Coral reefs are true treasures of biodiversity. It is estimated that over half a billion people worldwide depend on them for food, income and protection. However, due to human activities such as overfishing, pollution and global warming, biodiversity and coastal habitats are under severe pressure, and have been reduced by between 30% and 60%.
Protecting ecosystems as exceptional as coral reefs is now an urgent imperative. In the Indian Ocean, new solutions are being tested, combining observations at sea, technological innovations and digital developments to develop new indicators and effective decision-making tools.
by Carole Saout-Grit et Laurie Henry
Using mathematical and digital tools to protect the oceans
In October 2022, Matteo Contini moved to Réunion Island to join the team at Ifremer’s Indian Ocean Delegation (DOI).
A graduate of Milan Polytechnic in mathematical engineering and applied statistics, he had already worked on the island in September 2020, just after the Covid period, as a Civic Service Volunteer (VSC). At that time, he spent two years developing the Seatizen citizen science project, with the aim of developing data collection on tropical reefs through the participation of sea users in order to quantify changes in habitats and species in these ecosystems.
But this time, his task looks set to be longer and more complex. He is starting a new three-year project, a thesis during which he will use his skills as a mathematician and data scientist to monitor and protect the Indian Ocean.
Matteo Contini © M. Ravitchandirane
An in-depth analysis of the evolution of habitats in the Indian Ocean
Often referred to as the ‘rainforests of the sea’, coral reefs are home to thousands of marine species. They are also the source of many vital ecosystems that protect coastlines from storms and erosion, and provide local communities with a wealth of resources and recreational opportunities.
Faced with the challenge of protecting these exceptional ecosystems, traditional surveillance methods are showing their limitations. The underwater visual surveys (UVC) often carried out require the experience of professional divers. They are often costly and only cover small geographical areas. To overcome these shortcomings, new technologies incorporating hyperspectral imaging*, photogrammetry* and artificial intelligence offer promising prospects for large-scale, more accurate monitoring of coral reefs.
During his thesis, Matteo Contini will develop an innovative approach to monitor more specifically the evolution of tropical reefs in the Indian Ocean. He will attempt to produce detailed temporal maps of marine habitats (corals, seagrass, sand, rocks) and classes of the Global Coral Reef Monitoring Network (GCRMN), and to identify emblematic species such as sea cucumbers and sea urchins.
Diagram of the principle of image analysis using Deep Learning
To achieve this objective, it will collect a wide range of data at different scales: large-scale satellite data (over 100km) accessible free of charge since the first Landsat satellite programme in 1972; medium-scale data (less than 10km) collected by aerial drones; fine-scale data (less than 1km) collected by autonomous surface surveys (ASV) and a citizen science project to identify species and habitats.
Part of the underwater data will be annotated and used to train a fine-scale deep learning algorithm. The predictions made by this first neural network will be used to feed a second network, which will analyse images on the scale of drones. The same process will then be applied to satellite images, capable of providing precise clues as to the evolution of coral habitats and the distribution of species.
Funded for 3 years by the PPR Ocean & Climate Priority Research Programme, this thesis is a response to the challenge of developing innovative, multi-disciplinary, multi-parameter, multi-scale and multi-actor observation and modelling programmes. Co-directed by Sylvain Bonhommeau (Ifremer/RBE DOI) and Alexis Joly (INRIA/LIRMM), the overall aim is to provide better information to marine environment managers and, in the longer term, to strengthen coral reef conservation efforts worldwide.
3 Questions to Matteo Contini
Why did you decide to do a PhD in marine sciences?
‘Ever since I was a child, I’ve been fascinated by the oceans and their complex ecosystems. During my studies, I discovered how mathematics could be applied to real-life problems, including in the environment. Doing a thesis in marine science combines my technical skills with my deep interest in marine conservation.’
What made you want to do this thesis? What were your motivations?
I developed this thesis topic with Sylvain Bonhommeau and Alexis Joly attracted by his interdisciplinary approach, combining marine ecology, AI and innovative technologies. My previous missions at Ifremer awakened in me a passion for the application of Deep Learning and image processing techniques. The opportunity to monitor the evolution of tropical reefs and come up with practical solutions to protect them particularly appealed to me.
How do you see your future after this thesis?
After this thesis, I want to continue in marine research, focusing on the conservation and sustainable management of marine ecosystems. I’d like to use my skills in AI and data analysis to develop innovative tools.