Machine-learning predicted logarithmic values of marine mass accumulation rates on the seafloor

This data set consists of log10 values of machine-learning-based mass accumulation rates and linear uncertainty on the seafloor, globally. The Geospatial Predictive Seafloor Model (GPSM) of the United States Naval Research Laboratory was trained on real-world observations from 43 peer-reviewed sources (n = 1744) to predict marine mass accumulation rates on a 5-arc-minute map using a k-nearest neighbor algorithm. Original (non-log space values) are in g/cm2/yr. The model and results are described in Restreppo et al. (2021). The data files are in netCDF (.nc) grid format. The file called Log10_prediction_5m.nc contains the predicted sediment accumulation rate in log10(g/cm2/yr). The file called Linear_uncertainty_5m.nc is the uncertainty, in g/cm2/yr.