3d. Approach

1969/12/31 - 20:00

The approach consists of three tasks: (1) Determining sentinel site locations, (2) The technical use of optical and acoustic imaging, and (3) Production of statistical and data visualization products, and socio-economic metrics for end users. Persons primarily responsible for each of these tasks are indicated in brackets.

3d1. Sentinel Site Selection [Taylor, Fogarty, Auster, Ford, Wiley, McGee, Noji, Incze]
Purpose: To establish sentinel locations along the northeast coastline where benthic community structure, the coupling between the water column and benthic community, and system change over time (at scales of days to years) will be quantified using the instrumentation described in Objective 2
Areas both open and closed to fishing on Georges Bank, Stellwagen Marine Sanctuary, and the New England and Mid-Atlantic shelf are primary targets. Surveys of these areas will be conducted using instruments towed from a commercial fishing vessel over defined tracklines twice per year over a three year period to quantify benthic and plankton community structure repeatedly in the same areas over time.
Six separate criteria were reviewed in order to approach the selection of areas in a structured manner: (1) does the area exhibit a high composite score using the NEFMC/NMFS Essential Fish Habitat methodology, with higher values reflecting the presence of more managed species, as documented by the NMFS trawl survey data (Fig. 3), (2) is the area used by more than one major fishery, with weighting given to areas with multiple fisheries, such as groundfish, scallop, herring, surf clam, or lobster, (3) does the area demonstrate significant usage by protected species, (4) does the area have existing or planned multibeam backscatter data in order to compare with bottom imaging data to be collected during our cruises, (5) does the area contain one or more oceanographic features (i.e., gyres, fronts, stratification, shear zones) where variability in the feature may drive benthic community structure, and (6) is the area co-located with existing or proposed NERACOOS assets? Areas exhibiting significant overlap of the above criteria were selected to maximize applicability to the widest range of potential users.
Proposed Areas (refer to Table 1, Appendix)
Each of the Study Sites will be visited twice per year in both the winter and summer for a period of 2 days of data collection. Tracklines will be surveyed both optically using our high resolution, color stereo imaging system and acoustically using multibeam (300mHz) during the study period. Two days of continuous sampling at 4-5 kts translates into a 350 linear km by 1 m wide swath for optical data and 350 linear km at about 350 m swath (122 km2) for the acoustic data.
Four areas are proposed for repeated benthic imaging cruises (Fig. 1):
Sentinel Site 1. While each of the named areas within this first group are distinct, they are spatially contiguous and are to be integrated into an extended whole for the purpose of cruise planning (Fig. 4). (1) Cape Cod Bay and other areas within state waters with recent USGS multibeam efforts, from the Merrimac River to the Cape Cod Canal, and NOAA multibeam efforts planned for 2007 in Cape Cod Bay. (2) Stellwagen Bank National Marine Sanctuary: We will leverage three existing multibeam datasets covering the Sanctuary and Jeffreys Ledge. There will be a diverse focus on multiple areas, including the area where the shipping channel is being moved to avoid known high density areas of Right Whale usage. (3) Jeffreys Ledge / Western Gulf of Maine Closure Area, characterized by high relief and varied habitats. To date we have only short transects within the area. GoMOOS and the Mass Water Resources Authority operate several hydrographic buoys in this area.
Sentinel Site 2. Northern Edge, Georges Bank, Habitat Area of Particular Concern (Fig. 5). There is no known multi-beam or side scan data in the area. This small (~200 nm2) area is the only example of an area designated a Habitat Area of Particular Concern to date, a special designation above all others in the US Northwest Atlantic EEZ outside of the Sanctuary system. Substrate features include flat sand, sand waves, and gravels, some mesas, and a long ridge to the east, and offer direct comparison with the heavily fished adjacent areas A longitudinal jet rips across northern boundary of the HAPC providing a high energy environment and substantial scour. To date we have a single cruise into the area with about 150 nm of transect, with additional time scheduled for August 2007.
Sentinel Site 3. Great South Channel/ Closed Area I (Fig. 6). Multibeam data (95 mHz) is available (Valentine, USGS, 2000) for a portion of the deeper waters (>70m) mostly within the boundaries of Closed Area 1. This area is characterized by high current flow, sand, sand ridges (central) and gravel pavements (west), with the finer sediments scoured away in the shoaler areas by strong currents. A strong, persistent front is a prominent feature resulting in concentration of plankton and subsequently this is an important area for feeding right whales. A NERACOOS hydrographic buoy is located in this area. To date we have about 300 nm of photographic transects within the area.
