Sub-contract proposal ADVANCING REMOTE MARINE MAMMAL by fjzhangweiyun


									                             Sub-contract proposal


                                 Submitted to

                     Pacific Islands Fisheries Science Center
                National Oceanic and Atmospheric Administration

                           Funding Opportunity
                  Number NOAA-NMFS-PIFSC-2010-2002304

                                19 March 2010


                                 Bruce M. Howe
University of Hawai’i at Manoa, School of Ocean and Earth Science and Technology
               2540 Dole St., Holmes Hall #402, Honolulu, HI 96822
                                 (808) 956-0466

                  Project Duration: 1 May 2010 – 30 April 2012

              Amount Requested: Year 1: $64,000 Year 2: $96,000
Project Summary

Traditional methods of cetacean stock assessment often require resources beyond those
available in most regions and institutions. As such, stock assessments are inadequate for
many species and have not been completed at all in many regions. We propose an
alternative platform for cetacean assessments. The acoustic Seaglider is an autonomous
vehicle capable of carrying acoustic instrumentation as well as oceanographic sensors for
periods of up to 200 days. In the first year of our proposed project we will test the
cetacean detection capabilities of the Seaglider relative to visual methods during a
planned cetacean assessment around Hawaii. In response to the outcome of the test
mission, we will also begin developing new hardware and software aboard the glider for
detection, classification, and localization of vocal animals. In the second project year we
will implement the acoustic design changes and again test the effectiveness of the glider
versus a traditional visual survey. Although the experimental design of a glider-based
survey must be modified from that of visual line-transect surveys, we are confident that
established statistical methods can be used to evaluate cetacean distribution and
abundance based on acoustic detections and localizations from the glider. Alternatives to
traditional methods are sorely needed, particularly in the Pacific Islands Region where
the requirements across the survey area vastly exceed the resources available to conduct
robust assessments.

The project will be a joint effort of investigators at the PIFSC (E. Oleson and J. Polovina)
and the University of Hawaii (B. Howe). The UH component of the work, proposed
herein, will consist primarily of preparing and operating the acoustic Seagliders,
providing acoustic, temperature and salinity data for colleagues at PIFSC for analysis.
Project Description

Problem Statement

Under the Marine Mammal Protection Act, NMFS is required to assess the status of cetacean
stocks occurring within US waters every three years. Cetacean stock assessments must include
description of the stocks geographic range, minimum population estimate, the population trend
and maximum productivity rate, Potential Biological Removal (PBR) rates, the status of the
stock relative to the PBR, and estimates of human-caused mortality. Existing assessments have
been classified into two tiers, Tier I and Tier II, based on the accuracy of stock identification,
abundance, and mortality estimates, and on the assessment frequency and data quality (Merrick
et al. 2004). As of the last Tier I and Tier II status update, only 18% of identified cetacean stocks
were classified as Tier II. New requirements for more advanced ecosystem-based assessments
(Tier III- Merrick et al. 2007) are now being considered.

Cetacean stock assessment surveys require a very large investment of resources. Regions with
significant annual allocations of ship time and ample personnel often struggle to maintain up-to-
date assessments for all species, with many of those assessments including only the most basic
biological and distributional information (Merrick et al. 2004). The NMFS Pacific Islands
Region (PIR) encompasses the U.S. Exclusive Economic Zone (EEZ) around the Hawaiian
archipelago, Johnston Atoll, Kingman Reef and Palmyra Atoll, Baker and Howland Islands,
Jarvis Island, American Samoa, Wake Island, Guam, and the Commonwealth of the Northern
Mariana Islands (CNMI), the largest region under NMFS management and representing
approximately 1.8 million nmi2 of ocean. Stock assessments for marine mammals in this region
are currently available only for the Hawaiian EEZ (Carretta et al. 2009), and with the exception
of single survey in the Palmyra and Johnston EEZs (Barlow et al. 2008), cetacean stocks in the
remaining PIR have never been surveyed. Known cetacean stocks in the PIR are all at the Tier I
stage. Over 500 days of ship time are required to conduct one line-transect survey of each PIR
EEZ. With limited ship time and resources, this represents a significant challenge to using
traditional ship-based methods, thus requiring new approaches and new technology in order to
meet our assessment mandates. As NMFS moves toward next generation assessments of
protected species (Tier III assessments), the PIR is in a unique position to design and implement
a more advanced, mobile, and flexible system for cetacean assessment, even during the first
surveys of many of the region’s EEZs.

