Wild dolphins live in a world of sound. Of all forms of energy,
sound is the one form that propagates most effectively in
water, much better for example than light. Dolphins use this
to their full advantage and have excellent hearing and echolocation
capabilities, which they use to communicate with each other
and to probe their surroundings. Therefore, to better understand
wild dolphin behavior and ecology, an important first step
is is to accurately record and analyze the sounds they produce.
However, this has proven very difficult in the past.
One of the main problems with recording their sounds is the
high frequency content of dolphin signals. Most dolphin species
produce three types of signals: echolocation clicks, the faster
paced burst pulse clicks, and whistles. Echolocation clicks
are used by dolphins to detect and recognize objects in the
water from the returning echoes, akin to a biological (and
much superior) version of man-made SONAR. Burst pulse clicks
and whistles on the other hand are thought to be used mainly
for communication. Click signals are extremely short (50 microseconds)
and broadband high frequency signals, ranging from 0 to over
200 kiloHertz. Whistles are generally within human hearing
range, but also have ultrasonic components called harmonics
which can go up to over 100 kHz. However, conventional audio
recording systems only go up to about 20 kHz (also the upper
limit of human hearing), and therefore miss the majority of
energy of these signals.
Another major problem is that it is usually impossible to
tell which, or how many dolphins produced the recorded sounds.
This information is essential, however, to associate acoustic
features of the sounds with behaviors of individual dolphins,
or to find out how communication signals are exchanged among
them. Also, the distance of the sound producing dolphin, needed
to determine the amplitude of the sound, is usually unknown,
as well as the orientation of its head with respect to the
recording device. The reason it is important to know head
orientation is because dolphins emit sounds from their foreheads
in a directional “beam” of sound, somewhat comparable
to the light on a miner’s helmet. Only recordings from
the beam axis represent the actual signals, while signals
recorded off-axis can be highly distorted. Since recording
systems are usually operated from boats, it is hard to control
for any of these factors.
The Underwater Dolphin Data Aquisition System (UDDAS,
designed by Dr. Marc Lammers) and 4-channel
UDDAS (designed by Michiel Schotten) were developed at
OSI as tools to deal with each of these challenges. Their
main asset is that they are self-contained units and can be
taken under water by a diver to record the entire frequency
bandwidth of dolphin signals. With simultaneous video, behaviors
of the sound producing dolphins and their head orientations
can be recorded as well. Using the 4-channel UDDAS, one can
also localize recorded sounds to a 3-D position in the water.
This allows one to determine the amplitude of the sounds,
and to attribute them to individual dolphins on the video.
UDDAS
Specifications
Broadband
recording unit:
The heart of the recording unit is a single-board computer
by Aaeon electronics (model: GENE 4310) with a low-power 233
MHz processor. It conforms to a PC-104 format. Stacked onto
the computer are a PCMCIA module, a hard drive module and
a power supply module, all connected via the PC-104 bus. The
PCMCIA module holds an A/D board by Measurement Computing
(model: DAS16/330) capable of sampling up to 330 kHz. The
hard drive module carries a 6 GB 2.5" Toshiba hard drive.
The computer operates on Windows 98 and the data acquisition
software is written in LabView. The acquisition program waits
for a user "TTL-high" trigger to begin recording.
Typically, data is acquired at a sampling rate of 220 kHz
(110 kHz Nyquist) in 10-second blocks.
A custom-built directional transducer made of a piezocomposite
material receives incoming signals. It has a flat sensitivity
of approximately -180 dB re 1 uPa up to at least 150 kHz.
A signal-conditioning unit in the housing is mounted below
the computer stack to amplify and filter incoming sounds before
recording. Signals are amplified by 40 dB with a user option
to amplify an additional 10 dB. A 5th order low-pass filter
by Lattice Semiconductors (model: ISPpac-80) is used as an
anti-aliasing filter. It is programmed to have a cutoff frequency
of 100 kHz.
Video camera unit:
A Sony digital video camera (model: DCR-PC110) is used to
record simultaneous behavioral data. The camera is housed
in a customized housing manufactured by Ocean Images. A trigger
switch near the left handle connects via a cable to the electronics
housing mounted below the camera housing. An omnidirectional
hydrophone (built by Don Norris) with a sensitivity of - 205
dB provides acoustic input for the video camera. Inside the
camera housing, a custom-built pre-amp provides 20 dB of gain
before the signal is fed into the camera.
4-channel UDDAS
Specifications
The
4-channel UDDAS is a digital acoustic/video recording system
that is battery powered and packaged in an underwater housing.
