A machine-learning derived model of seafloor sediment accumulation

Abstract

Previous studies regarding the depositional pattern and quantity of accumulated seafloor sediment tend to be regional, limited in scope and involving costly and time-consuming geologic field campaigns and laboratory work. Presented herein is a global map of predicted modern (postindustrial, 20th and 21st century) oceanic mass accumulation rates of 5-arc-minute pitch and in log10-space, trained on observed marine mass accumulation rates from 43 peer reviewed sources (n = 1744) and predicted using a k-nearest neighbor geospatial algorithm. The resultant model predicts ~3.3 × 104 Mt. yr−1 of sediment accumulating onto the sea floor (R2 = 0.88). Most sediment accumulates proximal to major river outlets and deltas. Continental regions with the highest sediment accumulation are Asia and Oceania. This model is the first of its kind to predict the rate and quantity of sediment accumulating on to the ocean floor, globally, using decades of regional real-world observations. The generated global map of modern, benthic mass accumulation rates also serves to highlight areas of interest for future study in related fields, such as sediment dynamics and seafloor stability.