Scope and Objectives

Distribution and abundance of cetacean stocks is typically surveyed using visual line transect
techniques (Buckland et al. 2001), where density of animals is a function of the number of
animals seen along systematically placed track lines and the probability of detecting the target
animal at various distances from the track line. Successful execution of this technique often
requires months of continuous ship time and the support of dozens of personnel, particularly
when used to assess cetacean abundance over broad geographic areas. In some regions, surveys
include the collection of oceanographic datasets that may be used to place the occurrence of each
cetacean species into an ecosystem context, thereby improving stock assessments with habitat
The field of passive acoustic sensing for marine mammals has grown immensely over the past 10
years. Towed hydrophones have become a regular addition to the traditional line transect visual
survey design as they provide an alternative method for detecting vocal animals that are not
available to the visual observers – because they are submerged, difficult to see, or further from
the survey vessel – therefore producing more precise abundance estimates. The addition of towed
hydrophone arrays has provided correction factors for the number of animals missed by the
visual team, increasing the accuracy of the abundance estimate (Barlow & Rankin 2007), and for
some species, has provided independent estimates of species abundance (Barlow and Taylor
2005). Stationary acoustic sensors have provided opportunity for year-round assessment of the
occurrence of several cetacean species (partially reviewed by Mellinger et al. 2007). Under
certain environmental conditions or when placed in the appropriate configuration, stationary
sensors may also provide the capability to assess abundance of cetaceans over time (e.g.
McDonald & Fox 1999, Marques et al. 2009). However, both towed and fixed acoustic methods
are limited in their capabilities; towed surveys by ship time allocation and the timing of ship
surveys, stationary surveys by the small geographic range of the assessment. The success of
passive acoustics in detecting cetaceans suggests that deployment of acoustic sensors on
alternative platforms may provide a robust, cost-effective, and flexible means of surveying for
many species.

Gliders are autonomous, re-usable, underwater vehicles designed to glide along a saw tooth
trajectory from the ocean surface to a programmed depth, measuring a suite of oceanographic
quantities along their path. Gliders can be deployed from a pier or from a small boat requiring far
less logistical support than a large-scale ship survey. In addition, the sensor suite of a glider
provides the capability to collect the oceanographic measurements that allow for assessment of
cetaceans within an ecosystem-based context. Incorporation of oceanographic variables into
models of cetacean occurrence has already proven useful in reducing the uncertainty of cetacean
density estimates (Ferguson et al. 2006b, Forney 2000).

We propose an experiment to assess the current acoustic sensing capabilities of existing gliders
to provide an alternative means of measuring cetacean occurrence, and development of more
advanced acoustic sensing and processing methods to allow for estimation of distance to vocal
animals, a key component when using distance sampling statistical methods. By project year, our
goals are the following:

Year 1
   1) assess the detection capability of the acoustically-capable Seaglider versus a visual line-
       transect survey
   2) assess oceanographic correlates of cetacean occurrence based on glider detections and
       evaluate further development of glider-based oceanographic sensors
   3) initiate development of advanced acoustic detection and localization capabilities on the
       glider to provide the data sets necessary to compute abundance estimates from glider
Year 2
   1) implement new acoustic-sensing technology into the glider and test the new system
       versus a visual and acoustic survey
   2) evaluate new or enhanced existing statistical methods for estimating abundance from
       glider-based acoustic detections.