The housing, custom-made by The
Sexton Company LLC, holds a Samsung digital camcorder
(model: SCD103) with wide-angle lens, 2 sealed lead-acid 12V/
8.5 AmpHr batteries, and a custom-developed 4-channel acoustic
recording unit. Attached to the housing is an array of 4 omni-directional
hydrophones in the shape of the letter Y, with 1 hydrophone
in the middle (underneath the dome for the camcorder lens)
and 3 hydrophones sticking out at the ends of 0.5 m long,
hollow delrin pipes spaced 120º apart. The spherical
hydrophones (custom-built by Don Norris) have a sensitivity
of –214 dB re 1V/uPa and a flat frequency response up
to 300 kHz. Positions of sound emitting dolphins are determined
with the 4-hydrophone array localization technique. Range,
horizontal angle, and vertical angle are calculated from the
differences in arrival time of the sound at each of the 4
hydrophones, which gives the 3-D position of the dolphin.
Localizations are accurate up to a range of at least 15 m
(50 ft) away.
Incoming dolphin signals on the 4 channels are amplified
and filtered by programmable ispPAC80 filter chips (donated
by Lattice Semiconductor)
on a custom-made signal conditioning unit. The amount of amplification
is user controlled, with a choice of 20, 40, or 52 dB re 1
uPa through a switch outside the housing. In the ispPAC80
chips, a 5th order low-pass filter with a programmed cut-off
frequency of 220 kHz acts as an anti-aliasing filter. After
signal conditioning the signals are fed into the DAQ-2010,
a small 4-channel simultaneous analog-to-digital conversion
board from ADLINK Technology,
which digitizes the signals at a sampling rate of 667 kHz
(333 kHz Nyquist frequency) on 4 channels. The PCI data acquisition
board, which has been generously donated by Adlink, is stacked
onto a PC/104-plus computer stack with a PCI-to-PC/104-plus
adapter board (donated by Douglas
Electronics). PC/104-plus is the ideal format because
of its small size and the compact way the computer modules
stack together. PC/104-plus was chosen over PC/104 because
it conforms to the 33 MHz PCI data bandwidth, which is needed
to run the DAQ-2010 and to handle the extremely high data
throughput of 5.4 MB/s (i.e. 4 channels simultaneously sampling
at 667 kHz with 14-bit = 2-byte resolution).
After digitizing, data are transferred through the scatter-gather
DMA function of ADLINK’s DAQ-2010 to the onboard memory
of the MOPS/lcd7. This is a PC/104-plus single-board computer,
with a fanless 300 MHz Celeron processor and 512 MB of SDRAM
memory, and has been donated to this project by Kontron
America. Subsequently, data are written from the onboard
memory onto the Hitachi Travelstar, a 2.5” notebook
hard disk of 60 GB and 7,200 rpm. The computer operates on
Windows 2000 and a hand-written LabView program drives the
data acquisition and data storage process. The program waits
for a TTL trigger to begin recording, which is user controlled
from outside the housing. The entire system runs off two 12V
batteries, converted from +24V to +5V by a PC/104-plus power
supply module, which has been custom-adapted and donated by
Sanmina-SCI. The
PCI-to-PC/104-plus adapter, PC/104-plus power supply, and
single-board computer are all stacked on top of each other
and connected via the PC/104-plus bus.
Completion
of the 4-channel UDDAS was possible because of the Oregon
based Sexton Company, one of the very few manufacturers in
the world building custom underwater housings for scientific
research. The scientific expertise and many years of experience
of its director, Ken Sexton, as well as his top-of-the line
3-D designing software and manufacturing equipment, guaranteed
the best possible housing for this project. Ken came up with
many great ideas for the final design of the 4-channel UDDAS:
all parts were mounted on a tray, which can be taken out of
the housing entirely for easy access; a semi-spherical dome
in front of the housing accommodates the wide-angle lens of
the video camcorder; a Noren heat pipe transfers heat from
the heatsink on the MOPS/lcd7 to a heatsink on the rear outside
the housing, to cool the computer processor to the sea water;
a tv-monitor (model: Liliput) connected to the camcorder was
mounted in the back of the housing, so that a diver is able
to see what is being recorded. There are controls for camcorder
record, zoom, on/off, start acoustic recording, and amount
of amplification. The system is slightly positively buoyant
and easy to swim with.
Fieldwork
The 4-channel UDDAS is a welcome new tool to study the acoustic
behavior of Hawaii’s resident spinner dolphins (Stenella
longirostris) in OSI’s Nai'a Kuwili Project. Additionally,
in an ongoing collaboration with Dr. Denise Herzing of the
Wild Dolphin
Project, the 4-channel UDDAS is also being used to study
wild Atlantic spotted dolphins (Stenella frontalis)
and bottlenose dolphins (Tursiops truncatus) in The
Bahamas. Dr. Herzing has studied this pod of dolphins since
1985 using underwater video, and has currently identified
over 200 individual dolphins which can be recognized from
their specific patterns of body spots. WDP’s knowledge
of the individual spotted and bottlenosed dolphins in its
catalog acquired over 20 years of fieldwork is unique in the
world, and present an ideal setting to employ a system as
the 4-channel UDDAS, which focuses on getting recordings from
individual dolphins. After all, if the life history and family
relationships of a dolphin are well known, it puts the high
frequency sounds recorded from that dolphin into a very meaningful
perspective.
