Rapid and large-scale reservoir modelling for offshore freshened groundwater applications: The North Sea story

With extreme climatic events and increasing populations, water stress in regions of Europe is becoming critical. Offshore Freshened Groundwater (OFG) is increasingly being identified within continental margin sedimentary sequences worldwide, and has potential to be used as an industrial, agricultural, or domestic/potable resource, especially as a mitigation to drought during extreme climatic events. As part of an international effort under the Horizon Europe Water4All project RESCUE (RESources in Coastal groundwater Under hydroclimatic Extremes), we have used extensive subsurface petrophysical and geophysical datasets, alongside machine learning approaches, to generate detailed static reservoir models for a region of the Southern North Sea. Neutron, density and sonic porosities from well log data are used to train the spatially aware EMBER machine learning algorithm against acoustic impedance data, which is derived from 3D seismic reflection and well data. We demonstrate a strong predictive capacity of the trained algorithm to predict porosity from acoustic impedance for the interpreted formations by blind well testing. Permeability is also derived from well logs using the Timur and Holmes-Buckle relationships, before also training EMBER for permeability prediction. Our results provide a detailed, strongly data based, and fully spatially constrained determination of the porosity and permeability distribution for an area of the Southern North Sea, which can be used for dynamic modelling of OFG emplacement during sea level lowstands associated with the last glacial maximum.