Detection Of Whales Research Articles

Monitoring cetacean occurrence and variability in ambient sound in Scottish offshore waters

The characterisation of marine soundscapes allows observation of spatiotemporal distribution of vocalising species and human activities, which can inform an assessment of their interactions. Such data are important for monitoring the ecological status of marine habitats. The Scottish Atlantic Frontier is an important habitat for a variety of cetacean species. Historic whaling has heavily impacted several species inhabiting these waters and current comprehensive information about seasonal occurrence and distribution is lacking for all species. This study presents year-round passive acoustic monitoring data from ten sites in this understudied region. The three most offshore sites were examined for baleen whale vocalisations, and four species were regularly detected. Fin whale detections peaked from October to January and were at their lowest during May and June. Humpback whale song was detected as early as January but showed a strong seasonal peak in March and April. In contrast, minke whales were detected regularly throughout the year but with a peak in detections from October to November, when sei whales were also detected. All monitoring sites showed frequent occurrence of odontocete echolocation clicks and whistles. Comprised mainly of delphinid vocalisations, whistles and clicks were detected on an almost-daily basis among the offshore sites, with a slight reduction in detections from May to July particularly among the more inshore sites. Ambient sound levels (root-mean-square sounds pressure level; SPL) varied by site and season in relation to species presence, anthropogenic contributions, and environmental conditions. Monthly median SPL across the array varied up to 18 dB within 1/3-octave bands. Throughout the year, variability in median SPL was lowest in the higher frequency bands (>10 kHz), while highest variability was found between January to July in specific lower frequency bands (<1 kHz). Results from this study demonstrate the value of passive acoustic data in providing novel baseline information about cetacean occurrence and distribution in Scottish offshore habitats where data are limited and outdated. The results will feed into statutory reporting on underwater noise, support the identification and designation of future marine protected areas for cetaceans, and help guide management of future human-marine mammal interactions in Scottish offshore waters.

Acoustic presence and vocal repertoire of bowhead whales (Balaena mysticetus) in eastern and central Fram Strait

Bowhead whales (Balaena mysticetus) of the East Greenland-Svalbard-Barents Sea (Spitsbergen) population are still considered endangered, but knowledge on spatio-temporal distribution patterns and behavioral aspects remains scarce, yet crucial for this population’s conservation. Long-term passive acoustic recordings were collected at five locations in central and eastern Fram Strait (78-79°N, 0-7°E) as part of the Ocean Observing System FRAM (Frontiers in Arctic Marine Monitoring). Data recorded in 2012 and 2016/2017 were analyzed for the acoustic occurrence of bowhead whales at hourly resolution using a combination of automated and manual analyses. Bowhead whales were acoustically present from autumn throughout the winter months (October-February) and occasionally in spring (March-June), supporting hypotheses that Fram Strait is an important overwintering area. Acoustic presence peaked between mid-November and mid-December with bowhead whale calls recorded almost daily, often hourly for several consecutive days. The observed peak in acoustic presence coincided with the presumed mating period of bowhead whales, starting in late winter, indicating that Fram Strait may also serve as a mating area. Detailed analyses of recordings of a single year and location revealed eight distinct bowhead whale song types, comprising simple songs and call sequences. No bowhead whales were recorded in summer (July-September), indicating that they had migrated to summering areas or resided outside the detection range. Compared to previous studies in western Fram Strait, bowhead whale detections in our recordings were less frequent and recorded songs were less complex. The observed regional differences in bowhead whale acoustic behavior across Fram Strait suggest that eastern Fram Strait may represent a less favorable part of the bowhead whale overwintering area.

Repatriation of a historical North Atlantic right whale habitat during an era of rapid climate change

Climate change is affecting species distributions in space and time. In the Gulf of Maine, one of the fastest-warming marine regions on Earth, rapid warming has caused prey-related changes in the distribution of the critically endangered North Atlantic right whale (Eubalaena glacialis). Concurrently, right whales have returned to historically important areas such as southern New England shelf waters, an area known to have been a whaling ground. We compared aerial survey data from two time periods (2013–2015; 2017–2019) to assess trends in right whale abundance in the region during winter and spring. Using distance sampling techniques, we chose a hazard rate key function to model right whale detections and used seasonal encounter rates to estimate abundance. The mean log of abundance increased by 1.40 annually between 2013 and 2019 (p = 0.004), and the mean number of individuals detected per year increased by 2.23 annually between 2013 and 2019 (R2 = 0.69, p = 0.001). These results demonstrate the current importance of this habitat and suggest that management options must continually evolve as right whales repatriate historical habitats and potentially expand to new habitats as they adapt to climate change.