Sentinel Site 4. New Jersey shelf, including Elephant Trunk Access Area (Fig. 7). No known multibeam or sidescan acoustics exist for this area. The substrate is generally silty sand in a relatively low energy environment in the absence of storm events. The water column becomes strongly stratified in summer with potential for anoxic conditions and resultant fish and shellfish kills. Current fisheries resources include extremely high scallop biomass (100 million pounds), historically a large winter aggregation area for summer flounder, along with squid, mackeral, with tilefish in the deeper (> 100m) mud bottoms. Protected species present include sea turtle and whale populations. Other areas along the trackline to and from port can be sampled with little addition of time to cruises, for example Hudson Canyon Trench, Hudson Canyon Scallop Area, Southern New England shelf south of Long Island. As part of NERACOOS, Rutgers operates a glider track out to the Shelf break directly through this area so water column hydrographic and bio-optical data will be available. There is an existing HabCam transect of 50 nm, with more effort scheduled under our RSA program in the area for July 2007.
NOAA Fisheries NEFSC is currently developing a benthic research and monitoring program for the Northeast Shelf Ecosystem. The products from the proposed activities will directly support their mission and help identify optimal sites and transects for initiating a benthic monitoring program.
3d2. Technical approach:
Purpose: To develop the tools necessary to collect, integrate, and fuse disparate fisheries-relevant data sets, to segment and classify epi-benthic targets and substrate, and, to visualize the results in near real-time using state-of-the-art visualization tools.
The fishing vessel F/V Kathy Marie, which we have been using as part of our scallop survey work, has been outfitted with an overboarding a-frame and fiber optic winch system. The optical and acoustic imaging systems along with hydrographic and biological sensors will be mounted on the tow vehicle and operated simultaneously at an altitude of 3m off the bottom (Fig 8). The imaging vehicle, HabCam, was developed using funds from the Northeast Consortium and the National Marine Fisheries Service administered Research Set Aside program. Details of the vehicle and its capabilities can be found in Appendix I and in Gallager et al. (2005) and Howland, et al. (2006).
Data Flow and Processing [Gallager, Prasad, Auster, Vine, Taylor, Valentine]
The flow chart (Fig. 9) describes data flow from image capture to distribution of the final data products. Each image is georeferenced to a discrete timestamp, latitude and longitude so that ancillary data such as hydrographic information, bathymetry, vehicle engineering, plankton abundance, etc. may be attributed to each image. Raw 16 bit TIFF images require both lightfield and color correction before being stored in the processed data base. Following processing and convertion to 8 bit jpegs, the images are segmented by several algorithms currently undergoing research and development. Region of interest (ROI) extaction of the target from the background is followed by extraction of features critical to both identifying the target and classifying substrate type. Both supervised and unsupervised classifiers are currently being used to identify more than 60 taxonomic categories from HabCam images (Table 2, Appendix). Texture energy is used to classify substrate based on a modification of the structure proposed by Valentine et al (2005). Each of these processing tasks are explained in detail in Appendix II.
Water Column Measurements [Gallager]
Coupling water column properties to the benthic environment is critical to understanding ecosystem response to natural and anthropogenic impacts. The NERACOOS observatory will have stations and data products available for water column measurements throughout our study areas (Fig. 1). The focus for NERACOOS is water quality and transport of Harmful Algal Blooms (HABS) so a variety of measurements from 4D current structure, and nutrients to primary and secondary production will be available. However, processes in the benthic boundary layer (BBL) are in many ways decoupled from the water column so we will need to make measurements of hydrography, chlorophyll, and plankton directly from our towed imaging vehicles. The HabCam vehicles will contain a SeaScan, Inc. digital, color Video Plankton Recorder similar to the two units currently in use by NMFS in Woods Hole, but with real-time data transmission of images to the surface. On shipboard, the image will be processed according to standard methods (Davis, Gallager et al. 2005) to provide abundance of major taxa (e.g., copepod, ctenophore, diatom) and species in some cases. The vehicle will also carry a YSI 6600 Sonde with probes for temperature, salinity, pH, dissolved oxygen, turbidity, and chlorophyll to provide a proxy for phytoplankton abundance. A SonTek Acoustic Doppler Velocimeter will provide information on mean water motion and turbulence in the benthic boundary layer. Since the vehicle travels 2-3m off the bottom, near bottom water characteristics will provide evidence of hydrographic conditions, mixing and food supply to the benthos.