Technical Approach

Our proposal seeks to advance the capabilities of an existing platform to increase opportunity
and reduce cost for stock assessment surveys for cetaceans, particularly when use of traditional
methods is limited by available resources. We will use Seaglider (Eriksen et al. 2001), a
commercially-available ( unmanned underwater
vehicle whose capabilities have been expanded to provide duty-cycled continuous acoustic
sampling over the course of the deployment. Although Seaglider’s ability to detect cetacean
vocalizations has already been proven (Moore et al. 2007- Figure 1.), the comparison of those
capabilities to existing assessment methods has not been evaluated. Our goal will be to evaluate
the use of autonomous gliders, and by extension, other autonomous vehicles, for providing the
type and quality of data needed to carry out a quantitative assessment of cetacean populations.
This will be achieved through comparison of temporally concurrent glider and shipbased
cetacean surveys off the Kona coast during each of the project years.

The Seaglider: Seagliders are currently capable of travel at roughly ½ knot using ½ W of power.
With basic temperature and salinity sensors, glider missions can last over 200 days, including
600 dives to 1000 m, and cover distances up to 3500 km. The Seaglider has been designed to
make CTD and other measurements as it yo-yo’s through the water column, navigating along a
prescribed route (Eriksen et al., 2001; Davis et al., 2001). On a steady glide slope, the Seaglider
travels at a nominal speed of 0.25 m/s through the ocean, propelled by buoyancy. Buoyancy and
internal mass control, combined with body and wing hydrodynamics, cause Seaglider to
alternately dive and climb along slanting glide paths. The Seaglider dead reckons underwater
between sea surface GPS navigation fixes to glide through a sequence of programmed
waypoints. Via Iridium satellite data telemetry, Seaglider transmits collected environmental
(high resolution profiles of physical, chemical, and bio-optical) variables and internal system
data and receives command files. The antenna is exposed above the sea surface for a few minutes
between dive cycles. This ability to receive data and change mission goals in near real-time
provides tremendous operational flexibility. Gliders can be readily launched and recovered from
a small boat with a crew of two, avoiding costly reliance on ships.
                                                     Figure 1. From Moore et al. 2007. Example
                                                     of (a) blue A and B calls, (b) humpback
                                                     upsweeps and (c) sperm whale echolocation
                                                     clicks recorded on acoustic Seagliders
                                                     deployed in Monterey Bay.

An acoustic recording system (ARS), recently added to the basic Seaglider, uses a CF2 processor
combined with a 4 GB flash memory and a 60 GB hard disc. With each surfacing, the Seascan
clock in the ARS is synchronized to GPS providing absolute timing accuracy of about 1 ms over
a dive. The current Seaglider uses a single hydrophone and can maintain sampling at 4 kHz with
the low pass filter set at 30 kHz. The 64 kHz bandwidth is adequate for detection of the sounds
produced by mysticetes and many odontocetes. Software parameters (both acoustic recording
system and glider parameters such as waypoints) can be changed on the fly each time the glider
surfaces and communicates with the pilot on shore. The current Seaglider configuration is
limited to recording data just over 50% of the time due to the requirement to write data to disk
from flash when sampling is not occurring. This limitation is not expected to impact our results
and can be accounted for during data analysis.

Figure 2. The acoustic Seaglider. In the current configuration, the hydrophone is in the top of the
tail of the glider.
The second project year will include the incorporation of new acoustic technology (DMON)
capable of continuous broadband (up to 300kHz) acoustic sampling of three hydrophone
channels. The hydrophones can be placed at different points along the glider, or on the wings,
providing the capability for 2-dimensional localization of a vocalizing animal.