Machine-Learning Approach for Automatic Detection of Wild Beluga Whales from Hand-Held Camera Pictures.

A key aspect of ocean protection consists in estimating the abundance of marine mammal population density within their habitat, which is usually accomplished using visual inspection and cameras from line-transect ships, small boats, and aircraft. However, marine mammal observation through vessel surveys requires significant workforce resources, including for the post-processing of pictures, and is further challenged due to animal bodies being partially hidden underwater, small-scale object size, occlusion among objects, and distracter objects (e.g., waves, sun glare, etc.). To relieve the human expert’s workload while improving the observation accuracy, we propose a novel system for automating the detection of beluga whales (Delphinapterus leucas) in the wild from pictures. Our system relies on a dataset named Beluga-5k, containing more than 5.5 thousand pictures of belugas. First, to improve the dataset’s annotation, we have designed a semi-manual strategy for annotating candidates in images with single (i.e., one beluga) and multiple (i.e., two or more belugas) candidate subjects efficiently. Second, we have studied the performance of three off-the-shelf object-detection algorithms, namely, Mask-RCNN, SSD, and YOLO v3-Tiny, on the Beluga-5k dataset. Afterward, we have set YOLO v3-Tiny as the detector, integrating single- and multiple-individual images into the model training. Our fine-tuned CNN-backbone detector trained with semi-manual annotations is able to detect belugas despite the presence of distracter objects with high accuracy (i.e., 97.05 mAP@0.5). Finally, our proposed method is able to detect overlapped/occluded multiple individuals in images (beluga whales that swim in groups). For instance, it is able to detect 688 out of 706 belugas encountered in 200 multiple images, achieving 98.29% precision and 99.14% recall.

Weathering the storm: Impacts of hurricane-induced noise on the probability of detecting cetaceans

Hurricanes Dorian (Category 5; 2019), Florence (Category 4; 2018), and Humberto (Category 3; 2019) impacted the soundscape as observed at the Atlantic Deepwater Ecosystem Observatory Network (ADEON) locations in the US Mid- and South Atlantic Outer Continental Shelf. Passive acoustic data obtained from bottom-mounted hydrophones at the ADEON locations were examined to assess changes in ambient sound before, during, and afterhurricane presence. Ambient sound level increased up to 25 dB in the 0.1–7 kHz band which could impact the detectability of cetaceans vocalizing at those frequencies. The probability of detection of fin whales, minke whales, and pilot whales was estimated using empirical ambient sound levels, modelled propagation loss, and pre-defined detection parameters. Detection area, as defined by range of minimum detectability of the cetacean from the receiver, was estimated before, during, and after hurricane presence at each ADEON location. Detection area changed considerably during hurricane presence with site-specific impacts for each of the cetaceans, which may affect estimates of their abundance from passive acoustic recordings. [Study concept, oversight, and funding were provided by ONR Award N00014-16-1-2594 and BOEM under Contract No. M16PC00003, in partnership with ONR and NOAA. Funding for ship time was provided under separate contracts by ONR, Code 32.]

Acoustic detection range of right whale upcalls identified in near-real time from a moored buoy and a Slocum glider.

