Project: Integral Habitat Mapping
An important consideration with marine renewable energy (MRE) systems is their overall impact on surrounding habitats. Integral Consulting, Inc. created an accessible ecosystem monitoring method that uses pre- and post-installation habitat mapping to assess the impact of marine energy systems on their surrounding benthic habitat. The method is two-part—photographic and acoustic surveys.
Sediment profile and plan viewing imaging (SPI/PV) is a photographic survey technology first developed in the 1970s as a benthic research tool. SPI/PV is useful for mapping sedimentological, geochemical, and ecological conditions on the seafloor. It obtains high-resolution images of the upper 20 cm of the sediment column with minimal physical or animal disturbance. It uses a prism camera that researchers insert vertically into the seafloor to capture a sediment profile. In this project, the technology was shown to be a cost-effective way to ground truth sonar seabed maps. A significant achievement of this project was the development of an AI-based image processing software that streamlines and standardizes the extraction of quantitative information from the imagery.
The project team completed an acoustic survey of Sequim Bay in spring 2017. The first run included a multibeam echosounder sonar survey and collection of surface sediment samples as well as SPI/PV images at 50 locations for ground truthing the acoustic data. The data provided baseline bathymetric imagery of the ocean floor, which allowed for estimates of grain size, and the sediment samples provided laboratory grain size measurements, which were compared to—and found consistent with—the data from the SPI/PV.
Researchers completed an additional survey at Sequim’s Dungeness Spit in 2018, and returned to MSL in 2019 to apply those findings to technical development. The upgrades include improved SPI camera operations that allow for better penetration in firm substrates, enabling photography of the sediment profile to greater depths. Software upgrades were made to allow for automated data processing with decreased reliance on manual substrate characterization. The SPI/PV habitat mapping project then completed its final phase at the PacWave South wave energy test site.
Project: PNNL Igiugig Fish Video Analysis
The Igiugig Fish Video Analysis project analyzed underwater video data collected around a river turbine and developed a suite of algorithms that allows automatic detection of fish from video data. Through these efforts, collaborators determined that using a combined approach of automatic detection software and human analysis can reduce labor time by half as well as improve reporting accuracy over sampling-based methods.
The Ocean Renewable Power Company, Inc. (ORPC) deployed their RivGen® turbine at Igiugig, Alaska, for two months in 2015. Researchers positioned five video cameras around the turbine and collected video for approximately one week each month, with lights that illuminated the turbine at night to allow for continuous monitoring.
During the deployment, researchers manually analyzed the video data the first 10 minutes of each hour to detect fish and describe their behavior. The Alaska Department of Fish and Game used the analyzed data for permitting requirements.
MSL researchers analyzed the video datasets to determine the number of fish that went through the turbine or made contact with the turbine or surrounding platform, and the difference in fish quantity seen during the day versus night. This effort also noted any detectable behavioral differences in the fish when they were around the turbine.
Previous work completed by the University of Maine informed the analyses to ensure compatible analytic results between acoustic cameras observing fish around turbines, and the use of video data for the same purpose. In parallel, drawing on collaboration with the University of Washington (UW), PNNL researchers created EyeSea, a machine-learning tool that automatically identifies the presence of fish in the video data and helps reduce the time for manual analysis.
The EyeSea software is being further developed to support the DOE-funded Igiugig Hydrokinetic Project at the village of Igiugig, AK. The software will be used by University of Alaska – Fairbanks fish biologists to monitor migrating salmon interactions with the installed Ocean Renewable Power Company’s RivGen Power System.
Over the course of the project, collaborators discovered that most interactions between fish and the turbine occurred at night, and the frequency of fish interactions did not appear to be affected by whether the turbine was spinning or static. Not surprisingly, the analyses also found that adult fish are more likely to avoid collision and show avoidance behavior compared to juveniles.
The fish-detection algorithms helped eliminate most video that did not contain fish, allowing subsequent human analysis to focus on segments most likely to contain fish. This helped researchers focus the sampling of videos on only those that were most likely to include fish.
Project: University of Washington 2G AMP
Nicknamed the “Millennium Falcon,” the second-generation intelligent Adaptable Monitoring Package (2G AMP) collects an integrated set of environmental data for marine energy devices. The 2G AMP supports multiple oceanographic sensors, including stereo-optical cameras and lights, an acoustic camera, a multibeam sonar, an acoustic Doppler wave and current profiler, four passive hydrophones, and a passive fish tag receiver using acoustic telemetry. It connects to a docking station that communicates to the shore for power and data connectivity. As it continuously streams sensor data, only data on detected targets is archived—the rest goes to temporary storage, thereby saving storage space so that the instrument can use its high-resolution sensors without the massive data storage and analysis requirements associated with constant archiving.
Initially tested at the University of Washington (UW), the 2G AMP was deployed in the inlet channel to Sequim Bay from August to November 2015 and January to May 2016. This long-term endurance monitoring enabled more efficient integration between the different sensors and demonstrated the package functioned well over many months, with greater than 90 percent operating time in 2016. Due to its success, UW has continued to improve the package with its third-generation model. See the 3G AMP project for details.
UW developed the instrument package in collaboration with Oregon State University, Sea Mammal Research Unit Ltd, and PNNL.