Abstract

Cold-water coral (CWC) reefs represent significant deep-sea carbon reservoirs, yet province-scale storage volumes remain poorly quantified. This study develops two bathymetry-based models to estimate the total carbonate volume and sequestered carbon of the Logachev Mound Province (LMP) on the southern margin of Rockall Bank.

The Linear Approximation Model (LAM) reconstructs the basal seafloor by interpolating across generated coral gaps in a 25 m resolution multibeam bathymetric grid, yielding a total mound volume of 10,387,020,800 m³. Carbon mass estimates derived from three drill cores, partitioned into Holocene and pre-Holocene components, produce an average of 2.64 Gt C stored across the province.

The Sigmoid Approximation Model (SAM), developed to isolate positive-relief coral volume from geometric error, returned an unreliable volume of 17 m³ but confirmed a strong sigmoidal fit to the Rockall margin bathymetry with a mean R² of 0.9943. These results provide the first province-scale carbon storage estimate for the LMP, offering a computationally efficient, reproducible baseline for stochastic carbon flux modeling within the LMP and for other CWCs.

Introduction

Cold-water corals are a vital ecosystem in the deep sea. The reefs form biodiverse oases in an otherwise sparsely populated region of the oceans, supporting rich assemblages of fish and invertebrates. Despite an increased interest by the scientific community, CWCs remain understudied and largely unknown to the public. As CWC ecosystems are degrading globally, more studies should be conducted for the preservation of data before these archives are lost.

Quantifying their role in carbon storage carries particular urgency. Determining the volume of a geological reservoir is a requirement for evaluating how effectively it buffers stochastic fluxes of carbon through time, because storage capacity directly controls the system's ability to absorb shocks without losing accumulated material. For climate models, this provides better constraints on the deep-sea carbon cycle. For conservation, it helps determine what is worth preserving and provides incentive for increased research while reefs remain intact.

Carbon reservoir size is largely undetermined for CWC reefs. This limits stochastic flow models that seek to predict how carbon moves through the deep ocean. Studies on CWC carbonate budgets have advanced considerably, with mass accumulation rates calculated for individual mounds and full carbonate budgets constructed for specific sites. Yet total province-scale storage volumes remain poorly constrained.

The challenge lies in moving from local accumulation rates to whole-reef volumes. Approaches have historically looked at surface area carbon storage, but volume estimation is challenging due to the ambiguity present in seismic profiles of the region. A purely morphological alternative was demonstrated by de Haas et al., who modelled an approximated seafloor surface beneath the reef from multibeam bathymetry by interpolating the surrounding ambient seabed beneath individual mound clusters. This bypassed the seismic problem entirely, though it captured only the positive relief volume and not any buried mound root. To date, no methodology has been tested to estimate the total carbon storage of the entire LMP at province scale.

This study seeks to generate a model capable of bypassing the seismic ambiguity by generating a basal surface in order to produce a volume value, and by using drill core data, quantify the sequestered carbon the LMP contains.

Study Area

The Logachev Mound Province is situated on the southern margin of Rockall Bank, at the southwestern edge of the Rockall Trough in the northeast Atlantic. The LMP occupies the upper continental slope at depths between approximately 500 and 1,200 m, where approximately 500 individual mounds have been mapped. The structural composition of the mounds reflects their biological origin: sediment samples show that the mounds consist primarily of pale-coloured carbonate muds, mainly aragonite, with live and dead cold-water corals at the seafloor and buried dead coral frameworks beneath.

This vertical stratigraphy, with living reef at the surface transitioning to compacted dead framework and carbonate sediment buried within the reef, is central to the carbon quantification approach developed in this study. Surface surveys of the LMP indicate that approximately 45% of the mound surface is covered by dead coral framework, 3% by live coral framework, and the remaining 51% by fine sediments.

Mound growth at the LMP has been continuous for approximately the last 11,000 years. Prior to this Holocene growth phase, several hiatuses and nonconformities exist within the mound record, with the most recent unconformity correlating across multiple mound provinces at the Irish margin, providing an end-Pleistocene marker. Since carbonate composition and sedimentation rates differ between the pre-Holocene and Holocene layers in the LMP, the carbon calculations must account for this.

Why This Matters

The deep sea remains poorly represented in global carbon cycle models. Deep-sea processes are largely absent from most carbon cycle models and rarely considered in climate discussions, and Earth System Models rely on highly simplified representations of marine ecology. Province-scale carbon stores such as the LMP are therefore not fully accounted for in global models. Providing a quantified volume for the LMP contributes to constraining deep-sea carbon stocks that global models currently have not integrated.

Cold-water coral mounds warrant specific attention as carbon reservoirs because they accumulate significantly more carbon than the surrounding ambient seafloor. Previous work found that during mound formation phases, CWC mounds accumulate between 15 and 19 times more carbon than the adjacent seafloor. This justifies treating the mound province as a discrete reservoir rather than averaging carbon storage across the broader seabed.

Generating a reproducible volume for the LMP also provides a metric that can inform marine spatial planning and conservation prioritisation. In a rapidly changing ocean, knowing how much carbon a given reef province stores helps determine what is worth protecting. While this study does not prescribe specific policy, the result offers an evidence base that managers and policymakers can use when evaluating the climate-regulation services provided by cold-water coral ecosystems.