The goal of this study was to characterize the detection range of a near real-time baleen whale detection system, the digital acoustic monitoring instrument/low-frequency detection and classification system (DMON/LFDCS), equipped on a Slocum glider and a moored buoy. As a reference, a hydrophone array was deployed alongside the glider and buoy at a shallow-water site southwest of Martha's Vineyard (Massachusetts, USA) over a four-week period in spring 2017. A call-by-call comparison between North Atlantic right whale upcalls localized with the array (n = 541) and those detected by the glider or buoy was used to estimate the detection function for each DMON/LFDCS platform. The probability of detection was influenced by range, ambient noise level, platform depth, detection process, review protocol, and calling rate. The conservative analysis of near real-time pitch tracks suggested that, under typical conditions, a 0.33 probability of detection of a single call occurred at 6.2 km for the buoy and 8.6-13.4 km for the glider (depending on glider depth), while a 0.10 probability of detection of a single call occurred at 14.4 m for the buoy and 22.6-27.5 km for the glider. Probability of detection is predicted to increase substantially at all ranges if more than one call is available for detection.

Monitoring fin and blue whales in the lower St. Lawrence Seaway with onshore seismometers

AbstractThe Lower St. Lawrence Seaway (LSLS), in eastern Canada, is an important habitat for several species of endangered baleen whale. As we seek to reduce the hazards that these endangered species face from human activity, there is increasing demand for detailed knowledge of their habitat use. Only a sparse network of hydrophones exists in the LSLS to remotely observe whales. However, there is also a network of onshore seismometers, designed to monitor earthquakes, that have sufficiently high sample rates to record fin and blue whale calls. We present a simple method for detecting band‐limited, regularly repeating calls, such as the 20 Hz calls of fin and blue whales, and apply the method to build a catalog of fin and blue whale detections at 14 onshore seismometers across the LSLS, over approximately a 4‐year period. The resulting catalog contains over 600 000 fin whale calls and almost 60 000 blue whale calls. Individual calls are rarely detected at more than one seismometer. Fin whale calls recorded onshore often consist of multiple seismic phases arriving as a ∼2 sequence. Onshore seismometers provide a valuable, previously unused source of data for monitoring baleen whales. However, in the LSLS, the current seismometer network cannot provide high‐precision whale tracking alone, so a denser deployment of onshore and/or offshore seismometers is required.

Effects of duty cycles on passive acoustic monitoring of southern resident killer whale (Orcinus orca) occurrence and behavior

Long-term passive acoustic monitoring of cetaceans is frequently limited by the data storage capacity and battery life of the recording system. Duty cycles are a mechanism for subsampling during the recording process that facilitates long-term passive acoustic studies. While duty cycles are often used, there has been little investigation on the impact that this approach has on the ability to answer questions about a species' behavior and occurrence. In this study, the effects of duty cycling on the acoustic detection of southern resident killer whales (SRKW) (Orcinus orca) were investigated. Continuous acoustic data were subsampled to create 288 subsampled datasets with cycle lengths from 5 to 180 min and listening proportions from 1% to 67%. Duty cycles had little effect on the detection of the daily presence of SRKW, especially when using cycle lengths of less than an hour. However, cycle lengths of 15–30 min and listening proportions of at least 33% were required to accurately calculate durations of acoustic bouts and identify those bouts to ecotype. These results show that the optimal duty cycle depends on the scale of the research question and provide a framework for quantitative analysis of duty cycles for other marine species.

Effectiveness of surface-based detection methods for vessel strike mitigation of North Atlantic right whales

Increasing commercial and recreational use of the world’s oceans has led to growing concerns about vessel and marine mammal encounters. For endangered species such as the North Atlantic right whale Eubalaena glacialis, reducing the number of vessel strikes is key to improving their protection. In this study, we developed an agent-based model to assess the efficacy of thermal imaging systems as a surface-based whale detection method for vessel strike mitigation. We found that the detection range of such systems is the determining factor for their efficacy and needs to be chosen according to vessel characteristics, such as speed and maneuverability. Furthermore, we found that combining large-scale (e.g. protected zones) and small-scale (e.g. on-board detection systems) mitigation strategies increases protection. Finally, technological improvements are needed to achieve reliable detection ranges beyond what is currently possible so that fast and poorly maneuverable vessels such as ultra-large container ships could benefit from on-board detection systems.

