Although marine controlled source electromagnetic (CSEM) methods are effective for investigating offshore freshened groundwater (OFG) systems, interpreting the spatial extent and salinity of OFG remains challenging. Integrating CSEM resistivity models with information on sub-surface properties, such as host-rock porosity, allows for estimates of pore-water salinity. However, deterministic inversion approaches pose challenges in quantitatively analyzing these estimates as they provide only one best-fit model with no associated estimate of model parameter uncertainty. To address this limitation, we employ a trans-dimensional Markov-Chain Monte-Carlo inversion on marine CSEM data, under the assumption of horizontal stratification, collected from the Canterbury Bight, New Zealand. We integrate the resulting posterior distributions of electrical resistivity with borehole and seismic reflection data to quantify pore-water salinity with uncertainty estimates. The results reveal a low-salinity groundwater body in the center of the survey area at varying depths, hosted by consecutive silty- and fine-sand layers approximately 20 to 60 km from the coast. These observations support the previous study’s results obtained through deterministic 2-D inversion and suggest freshening of the OFG body closer to the shore within a permeable, coarse-sand layer 40 to 150 m beneath the seafloor. This implies a potential active connection between the OFG body and the terrestrial groundwater system. We demonstrate how the Bayesian approach constrains the uncertainties in resistivity models and subsequently in pore-water salinity estimates. Our findings highlight the potential of Bayesian inversions in enhancing our understanding of OFG systems, providing crucial boundary conditions for hydrogeological modeling and sustainable water resource development.

Accessible seafloor minerals located near mid-ocean ridges are noticed to mitigate projected metal demands of the net-zero energy transition, promoting growing research interest in quantifying global distributions of seafloor massive sulfides (SMS). Mineral potentials are commonly estimated using geophysical and geological data that lastly rely on additional confirmation studies using sparsely available, locally limited, seafloor imagery, grab samples, and coring data. This raises the challenge of linking in-situ confirmation data to geophysical data acquired at disparate spatial scales to obtain quantitative mineral predictions. Although multivariate datasets for marine mineral research are incessantly acquired, robust, integrative data analysis requires cumbersome workflows and experienced interpreters. Here, we introduce an automated two-step machine learning approach that integrates automated mound detection with geophysical data to merge mineral predictors into distinct classes and reassess marine mineral potentials for distinct regions. The automated workflow employs a U-Net convolutional neural network to identify mound-like structures in bathymetry data and distinguishes different mound classes through classification of mound architectures and magnetic signatures. Finally, controlled source electromagnetic data is utilized to reassess predictions of potential SMS volumes. Our study focuses on the Trans-Atlantic Geotraverse (TAG) area, which is amid the most explored SMS area worldwide and includes 15 known SMS sites. The automated workflow classifies 14 of the 15 known mounds as exploration targets of either high- or medium-priority. This reduces the exploration area to less than 7% of the original survey area from 49 km2 to 3.1 km2.

Integration of multiple geophysical methods in combined data analysis is a key practice to reduce model uncertainties and enhance geological interpretations. Electrical resistivity models resulting from inversion of marine magnetotelluric (MT) data, often lack depth resolution of lithological boundaries, and distinct information for shallow model parts. This is due to the nature of the physics i.e. diffusive method, model regularization during inversion, and survey setup i.e. large station spacing and missing high frequency data. Thus, integrating data or models to constrain layer thicknesses or structural boundaries is an effective approach to derive better constrained, more detailed resistivity models. We investigate the different impacts of three cross-gradient coupled constraints on 3D MT inversion of data from the Namibian passive continental margin. The three constraints are a) coupling with a fixed structural density model; b) coupling with satellite gravity data; c) coupling with a fixed gradient velocity model. Here we show that coupling with a fixed model (a and c) improves the resistivity model most. Shallow conductors imaging sediment cover are confined to a thinner layer in the resulting resistivity models compared to the MT-only model. Additionally these constraints help to suppress vertical smearing of a conductive anomaly attributed to a fracture zone, and clearly show that the seismically imaged Moho is not accompanied by a change in electrical resistivity. All of these observations aid interpretation of an Earth model indicating involvement of a plume impact in continental break-up during the early Cretaceous.