Experimental Design: The first phase of this project will include a multi-day field assessment of
the detection capabilities of Seaglider relative to a visual survey off the Kona coast of the Island
of Hawai’i. The leeward coast of Hawaii is surveyed several times per year for a few weeks each
survey. The glider will be preprogrammed to travel a series of transect lines that fall within the
visual survey area. Attempts will be made each day to conduct visual survey within the
immediate vicinity of the glider track in order to provide direct comparison of the two methods.
Because the glider travels at only ½ kt, while vessels generally survey at much higher speeds (8-
10+ kts), vessel and glider surveys will not occur in parallel, but will provide independent
measures of cetacean occurrence within the defined survey area. At the end of the survey, we
will compare local and regional detection rates between the visual survey and the glider
providing a measure of the ability of the glider to provide a representative sample of cetaceans in
the study area. Echosounders aboard the visual survey vessel will also be used to calibrate the
acoustic detection capabilities of the glider at various frequencies and distances. Recent
dualmode visual and acoustic surveys near the Hawaiian Islands and around Palmyra suggest
that the acoustic detection rate for most cetacean species will exceed visual detections for a large
number of species (Rankin et al. 2008), though some species, particularly those with
vocalizations higher than 30kHz, may not be detected as frequently or at all with the current
glider (Figure 3).

During the survey effort the glider will also collect high resolution temperature and salinity
measurements. The glider’s vertical temperature profile data will be analyzed to map the depth
of the top of the thermocline and any eddy or frontal features in the survey area. The detection of
cetaceans based on the glider acoustics will be compared to the vertical and horizontal
distribution of measured oceanographic variables to provide the basic data required to assess the
environmental context of cetacean occurrence within the study area.

We anticipate that it will be necessary to upgrade the current acoustic capabilities of the glider to
allow for higher bandwidth sampling in order to assess the occurrence of most cetacean species.
The DMON hydrophone system being developed by the Woods Hole Oceanographic Institution
is well suited to this goal and will be available for commercial purchase and incorporation into
other oceanographic systems by late-2010. The DMON system also has multiple hydrophone
channels providing the capability to obtain bearing estimates to vocal cetaceans. The DMON
processor is low power, but powerful enough to support sophisticated cetacean detection and
classification algorithms. As the DMON is integrated into the glider, we will begin testing sensor
placement to determine the optimal configuration to detect signals for distance and bearing
estimation, while minimizing interference with glider operation or other glider sampling
capabilities. We will aim for a 3-D hydrophone arrangement so that location of vocal cetacean
groups can be estimates based on the time of arrival of the sounds at each hydrophone.

In the second year of the project we will implement the glider improvements and repeat field
tests versus visual techniques. Analysis of the second year of field test data will focus on the
accuracy of bearing estimates generated from the glider hydrophones versus known animals or
test source locations and will include detailed model-based analysis of oceanographic correlates
to cetacean occurrence within the study area using a combination of methods outlined by
Redfern et al. (2006), including flexible generalized models (Hastie & Tibshirani 1990). We will
also compare of the detection function between the glider and a traditional ship survey. Detection
of cetacean vocalizations on a multi-hydrophone platform provides the capability to estimate
location, and therefore distance to the vocal group. This is the primary measure needed to apply
traditional distance sampling techniques to estimate density and abundance.

Figure 3. From Mellinger et al. (2007):
Known frequency ranges of cetacean
sounds. Large whales are listed by species
and toothed whales by family. The range of
the most common vocalizations (blue bar)
and the recorded extremes (thin line) are
The proposed project focuses primarily on the hardware and software improvements needed to
implement cetacean assessments using alterative glider technology. We are fully aware that
significant challenges remain for acoustic-only assessment of cetaceans, primarily including the
ability to classify detected sounds to specific species with a measured degree of certainty, and the
ability to estimate the size of detected groups based on the acoustic characters alone. Acoustic
detection methods contribute the greatest amount to visual surveys when species-specific calls
can be identified. The calls of many baleen whales are highly stereotyped and well-described
(Edds-Walton 1997) and a rapidly increasing number of tropical and subtropical delphinids can
be identified by the characteristics of their clicks and whistles (e.g. Oswald et al. 2003, Oswald et
al. 2007, Baumann-Pickering 2009). However, significant gaps still remain in our ability to
acoustically classify sounds to species. There are a great number of researchers and institutions
contributing data and techniques to address this short-coming, and we feel that once the glider
technology is proven capable of detecting and localizing cetaceans, new processing algorithms
can be integrated as they become available, either into the acoustic acquisition system onboard
the glider or during post-processing of the collected data.