Acoustic detections of beaked whales, narrow-band high-frequency pulses and other odontocete cetaceans in the Southern Ocean using an autonomous towed hydrophone recorder

Encased in a streamlined, flooded housing, a SoundTrap ST300HF hydrophone recording system was towed on voyages to South Georgia Island and the South Sandwich Islands and to the Antarctic Peninsula in December 2019–February 2020. Recordings were analyzed to identify acoustic detections of cetacean species. Acoustically identified species included sperm whales (Physeter macrocephalus), southern bottlenose whales (Hyperoodon planifrons), Arnoux's beaked whales (Berardius arnuxii), killer whales (Orcinus orca), and long-finned pilot whales (Globicephala melas). Acoustic detections also included several recognized types of beaked whale echolocation pulses (BW37/39 and BW58) as well as two likely beaked whale echolocation pulse types that do not match any previous descriptions. Narrow-band high-frequency echolocation signals (NBHF) (typical of porpoises and some dolphin species) were detected in many locations, and one of these coincided with a sighting of hourglass dolphins (Lagenorhynchus cruciger). This study shows the utility of an autonomous towed hydrophone system on a vessel of opportunity to study the distribution of cetaceans in rough seas that are difficult to study by visual survey methods.

An acoustic observatory for sperm whales in the Eastern Mediterranean Sea

A deep-water acoustic observatory for real-time detection and 3D localization of sperm whales has been developed, deployed and operated for 3-month periods in summer 2020 and 2021 off SW Crete in the Eastern Mediterranean Sea. Detection and localization rely on receptions of regular clicks at 3 hydrophones suspended from surface buoys at depths of ∼100 m and ∼1 km apart. Travel times of significant arrivals extracted onboard the buoys are transmitted, together with other supporting data, via mobile broadband to a land-based analysis center where the data from all buoys are combined to enable detection and 3D localization of vocalizing animals exploiting direct and surface-reflected arrivals and using a Bayesian approach. The large separations between hydrophones result in small localization uncertainties for ranges up to ∼6 km, whereas, on the other hand, they pose significant challenges related, e.g., to peak association and synchronization between the buoys. The design and implementation, as well as results from the operation of the system will be presented. [Work supported by Ocean Care – SAvEWhales project.]

Open-source deep learning models for acoustic detection and classification of orcas

Current efforts to acoustically study and monitor killer whales in British Columbia (BC), including the endangered population of Southern Resident killer whales (SRKW), are hampered by the lack of sufficiently accurate sound detection and classification algorithms. Recently, several research groups have reported significant improvements in algorithm performance utilizing deep neural networks. However, for most practitioners, these novel tools remain out of reach. To bridge this gap, we are developing a set of open-source deep learning models for acoustic detection and classification of BC's killer whales. These models will be made publicly available along with expert-curated training and test sets to facilitate further development and applications. We are collaborating with Orcasound to deploy the models on their live data. We are also working together with the PAMGuard developer team to ensure that our models can be seamlessly imported and used in PAMGuard, a widely used open-source software platform for passive acoustic monitoring. In this contribution, we will provide an overview of our deep learning methodology and present preliminary results on model performance, with particular attention to the model's ability to handle diverse and variable acoustic environments and generalize to new “unseen” environments.

Characterizing Southern Resident killer whale calls using a particle filter for frequency contour extraction

With only 74 Southern Resident killer whales (SRKWs) remaining in the waters of the eastern Pacific, understanding these marine mammals is vital to their conservation. Invasive techniques such as focal follows or tagging whales could introduce stress to SRKWs. Instead, passive acoustic monitoring that uses a network of hydrophones one preferred tracking method, particularly in their critical habitat that overlaps the shipping lanes in the Salish Sea. The acoustic data, if processed in real-time using modern deep learning methods, can be used to detect whales and alert ships of the potential for spatial overlap to reduce risks of collision. In this project, we consider the differences in SRKW whale call types from archived recordings of each of J, K, and L pods recorded on regional hydrophones. The whale calls are extracted as functional observations in the time and frequency space from underwater recordings using a particle filter. Functional data analysis (e.g., functional clustering) are performed to characterize the whale calls and represent the variations in call types. Such statistical insights between pod-level call types should be useful for improving machine learning whale detection algorithms, and for identifying SRKW movement in a high ship traffic region of their critical habitat.

