Organic carbon storage in northwest European shelf subtidal sediments: towards a consensus map

Quantifying the various pools of marine carbon is a fundamental initial step towards establishing the potential value of ocean carbon storage in climate change mitigation. While subtidal sediments hold less carbon per unit area than the coastal ‘traditional blue carbon habitats’ (mangrove, saltmarsh, seagrass), their large extent makes them an important component of the UK’s blue carbon inventory. Global and regional maps of seabed organic carbon have been published, but they do not provide a consensus view of northwest European Shelf subtidal sediment carbon storage. Improved seabed carbon maps, using predictive mapping approaches, require high confidence measurements of sediment organic carbon content with appropriate good spatial coverage as well and overall sampling of predictor ‘parameter space’. They also rely on availability and selection of appropriate high-resolution predictor variables. Our efforts to map seabed organic carbon have two, sometimes complimentary sometimes conflicting, objectives: (i) to generate the most accurate map possible, and (ii) to provide our process-focused understanding of how and why carbon content varies and which parameters control that variability in the context of seabed biogeochemical cycling.

We initially applied a Gradient Boosting Machine Learning method to generate a new predicted seabed carbon map for the northwest European Shelf using more than 2,000 observations of near-surface sediment (0-2 to 0-10 cm below seafloor) organic carbon content. We explored the importance of various methodological decisions on the importance of different predictors (e.g., mud content) and the predictive power and accuracy of the model. We discuss the differences and similarities of our data product and mapping approach with previous seabed carbon maps, highlighting the impact of knowledge gaps on potential use by policymakers.