NODC Archives Shipboard ADCP

(no figures)


     For the past decade, acoustic Doppler current profilers (ADCPs) have become steadily
more  common aboard the UNOLS, NOAA, and Navy fleets.  During the late 1980s, the data
quality was limited by lack of continuous Global Positioning System (GPS) coverage and
uncertainties in the ship's heading information.  However, for the past several years, the quality
has improved significantly due to the 24-hour GPS coverage, differential gps techniques, and the
advent of GPS heading sensors.  With reliable heading and navigation data , absolute currents
in the upper ocean are determined.  The data provide fine resolution in time (~5 minutes), depth
(~10 m), and horizontal distance (~2 km) throughout the duration of a cruise.  The accumulating
database allows a fresh view of upper ocean velocity structure on a variety of temporal and
spatial scales.  The National Oceanographic Data Center (NODC) has been working for several
years on a management scheme for this important new data set and is now ready to share the
plan with the scientific community and solicit contributions to the shipboard ADCP archive. 
Below, a brief review of the methodology for acquiring the absolute currents is given and
NODC's data management plan is discussed.

Methodology of data acquisition

     The hull-mounted ADCP  estimates horizontal and vertical velocity as a function of depth 
by using the Doppler effect to measure the radial relative velocity between the instrument and
scatterers in the ocean.  Three acoustic beams in different directions are the minimal requirement
for measuring the three velocity components.  A fourth beam adds redundancy and an error
estimate.  The ADCP transmits a ping from each transducer element roughly once per second. 
The echo arrives back at the instrument over an extended period, with echos from shallow depths
arriving sooner than ones from greater ranges.  Profiles are produced by range-gating the echo
signal, which means the echo is broken into successive segments called depth bins which
correspond to successively deeper depth ranges.  The operator configures the length of each depth
bin and the transmit pulse, which determines the degree of averaging in the vertical, depending
on whether one is interested more in vertical resolution or profile penetration. The noisy velocity
estimates from each ping are vector-averaged into 1- to 10-minute ensembles.  The relative
velocities are rotated from the transducer's to the earth's reference frame using the ship's
gyrocompass.  Finally, relative velocities and various ancillary parameters are stored on the ship
using a data acquisition system (DAS) which also optionally records navigation information, such
as provided by the GPS.  Specifics of the instrument capabilities and configuration options are
well documented (RDI, 1989).

     Routine processing, quality control, and calibration are performed at the host institute. 
Standard checks include detecting and correcting time errors, applying transducer-level
temperatures and salinities to obtain a better estimate of the sound speed for the velocity
calculation, editing out bad bins or profiles that have been contaminated by interference with the
bottom or some other physical object such as a hydro wire, and verifying the quality of the
gyrocompass and the navigation data.  The final gyrocompass estimates of ship heading and the
navigation information are the primary sources for calibrating the ADCP's relative current
velocities.  Typically, one is correcting for a "angle" error due to misalignment of the transducer
relative to the ship's hull and an "amplitude" component related mostly to minor imperfections
of the transducer geometry.  Relative current velocity errors caused by these components are
orthogonal; the angle errors lead to uncertainties of the athwartships velocity component while
the amplitude error introduces uncertainties along the ship track.  

      The navigation calculation is performed once calibration is complete.   Absolute currents
over a fixed depth range (reference layer) are obtained by subtracting the average of the ship
velocity relative to a reference layer (i.e. ADCP velocities) from the absolute ship velocity over
the ground (from navigation, i.e., GPS).  The raw absolute current velocities relative to the
reference layer are smoothed to reduce the effects of noise in the position fixes and combined
with the navigation data to obtain the  best estimates of ship positions and velocities, which are
stored into the data base.  Thus, absolute currents at any depth can be determined from the final
ship navigation data and the relative ADCP measurements.
     

NODC's archive plan

     A group of data producers and users of shipboard ADCP met with data management
experts at NODC in May 1992 to discuss the tasks at hand (Firing, 1992).  The meeting was
convened by Dr. Eric Firing, a professor of oceanography at the University of Hawaii (UH) who 
is a long standing expert in shipboard ADCP collection, processing, and analysis.  By this point
in time, many of the difficulties in calibrating and obtaining absolute currents had been overcome
and the number of scientific publications using this data was steadily rising.  With the sharp
increase in installation and attention to the shipboard ADCP in the early 1990s, it became clear
that a data management plan was pertinent for centralizing the data set into a well-documented,
quality-assured archive and for allowing easy access to the scientific community.

     Shortly thereafter, the NODC liaison assigned to the TOGA Sea Level Center at UH
began collaborating on a part-time basis with Dr. Firing,  the NODC data managers, and other
ADCP experts in the development of an archive strategy.  The primary logistical problem was
how to effectively handle the high-density data set consisting of currents and ancillary parameters
at the sampling interval with which the data were recorded and processed.  It is not merely the
volume of data collected on a typical month-long cruise (about 10 Mbytes) that makes this data
set complex, but rather the cruise-to-cruise variability (as well as the intra-cruise variability) of
the sampling rates and types of ancillary parameters. These parameters include the date-time-
group,  transducer temperature (and salinity), a variety of diagnostic values, heading information,
and navigational data.   Moreover, a method of flagging bad values and denoting the depth
penetration of reliable data was needed.  It was obvious that the traditional flat ASCII file
approach was inadequate and the use of a sophisticated processing and data management system
was required to facilitate fast, efficient access to the data.

