Improving marine sediment carbon stock estimates: the role of dry bulk density and predictor adjustments

Abstract

Continental shelves are critical for the global carbon cycle as they store substantial amounts of organic carbon (OC). Shelf sediments can also be subject to considerable anthropogenic pressures, offshore construction and bottom trawling for example, potentially releasing OC that has been sequestered into sediments. As a result, these sediments have attracted attention from policy makers regarding how their management can be leveraged to meet national emissions reduction targets. Spatial models offer solutions to identifying organic carbon storage hotspots; however, regional predictions of OC often rely on global scale predictors which may have biases on smaller scales, reducing their utility for practical management decisions. In addition, estimates of dry bulk density (DBD), an important factor in calculating OC stock from sediment OC content, are typically derived from an empirical relationship developed in one region and applied elsewhere, rather than from local in situ data, leading considerable uncertainty in regional OC stock estimates. We compared the performance of two spatial models of OC stock. The first used unadjusted predictors and a commonly used empirical relationship to estimate DBD. The second spatial model incorporated bias-adjusted predictors and a machine learning DBD model, trained on in situ DBD data. The adjusted model predicted a total OC reservoir of 46.6 ± 43.6 Tg in the top 10 cm of sediment in the Irish Sea, which was 31.4 % lower compared to unadjusted estimates. 70.1 % of the difference between adjusted and unadjusted OC stock estimates was due to the approach for estimating DBD. These findings suggest that previous models may have overestimated OC reservoirs and highlight the influence of accurate DBD and predictor adjustments on stock estimates. These findings highlight the need for increased in situ DBD measurements and refined modelling approaches to enhance the reliability of OC stock predictions. This study provides a framework for refining spatial models and underscores the importance of reducing uncertainties in key parameters to better understand and manage OC storage potential of marine sediments.