Acoustic estimates of group size are currently not possible for most species. General trends
indicate that larger groups are more frequently detected acoustically than small groups (Rankin
et al. 2008); however, this may vary based on animal behavior. The oceanographic data collected
with the glider may provide some insight into the range of expected group sizes (Ferguson et al.
2006a); however, much more data on acoustic behavior and the relationship between acoustic
characters and group size is needed before we will be able to estimate this important parameter
based on acoustics alone. Ongoing studies should eventually provide this data, and the
appropriate correction factors can them be applied to newly advanced glider technology. The
Cetacean Research Program at PIFSC actively contributes to these efforts.

Our primary goal is to develop a new tool that will contribute to existing (Tier I and II) and
advanced (Tier III) cetacean stock assessments. We feel the increased detection capabilities and
continuous environmental sampling will provide powerful new methods for conducting initial
assessments in the PIR and for augmenting ongoing assessment efforts in regions with
established assessment resources and protocols.

Scientific Merit and ASTWG Mission

The proposed research addresses the mission of the ASTWG by advancing current glider
technology to provide a promising alternative means of conducting cetacean stock assessments.
Acoustic sensing is a very effective way to detect the occurrence of cetaceans, providing a higher
rate of detection than standard visual surveys for many species (Rankin et al. 2007).
Incorporating passive acoustic sensors into autonomous underwater vehicles, such as gliders,
provide the capability to effectively survey a broad region, a cost-effective alternative to the
current visual line-transect standard. The glider also provides the capability to continuously
collect oceanographic data that can be incorporated into models of cetacean distribution and
abundance, increasing the accuracy and precision of stock assessments. Gliders equipped with
acoustic sensors and software capable of classifying and localizing cetacean groups while
collecting environmental data may be used in any NMFS region. We propose developing this
technology in the PIR where this type of technology can not only fill basic assessment gaps, but
can rapidly increase the quality of assessments. A broad range of species can be reliably found in
the protected waters off the Kona coast of Hawaii providing an excellent test location for
evaluating design changes and sampling schemes. Once the hardware and software capabilities
are stable and proven robust, this technology may be easily deployed elsewhere in the PIR and in
other regions.

Qualifications of the Principle Investigator

Glider technical expertise and the gliders themselves will be provided by co-PI Bruce Howe at
the University of Hawaii. Howe has had over 30 years of experience in ocean acoustics, most
notably in ocean observing technology (please see CV). While at the University of Washington
Applied Physics Lab, he initiated the Acoustic Seaglider program in 2003 with ONR support and
in 2005-2006 deployed Seagliders carrying a broadband hydrophone (10 Hz – 30 kHz). The
acoustic Seaglider proved its capability in marine mammal detection under Howe’s direction
during the 2006 Monterey Bay PLUSNet experiment (Moore et al., 2008). The University of
Hawaii is currently operating 8 Seagliders: one as part of the Hawaii Integrated Ocean Observing
System (HIOOS), operating south of Oahu and in the channel between Oahu and Molokai; four
as part of the Center for Microbial Oceanography Research and Education (CMORE) at Station
ALOHA 100 km north of Oahu; and three for the ONR-funded Philippine Sea 2010 experiment
(to be deployed there November 2010-April 2011). One or more of the latter gliders will be used
for this work.

Work Plan, Schedule & Milestones

A general timeline is provided for this project in Table 1 below. Exact dates cannot be provided
as the timing of some phases will be dependant on ship time allocations.

