Acoustic monitoring using bioacoustics and AI

Sound is revolutionizing the monitoring of biological diversity. Acoustic monitoring can be used both on land and in water, and gives us knowledge about nature in a fast and cost-effective way. NINA uses acoustic monitoring in several pilot projects. 

New and promising method

The acoustic monitoring of species, ecosystems and human activity is a new and promising method for efficiently monitoring the environment in general. With the help of sound recordings from an ecosystem, we can, for example, detect alien or endangered species, describe species phenology, and document changes in ecological status or human impacts on nature.

Acoustic monitoring also works when visibility is poor, such as in murky waters, dense forests, in fog or at night. Monitoring relies on small recorders equipped with a microphone or hydrophone, called sound loggers. The loggers are equipped with a battery and solar panel, so they can operate for a long time without supervision. Since the system requires little maintenance, researchers can avoid repeat visits that can disturb the area being monitored, which makes this approach particularly suitable for harsh Norwegian conditions.

Although the technology has been used for a long time in studies of individual species, it is only recently that acoustics have become more widely used. Advances in computer technology, digitization and artificial intelligence provide great opportunities to further develop and complement existing methods for data collection and analysis.

In a time of rapid environmental change, remote sensing methods are especially important for monitoring and nature management because they continuously produce data streams that can be analysed across different spatial and temporal scales.

NINA is testing bioacoustics in several pilot projects

Acoustics is a very useful method for addressing a number of research questions, and NINA leads or participates in many national monitoring projects where both bioacoustics and ecoacoustics could contribute new and important information.

The inclusion of sound can provide valuable additional information and a broader database for assessing biodiversity and the state of health in different ecosystems.

Insect monitoring

NINA recently conducted a pilot study in connection with a national insect monitoring project. Here, the diversity of the soundscape was compared to the diversity documented in DNA collected from insects.

Alien fish

In another project, the idea has been to identify and attract the round gobi (Neogobius melanostomus). This is an alien fish species that has not yet been detected in Norway, but which is expected to be introduced via ship ballast water. Since the round gobi makes a lot of noise, it is well suited to this kind of pilot project.

Several other projects are being planned, including microphone recorders set up in combination with a game camera, to see how human activity affects wildlife.

The digital revolution allows for many possibilities. With the technology and analysis methods available today, we can already monitor nature at a large scale using sound. By developing the technology further, we can find even better approaches for nature monitoring programmes.

Bioacoustics as a management tool

Nature is facing increasing threats from the influence of human activity, both in Norway and globally. This is why we need to know more about how human activity contributes to influencing the state of ecosystems.

Efforts are well underway to ensure that Norway’s ecosystems are in good condition, mainly through preparing comprehensive freshwater and marine management plans. Work on comprehensive management plans for terrestrial ecosystems remains in its infancy. In terrestrial ecosystems, we assess conditions using knowledge-based systems for determining what factors contribute to a good ecological status. Understanding the current state of ecosystems forms the basis for the preparation of management plans, for which we need more information.

The increased need for information requires better use of existing data, such as through the increased availability of relevant time series on biodiversity, but also better statistical methods and models. There is also a need to develop monitoring methodologies, such as those relying on remote sensing and ground-based sensors.

Acoustic monitoring of fauna and of human activity is a promising method to make monitoring more efficient, but also as a way to collect new information to increase our understanding of how nature is affected by human activity.

Recent developments in computer technology, digitization and artificial intelligence provide great opportunities to further develop and complement current methods for data collection and analysis.

Biodiversity is threatened by climate change, habitat degradation, invasion of alien species and overexploitation, all of which can be investigated using an acoustic approach. Changes in the phenology of plant and animal populations can lead to changes in the soundscape throughout the year, associated with many different ecological functions (such as reproduction, spawning, predation or migration). Habitat degradation, eutrophication events or overexploitation can cause changes in the amount of sound from both terrestrial and aquatic species. Alien species can be identified by the sudden presence of sounds that were not previously common. Monitoring both individual species and entire ecosystems can provide important information for management, and is particularly suitable for otherwise inaccessible environments.

Sensor technology and routines for data processing are continuously being developed, and are constantly improving. Modern machine-learning techniques make it possible to extract detailed information from the soundscape, which makes acoustics an important contribution to the toolbox for monitoring the Norwegian environment.

Contact

Publications

Birds of Norway

Live tweets from the Sound of Norway Project powered by Bugg | Research from @NINAnature & @cambridge_uni

Follow @norwaybirds on Twitter

Projects

SoundWell

Project leader: Carolyn Rosten (NINA)

Funding: The Research Council of Norway

Project period: 2018-2022

As fish farming is becoming increasingly industrialised and the industry is looking towards applications of emerging digital technologies and automated systems. Continuous data on the condition of the fish are necessary to monitor, control and document operational procedures or management regimes in real-time. However, acquiring such information from a net-pen with up to 200 000 fish is challenging. Sound production in fishes is gradually becoming better documented. Monitoring the soundscape in fish farming cages and observations of sound production represent an untapped resource for monitoring the status (e.g. hunger or stress) of the fish. Non-invasive monitoring of fishes’ status has great potential to improve both animal welfare and production efficiency. Passive sound monitoring can be performed when visibility or darkness is poor, and for large groups of animals in real time. The SoundWell II project will provide the basis necessary for the future development of a non-invasive acoustic status indicator for farmed salmon that can be used to improve fish welfare and feeding routines.

OWITOOLS

Project leader: Merete Bjørgan Schrøder / SINTEF

NINA-contact: Carolyn Rosten

Funding: FHF - Norwegian Seafood Research Fund

Project period: 2020-2021

The project will establish technological tools for objective and technology-neutral measurement of fish welfare, both for monitoring the fish through production and in connection with handling and treatment operations. 

Sound of Norway

Project leader: Carolyn Rosten

Funding: The Norwegian Environment Agency, NINA and others

Project period: 2020-ongoing

Natural ecosystems are under increasing pressures from large scale agriculture and climate change, leading to global biodiversity crisis. Collecting accurate and reliable biodiversity data is an essential first step to designing sustainable routes forward, yet it is surprisingly difficult to do. Whilst traditional ecological monitoring techniques are labour intensive, recently scientists have been using the sounds of an environment to map its biodiversity (eco-acoustic monitoring). Collecting audio using autonomous recording devices and analysing data using machine learning techniques allows ecological data to be collected on vast spatial and temporal scales, transforming our insight into species distributions, behaviours, and patterns. 

Since 2020, we have been piloting the Sound of Norway project; a first of its kind nationwide real-time eco-acoustic monitoring network. Our autonomous audio monitoring devices are powered by solar panels and transmit acoustic data in near real-time over a mobile internet connection. With the data we can identify birds by their calls and track how they move across the country and across the course of the seasons. We apply machine learning models to identify the bird calls automatically, allowing us to collect data on far larger scales than were previously possible in ecological studies of this kind. In 2021 we collected and analysed approximately 60,000 hours of data from over 40 sites across the network