To support a socially-licensed greener future, one of the biggest challenges of the next decade is to improve our ability to predict subsurface geology. For example, the mine of the future must have a reduced footprint and economic, socially-accepted mineral resource discoveries will depend on how well we are able to characterise the subsurface geology. This requires the ability to probabilistically forecast sub-surface geology, allowing for rapid model updates.

We present the current state of the Loop project, an open-source interoperable, integrative, probabilistic 3D geological modelling platform. Map2loop is a library that automatically extracts geological information from maps and generates parameters for the modelling library. In LoopStructural, we have defined a parameterisation of 3D geological models in a forward modelling sense. LoopStructural is based on the concept of the structural frame: a coordinate system defined for each object (faults, intrusions) or geological events (folding). These coordinate systems consist of 3 perpendicular scalar fields that are interpolated and fitted to data in 3D and then combined according to the geological history. The structural frames are conformable conformable to layering throughout the models. We present the concept for LoopResources, our proposed property modelling library. Using this deformed cartesian coordinate system, we propose to adapt geostatistical and interpolation methods to curvilinear coordinate systems using classical XYZ-UVW transformations. This will ensure that lithological anisotropies are enforced during resource estimation and property modelling to provide better digital twins of the subsurface and characterise geological uncertainty throughout the entire workflow.

In order to transition to more sustainable technologies, we as society need to improve our ability to find and manage natural resources. One of the biggest challenges for managing natural resources is our ability to characterise the subsurface distribution of geological objects including mineralisation, structures and stratigraphy. Standard approaches for quantifying the geometries of these objects interpolate these geometries using mathematical interpolation techniques which generally cannot incorporate geological rules and knowledge. Here we use a time aware modelling approach where the most recent geological feature is modelled first. The geometry of the first feature is then used to build a structural frame, a curvilinear coordinate system aligned the geometry of the feature for example capturing the fault surface and slip direction or fold axis and axial surface. The structural frame can then be used as a reference frame and combined with a conceptual model conditioned to geological observations to model the geometry of the older geological features. Using appropriate overprinting relationships and geological rules it is possible to combine multiple structural frames to characterise complicated geological objects for example refolded folds, overprinting fault networks and duplex faults. We demonstrate the application of structural frame to modelling folds, faults and intrusions with different case studies demonstrating how incorporating the structural frames allow for geologists to use models to test geological hypotheses to further understand subsurface geometries.

Coseismic responses in groundwater level have often been observed following earthquakes worldwide. These responses have often been attributed to coseismic static and dynamic changes in volumetric strain and pore pressure, caused by slip on the ruptured fault. On 12. September 2016, the M_L 5.8 Gyeongju earthquake ruptured a branch of the Yangsan fault network in southeastern Korea, triggering hydrological responses near the mainshock epicenter. To better understand the connection between volumetric strain, pore pressure and groundwater level (GWT) levels, we developed a hydro-mechanical coupled dynamic distinct element model (dyn-DEM) to simulate the Gyeongju earthquake rupture process and subsequent fluid pressure response, using 2D Particle Flow Code v7. The rock mass was modeled using an assembly of circular particles, bonded to each other by contacts with the potential to break, collectively simulating the hydro-mechanical effects of a seismic event upon application of an in-situ stress field. The hydraulic fracture process was represented by a pipe network model, where fluid flow was simulated through a network of flow channels which connected pore spaces storing fluid volume and pressure. During the simulation, the finite volume method was used to solve for the pore pressure evolution due to poroelastic effect. Overall, we observed a positive correlation between coseismic GWT level changes near the Gyeongju earthquake epicenter and modeled stress and pore pressure changes. This result supports the use of hydro-mechanical coupled dyn-DEM in reliably quantifying changes in the stress and pore pressure fields throughout the dynamic rupture process of a simulated seismic event.