Methodology

Seafloor bathymetry was retrieved through the Irish Government's public database for geospatial data. The dataset is the 25 m resolution product of Irish coastal waters, the highest resolution dataset available for the Logachev coral reef. Drill core data for M2001-28, M2001-43 and M2003-23 were sourced from van der Land et al., selected because the carbonate contents split between Holocene and pre-Holocene are presented clearly in that publication.

ArcGIS Pro was used to develop the models. The polygon tool was used to manually outline the area of the CWCs present within the LMP. The inverse selection generated the bathymetry raster with the coral area removed, for use as the ambient seafloor input to the model. For the coral polygon, the reef area was intentionally drawn slightly larger than the coral reef itself, biasing the model toward overprediction.

A Python script was written as a tool in ArcGIS Pro. The script searches for pixels within the bounds of the raster that border no-value pixels and linearly interpolates across the gaps in the dataset. This generated both horizontal and vertical interpolation lines, and the depth value for each pixel was taken as the average of these two lines, producing a continuous basal surface for the LMP. Volume was calculated by subtracting the depth of the basal surface from the depth of the reef surface at each pixel. Each pixel represents a 25 m × 25 m area, and these were summed to give the total volume.

A carbon mass calculator was constructed, taking as input the molar mass fraction of carbon in calcium carbonate, the density of aragonite and the density of calcite. The calculator fits each core to the entire volume separately and reports the extrapolation for each core.

Results

The linear approximation model successfully generated the basal surface. The LAM calculated a total volume of 10,387,020,800 m³ for the LMP. The LAM was designed to be computationally cheap and ran in under 1 minute on a computer with an Intel Core i5 CPU with integrated graphics as the primary processing unit.

The carbon sequestration calculation provided a range of 1.84–3.49 Gt C from core M2001-28, 2.32–3.39 Gt C from core M2001-43 and 1.29–3.53 Gt C from core M2003-23, with an average of 2.64 Gt C across all three cores.

To contextualize these figures, the global anthropogenic carbon emission for 2025 was 11.6 ± 0.9 Gt C. According to the LAM results, the LMP stores approximately 22.8% of the 2025 anthropogenic carbon emission.

The SAM calculated a volume of 17 m³, with a mean R² of 0.9943. To appreciate how small this prediction is: over the entire ~70 km-long reef province, each pixel is 25 m × 25 m. For the SAM to yield 17 m³, the average relief per pixel across the entire reef would need to be ~3 cm.

Discussion

The LAM provides a linear province-scale volume by stitching the generated coral gap line by line to construct the basal surface of the ambient seafloor. The finding that a province-scale linear approximation is feasible reinforces the central research question: a bathymetry-based model can bypass seismic ambiguity and generate a cheap and reproducible carbonate storage estimate.

However, the results also confirm that a linear interpolation over a curved margin will systematically incorporate the ambient seafloor into the mound volume. Since the LMP contains buried coral mounds, the amount of incorporated ambient seabed is unknown. The SAM was therefore designed to generate a volume for the positive relief of the corals, assuming the LMP lies atop a curved, convex-upward seafloor. While it did not return a usable volume, it confirmed the sigmoidal profile of the margin and identified why the LAM overpredicts.

Four major modeling limitations were identified. First, the models consistently overpredict volume, both because the 25 m resolution smooths small changes in reef surface and because the linear assumption captures ambient seafloor. Second, selecting the boundary of the coral reef proved challenging due to data resolution. Third, a byproduct of the linear approximation is the smoothing of buried topographic features beneath the reefs. Fourth, the cumulative error in the model is unknown, as the LAM has no independent dataset against which to validate its output.

On the calculation side, drill core data from a single mound were generalized to the entire LMP, the growth hiatus was taken as the pre-Holocene / Holocene boundary, the existence of buried coral reefs was not accounted for, and drill cores preferentially sample the Holocene cap rather than the reef body as a whole. Future studies can improve the calculations by incorporating CWC growth patterns and reef accretion geometries into the equations.

Conclusion

This study set out to quantify the volume and stored carbon within the Logachev Mound Province. Two models were constructed. The LAM successfully produced a volume of 10,387,020,800 m³. The SAM, while unsuccessful, demonstrated through an overall R² of 0.9943 that the Rockall margin follows a sigmoidal clinoform profile. The estimate of 2.64 Gt C represents the first province-scale carbon storage figure for the LMP, filling a gap that existing carbonate budget studies at the mound or single-site scale have not addressed.

The linear assumption introduces a geometric error that currently remains unconstrained due to the lack of independent datasets with which to validate the LAM. Future work should prioritise minimum curvature interpolation as an improvement on the sigmoidal approach and apply the LAM to other CWC reefs. Integrating the three drill cores as an averaged dataset rather than treating them individually would also strengthen the carbon mass estimates.

As cold-water coral ecosystems face increasing anthropogenic pressure, quantifying their role as carbon reservoirs becomes critical for both climate modeling and conservation policy. By developing more models for province-wide carbon storage, we can better understand the role CWCs play in the global carbon system and seek to preserve these unique ecosystems.

Full Document

📄 Download the full thesis (PDF)