The
first employment of the 4-channel UDDAS in The Bahamas was
during a fieldtrip with WDP in the summer of 2005. High frequency
sounds were successfully recorded from both spotted and bottlenosed
dolphins, engaged in a variety of different behaviors such
as crater feeding on the sand, night feeding, courtship, aggression,
and play. Analyzing these recordings, which involves localizing
the sounds and attributing them to specific individuals visible
on the simultaneous video, will take many months. It is our
hope that this information will shed new light on how communication
sounds are exchanged within a dolphin group, how echolocation
clicks might be adapted by dolphins, and whether specific
sounds are indicative of specific behaviors.
Documentation
The 4-channel UDDAS has been documented in JASA
117(4):2552 (2005) and won the Readers’ Choice Award
as best Research Category entry of the 2005
PC/104 Design Contest (see bottom of their webpage).
Sponsors and
Acknowledgements
Development of the 4-channel UDDAS has been made possible
by a number of sponsors, who willingly donated parts to support
this research project. The
Sexton Company LLC dba Sexton Photographics donated half
of the underwater housing and has been incredibly supportive
of this project. ADLINK
Technology America, Inc. made a major donation by providing
the DAQ-2010, a 4-channel data acquisition board. Because
of the combined small size and outstanding performance of
this board, development of the 4-channel UDDAS became possible
in the first place. Kontron
America donated the MOPS/lcd7, a passively cooled single-board
computer with enough processing power and memory to support
the extremely high data throughput of the 4-channel UDDAS.
Sanmina-SCI Corporation
donated and specially adapted a PC/104-plus power supply to
this project, one of the very few existing power supplies
in the PC/104-plus format. Douglas
Electronics, Inc. donated its PCI-to-PC/104-plus adapter
which allowed connecting the PCI data acuisition board to
the PC/104-plus stack. Finally, Lattice
Semiconductor Corporation provided 30 ispPAC80 filter
chips, an ideal chip for the 4-channel UDDAS because of its
capability to output differential signals (which reduces system
noise significantly) and because its gain and filter settings
can be easily programmed.
Michiel Schotten wishes to thank the Royal Netherlands Academy
of Arts and Sciences (KNAW, Schure-Beijerinck-Popping Foundation
) as well as the Royal Dutch Society of Sciences (KHMW, Pieter
Langerhuizen Foundation) for providing personal research funding.
Also, a big Mahalo goes out to a number of people who were
of special assistance to this project. The support of Ken
and Adrienne Sexton in the last phase of this project was
greatly appreciated. Dr. Whitlow Au is acknowledged for his
on-going support of this research and excellent electronics
coaching. Hank Lin and Roy Wan of Adlink Technology have been
extremely helpful with programming the data acquisition card.
Don Norris is acknowledged for his time and useful advices.
Furthermore, Bruce McGrath, Chuck White, Dwight Webster, Jim
Krebs, and Chad Pennebaker are acknowledged for their personal
dedication in realizing sponsorships. Co-investigator Dr.
Denise Herzing and all crew members of RV Stenella have been
extremely supportive of this research and have provided the
photos of the system in the field. Profs. Wim Wolff and John
Videler of Groningen University are thanked for their on-going
research feedback and logistics support. John and Willy Schotten
have always been very supportive. Natasja Baints is thanked
for her love and personal support.
Additional sponsors needed!
A number of additional items are still needed to be able
to analyze the recordings acquired with the 4-channel UDDAS.
The most urgent of these is a powerful LAPTOP,
with enough processing power and RAM to be able to handle
the very large data files which contain the dolphin recordings.
Other items that are also needed include a digital oscilloscope
and a signal generator.
Additionally, research funds are urgently needed in order
to be able to analyze the recordings made with the 4-channel
UDDAS in 2005, and to be able to continue using this device
in the field to acquire new recordings. Continuation of this
research in the near future strongly depends on our success
in finding new sources of funding..
If you’d like to make a contribution, either in the
form of a donation of an item that your company sells, or
in the form of research funds which will enable us to continue
this work, please contact OSI researcher Michiel Schotten
(mike_schotten@hotmail.com)
All donations to OSI (which is a 501(c)(3) nonprofit organization)
are 100% tax-deductible. Additionally, if you wish we will
add your company’s logo and/or weblink to this webpage
as a contributing sponsor, and will acknowledge your sponsorship
in any publications that will result from this research, i.e.
both publications in scientific journals and in the popular
media. Furthermore, if your company or organization wishes
to use sponsorship of this research for advertising purposes,
that can also be arranged..
If you would like to make a donation to this project directly,
please send your check using our donations
form and indicate the amount under ‘UDDAS research’.
If you wish to contribute to any other OSI project, please
indicate this on the form as well. Thank you for your interest
and support!
Introduction
