Coral Cartography: Unleashing AI for Unprecedented Cold-Water Coral Mapping

Project overview

Cold‑water corals are vital components of deep‑sea ecosystems across the Atlantic Ocean. They create structurally complex habitats that support high biodiversity and contribute to the resilience and functioning of ocean systems. Despite their importance, large areas of the Atlantic—particularly in deeper and less‑surveyed regions—remain poorly characterised, with significant gaps in our understanding of coral distribution, density and ecological role.

At the same time, decades of deep‑sea exploration have generated vast archives of underwater imagery that remain under‑utilised due to the time‑intensive nature of manual analysis. This limits the speed at which scientific evidence can be translated into effective conservation and management.

The Coral Cartography project addresses these challenges by combining marine ecology, data science and artificial intelligence to produce the first Atlantic‑scale, high‑resolution maps of cold‑water coral habitats. From Plymouth Marine Laboratory’s perspective, the project aligns strongly with our mission to deliver interdisciplinary, digitally enabled science that supports sustainable ocean management.

Project aims

The project aims to transform how cold‑water coral ecosystems are observed, analysed and represented at ocean‑basin scale by:

• Creating a standardised, quality‑controlled dataset of cold‑water coral imagery and annotations.
• Developing and applying AI‑based tools to detect, classify and quantify cold‑water corals from seabed imagery.
• Producing spatially explicit maps of coral distribution, density and diversity across the Atlantic Ocean.

PML’s contribution

Plymouth Marine Laboratory contributes expertise in marine biodiversity, environmental data integration and artificial intelligence for ocean science. PML researchers support the development of robust analytical workflows that combine biological observations with physical and environmental data, enabling scalable and repeatable mapping of vulnerable marine ecosystems.

Through this work, PML helps ensure that project outputs are scientifically robust, policy‑relevant and aligned with international best practice in open and reproducible science.

How the project works

Image data integration and quality assurance
Existing and newly collected deep‑sea imagery is brought together from across the Atlantic. Consistent annotation standards and quality‑control procedures are applied to ensure the data are suitable for large‑scale analysis and machine learning.

AI‑enabled image analysis
Deep‑learning models are trained using expertly annotated imagery to automatically identify and quantify cold‑water corals. These tools dramatically accelerate the analysis of large image archives while maintaining scientific accuracy through expert validation.

Habitat mapping and modelling
Biological data are combined with environmental variables such as depth, temperature and oceanographic conditions to generate habitat suitability models. These models underpin Atlantic‑wide maps of cold‑water coral distribution and density.