Integration of acoustics and optics [Mayer, Vine, Ford, Lerner]
A fundamental objective of this proposal is to develop approaches for using quantitative acoustic measurements to extrapolate, with confidence, habitat-relevant parameters between sparse lines of ultra-high resolution optical imagery. Our approach involves the collection of “high-resolution (typically at frequencies of 300 – 455 kHz) multibeam sonar bathymetry and backscatter data. At these frequencies, multibeam sonars can produce bathymetry with resolution on the order of 0.02 - .05% of the water depth (or height above the bottom in the case of a system on a towed or autonomous vehicle) and lateral resolution on the order 1 – 2 % of the water depth (or height above the bottom). While attaining these levels of resolution, these sonars also typically cover an area of between three and five times the water depth.
The ability to collect high-resolution bathymetry provides a first and fundamental component of habitat determination – a detailed depiction of the depth and general nature of the seafloor (e.g., flat, rocky, etc.) along with the ability to quantitatively analyze the roughness or rougosity of the seafloor (e.g. Cutter et al., 2003). Just as importantly, modern multibeam sonars also collect acoustic backscatter data which can easily be transformed into a backscatter mosaic that is a qualitative expression of changes in seafloor type. These backscatter data can also be used for quantitative analyses and characterization of seafloor properties.
The starting point in the process of using backscatter for quantitative seafloor characterization is the correction of the backscatter for radiometric and geometric factors (Fonseca and Calder 2005). These corrections involve the removal of the effect of changes in gain, power level and pulse width, and the effect of any residual beam pattern. The observed differences in acoustic backscatter are related only to differences in seafloor properties and the backscatter values will represent the actual backscatter cross-section returning from the seafloor. The acoustic backscatter values from different acquisition lines can be reduced to a near-calibrated scale of scattering strength, and can be compared directly to ground truth (HabCam images) or to a mathematical model (Fig. 14) (Fonseca and Mayer, in press).
Our approach will be to collect (or analyze previously collected) multibeam sonar data that is co-located with HabCam image data. We will run our Angular Range Analysis (ARA) analyses on the multibeam sonar data and then compare these results with the range of products derived from the analysis of the HabCam images. We will use state-of-the-art visualization and data fusion techniques to explore the relationships between the acoustically derived parameters and those derived from the HabCam. As relationships are found, for example between particular biotic associations and quantitatively-derived seafloor descriptors, we will then have the ability to extrapolate the ultra-high resolution HabCam-derived associations far beyond the limited range of optical measurements. As the HabCam will be run in a series of offset “skunk stripes” we will be able to test the veracity of our extrapolation at the next intersection of a camera track with the multibeam sonar data. Segmentation algorithms being developed for HabCam imagery (Prasad) will also be useful in segmenting acoustical information.

Standardized Image Test Set [Taylor, Gallager]
We have developed libraries of images that have been manually classified for foreground targets and background substrate, which are subsequesntly used as training sets for the automated classifier. Based on our HabCam field program, we have subsampled four important areas in the northeast for commercial scallop populations: Nantucket Lightship, Great South Channel Closed Area I, Northeast peak Closed Area II, and the Elephant Trunk area off of New Jersy. Using images from a varietry of areas is critical to optimizing the image processing algorithms since substrate and water quality differ extensively from one region to the next. About 100,000 images from each area were moved onto the HabCam server (http://habcam.whoi.edu) where image processing specialists the world over access the images and provide comparisons using different image processing algorithms. Manual processing of each image in this unique test set is nearly complete. As part of the proposed project, the Stellwagen Bank Mairne Sanctuary and Cape Cod Bay region will be added to this library. Ultimately, once a particular subset of images has been manually classified then the autoclassification can proceed in near real-time while at sea.
Automated Classification [Gallager, Prasad]
Currently, we are using supervised classification algorithms (Suppport Vector Machine [SVM]) to bin targets and substrate into trained categories by virtue of their color and texture. Results indicate we can automatically classify 60+ taxonomic categories at an accuracy of about 80% (Table 2, and Appendix II).
