the Arctic's well-mixed layer, we were able to cross calibrate the various sensors with the SBE9+ profiler to a much higher relative accuracy than was provided by the manufacturer’s calibrations. The following table is supplied as a courtesy only, and should be verified/revised by an investigator interested in minute T/S differences. These corrections have not been applied to the data in the Fmon.mat files.

The routine flux_allmonths.m arranges the 15-minute records in structure arrays month by month. So for example, the Fsep.mat file in realizations/ contains all of the 15-minute structures for the month of September 1998 for each cluster (where Cluster 3 is the same as Cluster 2 except with salinity calculated from the micro-conductivity). For someone unfamiliar with MATLAB structured arrays (akin to C+ structures), the following exercise from mfiles/ may help clarify their role:

>> load ../realizations/Fsep
>> whos
Name   Size       Bytes        Class
F1        823x1    20655468   struct array
F2        1646x1  40352328   struct array
F3        1646x1  40352328   struct array
Grand total is 11815420 elements using 101360124 bytes

>> f=F2(1000);
>> f
f =
hdr: [1x1 struct]
cluster: 2
time: 262.4546
flg: 1
dmn: [-1.6011 29.8015 0.2754 1.1935 7.3390 0.8186] T, S, speed (m/s), magnetic heading (rad), depth, θ
trb: [1x9 double] <u'u'>, <v'v'>, <w'w'>, <u'w'>, <v'w'>, <T'T'>, <w'T'>, <S'S'>, <w'S'>
data: [900x5 double] T, S, u, v, w [velocity units m/s]

>> f.hdr
ans =
filenumber: 900
comment: [1x51 char]
interface_flag: 1
scan_average: 6
no_clusters: 3
record_size: 73
no_records: 62893
Tsn: [7x1 double]
Csn: [7x1 double]
bearing: [7x1 double]
depth: [7x1 double]
start_time: [262.1682 1998 1.0043]
filename: 'A900C'
index_orig: [24301 25200]
>> load ../conv_data_structures/A900C
>> idx=f.hdr.index_orig(1):f.hdr.index_orig(2);
>> subplot(211);plot(f.data(:,3:5));grid on
>> uorig=[B(2).um(idx) B(2).vm(idx) B(2).wm(idx)];
>> subplot(212);plot(uorig);grid on
>> xlabel('Sample');ylabel('cm s^{-1}');subplot(211);ylabel('m s^{-1}');

5. References and Related Publications

McPhee, M. G. 2002. Turbulent stress at the ice/ocean interface and bottom surface hydraulic roughness during the SHEBA drift. Journal of Geophysical Research. 107(C10), 8037, doi: 10.1029/2000JC000633.

McPhee, M. G. 2001. Is ocean heat flux enhanced under rapidly growing ice? Proceedings of the Sixth Conference on Polar Meteorology and Oceanography. American Meteorological Society. 14-18 May 2001. San Diego. 267-270.

McPhee, M. G., T. Kikucki, J. H. Morison, and T. P. Stanton. 2003. Ocean-to-ice heat flux at the North Pole environmental observatory. Geophysical Research Letters. 30(24). 2274. doi: 10.1029/2003GL018580.

McPhee, M. G. 2004. A spectral technique for estimating turbulent stress, scalar flux magnitude, and eddy viscosity in the ocean boundary layer under pack ice. Journal Physical Oceanography. In press.

6. Document Information

List of Acronyms

The following acronyms are used in this document:

CCGS: Canadian Coast Guard Ship
CTD: Conductivity, temperature, and depth
MATLAB: Matrix Laboratory
NSF: National Science Foundation
NSIDC: National Snow and Ice Data Center
OPP: Office of Polar Programs
SBE: SeaBird Electronics
SHEBA: Surface Heat Budget of the Arctic Ocean
TIC: Turbulence Instrument Clusters
URL: Uniform Resource Locator
UT: Universal Time

Document Creation Date

May 17, 2006

Document URL

http://nsidc.org/data/docs/arcss/arcss142/index.html