Integrating visual and acoustic observations to build an “intelligent” killer whale movement forecast system

Shipping traffic continues to grow in the Salish Sea with considerable marine industrial developments planned for the region. With this comes an increase in the risk of cetaceans being disturbed, harassed, and potentially colliding with commercially operating vessels. One key conservation concern is these waters coincide with designated Critical Habitat for the endangered population of Southern Resident killer whales (SRKW). Monitoring technologies such as hydrophones provide the promise of real-time animal detection and localisation. To use these continuous streams of real-time whale location data, we are developing algorithms that automate acoustic detection and classification of SRKW calls using deep learning models trained on extensive new and updated annotated datasets from the region. We are developing open-source models for pod-level classification, while also differentiating SRKW from other ecotypes and species. We are developing a forecasting system to merge these whale detections with opportunistic visual observations. This is based on sequential data assimilation methods using advanced animal movement models to estimate SRKW pod locations with probabilistic predictions of whale directional movement. These methods will be transformative for shipping by providing forecasting with a lead time of a few hours allowing vessels to adjust their speed or pathway to minimize whale-vessel interactions.

Seasonal Occurrence of Sympatric Blue Whale Subspecies: the Chilean and Southeast Indian Ocean Pygmy Blue Whales With the Antarctic Blue Whale

There are multiple blue whale acoustic populations found across the Southern Hemisphere. The different subspecies of blue whales feed in separate areas, but during their migration to lower-latitude breeding areas each year, Antarctic blue whales become sympatric with pygmy and Chilean blue whales. Few studies have compared the degree of this overlap of the Southern Hemisphere blue whale subspecies across ocean basins during their migration. Using up to 16 years of acoustic data, this study compares the broad seasonal presence of Antarctic blue whales, Chilean blue whales, and Southeast Indian Ocean (SEIO) pygmy blue whales across the Pacific and Indian Oceans. Antarctic blue whales were sympatric with the other two blue whale subspecies during the migrating season of every year. Despite this overlap, Chilean and pygmy blue whale detections peaked earlier during the austral autumn (April–May) while Antarctic blue whale detections peaked later during the austral winter (June). Chilean (Pacific Ocean) and SEIO (Indian Ocean) pygmy blue whales showed similar seasonal patterns in detections despite occurring in different ocean basins. Though we have shown that Antarctic blue whales have the potential to encounter other blue whale subspecies during the breeding season, these distinct groups have remained acoustically stable through time. Further understanding of where these whales migrate will enable a better insight as to how these subspecies continue to remain separate.

Cetacean occurrence offshore of Washington from long-term passive acoustic monitoring

A variety of cetacean species inhabit the productive waters offshore of Washington State, USA. Although the general presence of many of these species has been documented in this region, our understanding of fine-scale habitat use is limited. Here, passive acoustic monitoring was used to investigate the spatial and temporal distributions of ten cetacean species at three locations offshore of Washington. Between 2004 and 2013, a total of 2845 days of recordings were collected from sites on the continental shelf and slope, and in a submarine canyon. Acoustic presence was higher for all species at sites farther offshore. Detections were highest during the fall and winter for blue (Balaenoptera musculus), fin (B. physalus), and humpback whales (Megaptera novaeangliae), likely related to reproductive behavior, while minke whales (B. acutorostrata) were only detected on two days. Odontocetes showed temporal separation, with sperm whale (Physeter macrocephalus) detections highest in spring, Risso’s (Grampus griseus) and Pacific white-sided dolphins (Lagenorhynchus obliquidens) highest in summer, and Stejneger’s beaked whales (Mesoplodon stejnegeri), Cuvier’s beaked whales (Ziphius cavirostris), and the BW37V signal type highest in winter or spring. There was interannual variation in detections for most mysticete species, which may be linked to oceanographic conditions: blue and fin whale detections increased during 2007 and 2008, and fin and humpback whale detections increased in 2011. These results inform our understanding of cetacean behavior and habitat use in this region and may aid in the development of conservation strategies suited to the dynamic conditions that drive cetacean distribution.

Acoustic classification of false killer whales in the Hawaiian islands based on comprehensive vocal repertoire.