Near Real-Time Data Products [Vine, Mayer, Lerner]
Results of the autoclassified images are then fed into a series of processes ranging from species-specific abundance and distributional plots, GIS mapping of organism distributions and associations with substrate, and development of mosaics in order to achieve true nearest-neighbor distances. Ultimately, the output from the near real-time products could be used for adaptive sampling to dynamically allocate ship resources most efficiently.
3d4. Integration of data and high-order data products [Fogarty, Noji, Auster, Mayer, Gallager, in collaboration with all end users]
Purpose: To establish metrics for quantifying change in benthic community structure, organism abundance and size distribution.
The results of the near real-time products such as species distribution and abundance will be used to develop high level synthesis products including patch dynamics and habitat associations over time. Ultimately, these data will be used in multispecies management and population dynamics models. Examples of data products available to end users or that will be developed in collaboration with our partners as part of the NEBO and NERACOOS projects include:
1. Taxon- specific abundance plots of macrobenthos and macrophytes (Fig. 15, 19),
2. Spatial plots of Species Diversity Indices (e.g., Shannon-Weiner) as a function of binning scale) (Fig. 16),
3. Species-specific patch size estimates using point process statistics (Fig. 17),
4. Organism-substrate association indices using cluster analysis (Fig. 18, 20),
5. Georeferenced species and substrate distributional map overlays using ArcGIS,
6. Water column properties- hydrography, chlorophyll, and zooplankton abundance in relation to habitat structure,
7. Temporal information including seasonal and inter annual variation in: species and substrate composition, organism-substrate associations, water column properties, and food availability,
8. Imagery data sets and mosaics for each of the sentinel sites to be available over the internet,
9. Estimates of annual variation in larval/juvenile recruitment for selected species,
10. Compile data to support characterization and designation of EFH,
3d5. Assessment of economic benefits of NEBO to management [Kite-Powell]
We propose to estimate the economic value of NEBO data products to the management of New England groundfish stocks. We will derive this estimate primarily from a model of groundfish management and economic yield. We will verify the results by observing how NEBO data products are used in practice, and resulting changes in management and fishing yield over the course of the project.
(Task 1) Review of area closure decision processes. Decisions about closing offshore areas to fishing have two primary characteristics: the geographic extent of the closed area, and the temporal duration. We will review recent area closure decisions by the New England Fisheries Management Council [e.g., Cod Conservation Zone between Boston and Salem, MA] with particular attention to the role of information about benthic conditions in these decisions, and develop a conceptual model of the role of information about benthic conditions in these decisions. We will select recent area closure decisions for detailed analyses.
(Task 2) Effect of NEBO data products on area closure decisions. For each of the case studies selected above, we will consider how NEBO data products would have changed the closure management decision. To do this, we will compare the NEBO information with the “status quo” information in terms such as content, coverage, and resolution, and use the conceptual model (Task 1) to determine how each closure decision might have differed with the NEBO information. We will discuss and review the work on these first two tasks with appropriate members and/or staff of the New England Fisheries Management Council [Leslie Ann McGee and Sally McGee] and the Northeast Fisheries Science Center [Michael Fogarty and Thomas Noji]. The product of Task 2 will be descriptions of how the closure dimensions (geographic and temporal) in each case study might have been different if NEBO information had been available to the decision makers.
(Task 3) Effort and yield under two scenarios. For each case study, we will estimate how fishing effort and catch (yield) might have been different under the NEBO closure scenario. For example, a different closure geography would likely have resulted in more or less fishing effort in certain areas. We will assume for simplicity that closures under both scenarios will have the same effect on recruitment and stock structure (although in principle it is possible that improved information about benthic conditions could lead to closure decisions that are both more effective and more efficient), and that all other fishing regulations remain constant across the two scenarios. Effort and yield information will be based on NMFS data on fishing trips and catch for the areas contained by the case studies (Hoagland and Kite-Powell 2003).
(Task 4) Estimate of economic value of NEBO. For each case study, we will estimate the economic value of NEBO information to the closure decision as the difference in economic yield from fisheries affected by this closure under the two scenarios (Hoagland and Kite-Powell 2003). We will extrapolate from these case studies to generate an estimate of the value of NEBO data to groundfish stock management decisions for the New England region as a whole. All economic value estimates will be annualized ($/year).
