Mapping of Algae and Seagrass using Spectral Imaging and Machine Learning

Informations

Funding country

Norway

Acronym

-

URL

-

Start date

1/1/2020

End date

12/31/2024

Budget

983,385 EUR

Fundings

Abstract

The MASSIMAL project aims to develop new methods for mapping marine underwater vegetation such as seagrass, macroalgae, and maerl. These vegetation types are part of the "blue forests" which are home to a large range of marine species. They also contribute significantly to primary production, carbon capture, and absorption of dissolved nutrients. The blue forests are threatened by human activity, climate change and overgrazing by sea urchins, and new tools are needed for monitoring and studying how and why these ecosystems change. Using a hyperspectral camera mounted on a drone, the seafloor is imaged from 20-100 meters above the sea surface. By combining the hyperspectral images with manual sampling of the vegetation, machine learning algorithms can produce detailed maps of e.g. the different species distribution, vegetation density and physiological state. Data has been collected from numerous locations along the Norwegian coast, in areas close to Bodø, Larvik and Vega. The datasets represent a large variation in plant species, nature types, weather conditions and optical water properties. Hopefully, this variation will enable training of machine learning algorithms that are robust and perform well in many different settings. Several different methods for documenting underwater vegetation and nature types (so-called "ground truth") have been tested during field campaigns. During early stages, such documentation was done by defining a small number of transects and photographing these in great detail (while snorkeling in the surface). In later field campaigns, the documentation methods have been modified to cover larger areas and more of the natural variation. Additional methods for filming the sea floor have also been used, including operating a camera from a boat, imaging while diving, snorkeling with a long pole (to position the camera close to the seafloor) and filming from underwater drones and autonomous surface vehicles. The data collected during campaigns must be organized, post-processed and annotated in order to use it for training machine learning models. Developing methodology to do this efficiently and accurately has been an important and time-consuming part of the project. Several data sources need to be combined in order to annotate the hyperspectral images. Annotated datasets will be published as part of the project results. Preliminary results from the project indicate that it is possible to distinguish between different vegetation and nature types based on their spectral "fingerprints". A master thesis based on data from the project also presents promising results regarding classification of species within a limited geographic area. However, the preliminary results also show that water surface waves, water reflections and loss of light in the water column all contribute to reduced signal-to-noise ratio, and that this can cause reduced classification accuracy. Further work on machine learning will focus on making the algorithms robust to such "noise" and general enough to work across several different locations. The last research campaigns in the project will be conducted in 2023, in areas close to Smøla and Larvik. The Smøla area is subject to commercial harvesting of kelp, and by mapping the area, the project is hoping to measure and quantify regrowth of kelp in the period 1-5 years after harvesting. The Larvik area was imaged during a campaign in 2021, and by revisiting it in 2023, the project is hoping to map and measure how the vegetation has changed over time. The last two years of the project (2023-2024) will mainly be focused on development of machine learning algorithms and publication of results. Public outreach will also be given high priority.