Use of underwater passive acoustic datasets for species-specific inference requires robust classification systems to identify encounters to species from characteristics of detected sounds. A suite of routines designed to efficiently detect cetacean sounds, extract features, and classify the detection to species is described using ship-based, visually verified detections of false killer whales (Pseudorca crassidens). The best-performing model included features from clicks, whistles, and burst pulses, which correctly classified 99.6% of events. This case study illustrates use of these tools to build classifiers for any group of cetacean species and assess classification confidence when visual confirmation is not available.

Robust North Atlantic right whale detection using deep learning models for denoising.

This paper proposes a robust system for detecting North Atlantic right whales by using deep learning methods to denoise noisy recordings. Passive acoustic recordings of right whale vocalisations are subject to noise contamination from many sources, such as shipping and offshore activities. When such data are applied to uncompensated classifiers, accuracy falls substantially. To build robustness into the detection process, two separate approaches that have proved successful for image denoising are considered. Specifically, a denoising convolutional neural network and a denoising autoencoder, each of which is applied to spectrogram representations of the noisy audio signal, are developed. Performance is improved further by matching the classifier training to include the vestigial signal that remains in clean estimates after the denoising process. Evaluations are performed first by adding white, tanker, trawler, and shot noises at signal-to-noise ratios from -10 to +5 dB to clean recordings to simulate noisy conditions. Experiments show that denoising gives substantial improvements to accuracy, particularly when using the vestigial-trained classifier. A final test applies the proposed methods to previously unseen noisy right whale recordings and finds that denoising is able to improve performance over the baseline clean-trained model in this new noise environment.

Cetacean occurrence in the Gulf of Alaska from long-term passive acoustic monitoring

The Gulf of Alaska is an important habitat for a diverse array of marine mammals, many of which were severely depleted by historical whaling. To study current cetacean distributions in this region, passive acoustic monitoring was used to detect species-specific call types between 2011 and 2015 at five locations spanning the continental shelf, slope, and offshore seamounts. Spatial and temporal detection patterns were examined for nine species to compare differences in behavior and habitat use. Mysticetes showed seasonal increases in calling that indicated possible behavioral shifts between feeding and breeding in blue (Balaenoptera musculus), fin (B. physalus), and humpback (Megaptera novaeangliae) whales, and matched known migration timing of gray whales (Eschrichtius robustus). Interannual changes in blue and fin whale calling may relate to the marine heat wave that began in 2013 and lasted through the end of the monitoring period. Odontocete detections revealed unique spatial distributions, with killer whales (Orcinus orca) most common on the continental shelf and sperm whales (Physeter macrocephalus) most common on the continental slope, where detections occurred year-round. Beaked whales showed both spatial and temporal separation: Baird’s beaked whale (Berardius bairdii) detections were highest at Quinn Seamount in the spring, Cuvier’s (Ziphius cavirostris) at Pratt Seamount in winter, and Stejneger’s (Mesoplodon stejnegeri) on the continental slope in the fall. The year-round presence of many species highlights the ecological importance of the Gulf of Alaska and the spatiotemporal information reported here should inform future conservation efforts.

Beluga whale detection in the Cumberland Sound Bay using convolutional neural networks

The Cumberland Sound Beluga is a threatened population of belugas and the assessment of the population is done by a manual review of aerial surveys. The time-consuming and labor-intensive nature of this job motivates the need for a computer automated process to monitor beluga populations. In this paper, we investigate convolutional neural networks to detect whether a section of an aerial survey image contains a beluga. We use data from the 2014 and 2017 aerial surveys of the Cumberland Sound, conducted by the Fisheries and Oceans Canada to simulate two scenarios: (1) when one annotates part of a survey and uses it to train a pipeline to annotate the remainder and (2) when one uses annotations from a survey to train a pipeline to annotate another survey from another time period. We experimented with a number of different architectures and found that an ensemble of 10 CNN models that leverage Squeeze-Excitation and Residual blocks performed best. We evaluated scenarios (1) and (2) by training on the 2014 and 2017 surveys, respectively. In both scenarios, the performance on (1) is higher than (2) due to the uncontrolled variables in the scenes, such as weather and surface conditions.