(Task 5) Verification and refinement of models. We will track use of the NEBO data by maintaining contacts with relevant staff in the New England Fisheries Management Council and the Northeast Fisheries Science Center. We will use these observations to verify the accuracy of the conceptual model (Task 1) and our estimates of likely effects of NEBO and NERACOOS data on area closure decisions (Task 2), and modify our analyses if we detect major discrepancies. The overall product of the economic assessment will be an estimate of the economic value of NEBO information to specific fisheries management (area closure) decisions, and an estimate of annual value of this kind of information to the New England groundfish and shellfisheries as a whole.
3d6. Role of federal and state partners
The NOAA, National Marine Fisheries Service, will provide expertise on developing fishery related data products, habitat characterization, and ecosystem based management. The National Undersea Research Center will provide expertise in benthic composition, and habitat structure and diversity. The US Geological Survey will provide multibeam data and expertise in substrate characterization. The Massachusetts Division of Marine Fisheries will supply the Kongsberg EM3002 multibean sonar to be used in conjunction with all HabCam tows. The Los Alamos National Laboratory will provide expertise on novel approaches to image segmentation and classification. The New England Fishery Management Council will provide guidance and specific requirements for data products necessary for multi-species fishery management. The Stellwagen Bank National Marine Sanctuary will guide development of data products specifically matched to management of marine sanctuaries. The Census of Marine Life will help develop data archiving tools and species diversity indicies. The UNH Center for Coastal Ocean Mapping will provide expertise on acoustic imaging the seafloor, visualization of large, disparate data sets, and an educational component coordinated through the UNH Coastal Observing Center program.
3d7. Education and outreach [Holt-Cline]
The seafloor imagery collected through the NEBO efforts has the potential of reaching a wide range of audiences through the existing outreach components of participating organizations, with each contributing directly to increased access and usage. The New England Fishery Management Council, Stellwagen Bank National Marine Sanctuary, the National Marine Fisheries Service Ecosystems Processes Division, the Census for Marine Life web portal, and national, state and regional Marine Educators Associations will each have direct access to this information. Education and outreach activities will be coordinated through the University of New Hampshire Coastal Observing Center (Amy Holt Cline, Education and Outreach Coordinator) and through the Cooperative Institute for Climate and Ocean Research, a Joint Institute of the Woods Hole Oceanographic Institution and the National Oceanic and Atmospheric Administration (Robert Weller, CICOR Director). All activities will be fully integrated with the NERACOOS program whereby an outreach team will be established that will coordinate through workshops with teachers, scientists and educators to provide a forum to develop products for the K-12 classroom along with other audiences. The image based nature of NEBO lends itself well to visualization and interaction by students of all ages. We will work with UNH Coastal Observing Center, the UNH Center for Coastal Ocean Mapping (Briana Sullivan) and CICOR to develop specific tools for use in the classroom setting to visualize the seafloor and interactions between benthic populations. Key concepts related to species diversity and habitat classification will be developed for use by students and the public.
CICOR supports workshops for teachers, involvement with NOAA's Teacher at Sea Program, outreach to schools, and summer students at WHOI. At least two summer students will be involved each year through this program.
We will also develop a visualization display for the WHOI Exhibit Center, which along with the WHOI Information office, serves more than 30,000 visitors each year. The New England Aquarium, Boston, MA will also be the target for a full motion display of scrolling mosaics of the sea floor describing various regions on Georges Bank and Stellwagen Bank. The Mass Fisherman’s Partnership will provide an additional outreach component by convening workshops and forums to present our findings to the general public and fishing community.
3d8. Project management and the relationship of NEBO to NERACOOS
WHOI will administer project funding both internally and to all subcontractors. A science and technology steering committee will be established with representatives from all organizations and end users. Regular meetings and/or conference calls (monthly) will allow dissemination of new information, discussion of any proposed changes to the program, and development of new data products to fulfill the needs of our stakeholders. Since NEBO is essentially the benthic component of NERACOOS, we will be fully integrated into planning sessions and team meetings. Also, as NERACOOS becomes established as the governing body for IOOS activities in the northeast, NEBO and its stakeholders will be represented on the governing board.