Prediction of the abundance of ferromanganese nodules using Gaussian Process Regression based on multisource geological data in the western Pacific deep-sea basin

Ferromanganese nodules, rich in cobalt (Co), nickel (Ni), copper (Cu), manganese (Mn), and rare earth elements (REEs), are important marine mineral resources with the utmost capacity for commercial employment in the future. Recently, the discovery of high abundant Co-rich nodules in the Western Pacific has attracted significant attention. The prediction of nodule abundance is a vital geological problem to be solved in marine mineral resource exploration. Based on the multisource geological data of the study area in the western Pacific Ocean acquired through acoustic, optic and geological sampling, a stochastic probabilistic prediction for nodule abundance was developed via Gaussian process regression (GPR). The results revealed that the predicted abundance of nodules ranged from 0 to 71.2 kg/m², with an average abundance of 26.3 kg/m². The high-abundance (~30.0 kg/m²) nodules are mainly distributed in the deep-sea basins around several seamounts, and they may be spatially coupled with the Co-rich crust distributed over seamounts in the targeted study area. Compared to traditional machine learning approaches, such as stepwise linear regression, regression trees and support vector machine, intelligent prediction of nodule abundance by GPR is achieved with improved accuracy. Moreover, with the predicted abundance, the prediction error is obtained simultaneously by GPR. The deep-sea basins between the Magallan and Marcus-Wake seamounts are considered potential areas for further exploration of Co-rich ferromanganese nodules in the western Pacific Ocean.