Comparing a Machine-Learning Based Model of Benthic, Marine Mass Accumulation with Fluvial Sediment Flux into the Global Oceans

We present a global map of modern (postindustrial, 20^th and 21^st century) oceanic mass accumulation rates (MAR) of 5-arc-minute pitch and in logarithmic space, based on marine mass accumulation rates from ~45 peer reviewed sources (n = 690) and predicted using a k-nearest neighbor algorithm using the United States Naval Research Laboratory's Global Predictive Seabed Model (GPSM). GPSM predicts (R² = 0.88) a substantial mass of sediment accumulating on the sea floor, ~3.1 x 10⁵ Mt yr^-1, one order of magnitude larger than previously published fluvial sediment fluxes to the global oceans. The predicted MAR pattern generally agrees with previously published literature, with the majority of sediment accumulating proximal to major river outlets and on wide coastal shelves, declining to functionally zero in oceanic basins. The GPSM prediction indicates significant sediment accumulation along the West African coast, contrary to previously published fluvial sediment discharge data. The amount of sediment exiting Africa into the Atlantic Ocean may be considerably underestimated by previous studies; as such, this region is highlighted as an area of interest for future research.