The year 1 objective is to deploy one acoustic Seaglider on the leeward coast of Hawaii to record
marine mammal vocalizations, and to compare these with visual observations. The acoustic
Seaglider work will be carried out under the direction of Howe, with Oleson (NOAA) being
responsible for the overall experiment, including the visual observations.

In year 1, we will prepare one acoustic Seaglider that includes a 1-day test dive off Honolulu.
From a NOAA supplied vessel, we will deploy the glider off the west coast of Hawaii and pilot it
for up to one month while it collects data. The track and acoustic recording schedule will be
determined in collaboration with Oleson. The tentative time for deployment is in the April-June
2010 timeframe. Upon completion of the mission, the glider will be recovered by UH personnel,
again, using a NOAA-supplied vessel.

The glider acoustic and engineering data will be processed and made available to all
collaborators. This will include glider position and depth as a function of time, temperature and
salinity as a function of depth/time, and acoustic data files appropriately time-stamped. A brief
data report will be delivered that provides a summary of the engineering data and initial acoustic
results (e.g., time traces and spectra from representative files). Oleson and colleagues will
process the acoustic data for marine mammal vocalizations.
In year 2 – an option, essentially the same will be repeated, but this time, with a more capable
acoustic recording unit (based on DMON, made by WHOI (Mark Johnson), and integrated into
gliders by iRobot). This system will support up to three hydrophones, from which we will be
able to determine bearing to a vocalizing animal. The standard system comes with one
hydrophone; part of year 2 work will be to add the second and third hydrophones and test.

Howe will manage and direct the UH portion of the project. Karynne Morgan will provide
administrative assistance. Jennie Mowatt is the resident glider technician.

                                    Table 1. Work Schedule
Time          Activity                                 Milestones
2010 –
Year 1
Spring –      Acoustic Seaglider deployed during         1. Report to ASTWG on detection
Summer        visual surveys of the Kona coast of        capability of existing acoustic Seaglider
2010          Hawaii                                     and anticipated capability of upgraded
Fall 2010 –   Comparision of Seaglider and visual
Winter        survey performance
Spring        Integration of new acoustic system and
2011          optimal sensor placement for multi-
              channel sensing
Summer        Acoustic Seaglider deployed with new           1. Produce Final Report to
2011          acoustic system during planned visual             ASTWG
              survey off Hawaii                              2. Report on new glider
                                                                capabilities and future directions
                                                                to clleagues at national
                                                                workshops and meetings
Fall 2011 –   Work with survey design colleagues to
Winter        modify existing statistical methods for
2012          use with acoustic Seaglider detections
Spring        Comparison of Seaglider and visual
2012          survey performance and evaluate
              habitat-associations based on
              oceanographic sampling onboard the
Budget Justification

The total requested University of Hawaii (UH) budget for year 1 is $64,000. The budget includes
the costs necessary to provide engineering support and glider expertise for the proposed project.
These costs include partial salary support for UH PI Bruce Howe, as well as a Project Engineer
and Assistant. The SOEST Ocean Gliders (SOG) Facility has set a nominal per mission cost of
$36,000, which includes batteries, sensor calibration, insurance, Iridium communication charges,
and piloting. Also included are travel costs for field deployment of the glider of the Kona coast
and the costs of materials and supplies.

The total budget for year 2 is $96,000. The cost largely mirrors that of year 1, with the addition
of the purchase of the acoustic recorder DMON for incorporation into the glider ($10,000), and
increases due to inflation and merit-based personnel salary increases.

The current average benefit rate for UH faculty and staff is 37.13% of salaries. Composite fringe
rates varies for UH and RCUH employees, but benefits are similar for both and consist of
vacation reserve, sick leave, medical/dental insurance, retirement/pension, group life insurance,
long-term disability and care, FICA, unemployment insurance, workers compensation, and flex
spending. The DMON acoustic recorder is a component to be added to the glider (non stand
alone item without a lease-to-purchase option).

To top