      A software package called the Common Oceanographic Data Analysis System (CODAS),
designed, documented, and maintained by Dr. Eric Firing and associates at UH, became the focus
of attention.  The system has been used at UH since 1988 and has been  distributed to over 30
agencies in 12 countries.  In addition to the processing tools,  this readily available public domain
software provides easy access to the data with a variety of options for averaging, regridding, and
selecting only data that meet specified quality criteria. CODAS stores arrays of flags 
corresponding to the velocity arrays; thus, the original data are not altered by editing.  CODAS
is a hierarchial data base which uses a "directory file" to keep track of binary "block files."  The
system  was written in standard C language and the package is  primarily used on workstations
and IBM-compatible PCs.   For the binary block files, the software provides translation between
machines with different binary numbering conventions and offers  a complete ASCII dump.

      Because of its flexibility, the NODC decided to adopt the CODAS system for the
archival of the high-resolution data set.  This move advances NODC's  goal to be not only an
archive center, but to maximize ease of access for the scientific community using the most up-to-
date technology.  The NODC now archives the high-density shipboard ADCP data as CODAS
files and a standard subset of each cruise at hourly and 10 m depth intervals as ASCII files.  The
CODAS files include current velocities and all ancillary data while the subset includes only the
absolute current velocities, transducer temperature, and ship velocity. For analysis purposes, the
standard subset is best suited for synoptic and climatological research and the high-density set
is valuable for fine-scale studies.

     The NODC has established the Shipboard ADCP Center (SAC) at UH for the acquisition,
review, documentation, archival, and distribution of shipboard ADCP data sets (Figure 1).   The
activities are overseen by the NODC liaison and the locality takes advantage of close proximity
to the ADCP and CODAS experts (Dr. Firing and associates).  A network of SUN workstations
maintain the archive online and facilitate the archiving steps.  Data producers are encouraged to
contribute the high-density data sets that have passed the quality control, calibration, and
navigation stages.  Metadata (information about the data) are vital for the archive.  The SAC can
provide a guideline for the type of metadata desired.  The incoming data sets are placed in the
CODAS format if necessary, reviewed, reduced to a standard subset,  documented, and backed
up.  The data producers will be contacted if suspect features are identified or if additional
metadata are required.  The data sets will be periodically passed on to the NODC headquarters
which will act as the final repository, assist in advertisement of data availability and in the
encouragement for submissions from producers, and prepare CDROMs for easy distribution of
large volumes of the high-density data sets. 

     The growing archive now contains 77 cruises including a majority of the TOGA COARE
legs, several WOCE lines,  all of the WEPOCS sets, and about 75% of the HOTs cruises.  Cruise
tracks of the present data base are shown in Figure 2.  The archive is being filled first primarily
with readily available data sets at UH and other institutes that have passed scientific scrutiny, are
well documented, and are already in the CODAS format.  The data base will expand rapidly over
the coming years.  RD Instruments of San Diego, the dominant source for instruments, has sold
~125 vessel-mounted units to date.  If a third of those are active for 10 months a year, that
represents ~400 cruise-months per year of potential data.

     A browse, inventory, and retrieval system (BIRS) has been developed by the SAC for
facilitating access to the global database.  Presently, the system is operated solely by the SAC
staff, but could easily be placed in a public domain location for anonymous users to query to
database and make requests over the Internet.  For the interim, please send data requests directly
to the SAC.

     For requests of the high-density data set,  the CODAS block files and software if
necessary, or the ASCII dump are available.
Requests for the standard subset (~500 Kbytes per cruise-month), metadata, inventories, cruise
tracks, or general information can be handled through electronic or paper mail.  An anonymous
ftp account is available at the SAC for Internet users.  Questions, comments, data submissions,
data requests, and CODAS software and manual requests should be directed to:

   
     Mr. Patrick Caldwell
     NODC Hawaii Liaison Officer
     Shipboard ADCP Center
     Dept. of Oceanography
     University of Hawaii, MSB 317
     Honolulu, Hawaii 96822
     USA

     Internet: caldwell@soest.hawaii.edu
     fax:  808-956-4104
     office phone: 808-956-4105

 

Firing, E., 1992. Notes from acoustic Doppler Current Profiler Workshop at the National
     Oceanographic Data Center, May 14-15, 1992.  "unpublished manuscript."
     Copies available from Mr. P. Caldwell


RD Instruments, 1989. Acoustic Doppler Current Profilers Principles of Operation: A Practical
     Primer.  Available from RD Instruments, 9855 Businesspark Av., San Diego, CA 92131