Geological modeling is central to understanding the subsurface—whether for resource exploration, groundwater management, geothermal energy, or assessing geological risks. These models aim to integrate diverse and often sparse data into coherent representations of subsurface structures and properties.

Traditionally, geostatistics has provided a powerful framework to create geological models and to handle spatial uncertainty, allowing interpolation and simulation of rock properties based on limited observations.
However, as geological settings grow in complexity and data volumes increase, classical approaches reach their limits. Recent advances in computational geometry offer new possibilities: they enable compact, low-dimensional representations of complex geological structures that remain interpretable and physically plausible. These representations are particularly well-suited for integration with geophysical inversion methods and allow for efficient probabilistic workflows with meaningful uncertainty quantification.

In this talk, we explore the evolution from traditional geostatistical techniques to modern approaches in geological modeling, including the use of machine learning and AI. We highlight how large-scale pre-trained models—such as foundation models and LLMs—can provide a path to include contextual understanding. Through examples from joint geological-geophysical modeling and inversion, we illustrate how these novel tools offer not only technical improvements but also a conceptual shift in the use of geological models as scientific tools to understand the subsurface.

3D geological models are an essential source of information for research as well as for the safe and efficient use of the underground. They provide not only a visualization of the subsurface structures, but also serve as geometry input for geophysical and numerical models, e.g., gravimetric, mechanical or thermal models. The set-up of a geological model for a numerical simulation is often a time-consuming task. During the last two decades several 3D geological models have been created for specific regions in Germany. However, up to now only one attempt has been made to combine several of them to a Germany-wide model. We present a new Germany-wide 3D geological model combining information of 27 individual models. The model has a resolution of 1 x 1 km² and is vertically and horizontally subdivided into 146 units. Where possible, the model has been extended to neighbouring states, e.g., Netherlands, France, Austria or Switzerland. In order to combine all models with their different sizes, resolutions and stratigraphic subdivisions, a point-set approach was chosen which has a number of advantages with regard to the flexibility and usability.

The shallow subsurface of the North German Basin consists of unconsolidated Quaternary sediments. During the Elsterian glaciation subglacial tunnel valleys were deeply incised below the ice sheets. These tunnel valleys commonly range in depths between 100-400 m, but may reach depths of more than 500 meters. Given that the BGE is focusing on depths of 300 to 1500 meters for high-level radioactive waste repository sites, the potential for future subglacial erosion must be considered to ensure the long-term safety of a selected site.

We develop hydrostratigraphic 3D reservoir grid models of the Northwest German Basin as input for numerical hydraulic subglacial erosion modelling to assess tunnel valley formation during future glaciations. These hydrostratigraphic 3D reservoir grid models are constructed with different resolutions. All models cover Permian to Cenozoic sediments and have a depth of 2000 m. They are constructed with a layered-structural-model, voxel-grid-models approach. To construct the layered structural model, we used existing stratigraphic 3D models (GTA3D, TUNB3D-NI) and additional borehole data. The reservoir grids are constructed as hydrostratigraphic grids, integrating constant permeability values based on hydrogeological properties of the stratigraphic units. This approach allowed fast and successful construction of large grid models (up to 40,000 km² in size) despite the heterogeneous database.

By implementing grid models with different resolutions in the numerical erosion model, we will test the impact of grid size on the outcome of the erosion modelling. The results will help in the process of site selection and long-term safety assessment for potential repository sites.

During the TUNB Velo 2.0 project, the Geological Surveys of the North German Federal States and the Federal Institute for Geosciences and Natural Resources (BGR) developed a seismic velocity model of the North German Basin. It is based on the TUNB structural model from the previous TUNB project, which covers 13 stratigraphic units from Zechstein to Tertiary, numerous salt structures and faults.

For Lower Saxony, the modelling was done in Aspen SKUA®. A volume model was derived from the structural model and parameterized with seismic velocities. The regional velocity approach is based on Jaritz et al. (1991), where the interval velocities of the stratigraphic units can be derived from their theoretical surface velocities and gradients (V0-K method).

The resulting regional velocity model of the TUNB Velo 2.0 project covers almost all of Lower Saxony, with the exception of the southernmost areas where the availability of seismic data is limited. At the LBEG, 75% of the 3D and 83% of the 2D seismic data is available only in time domain. With the velocity model now available, time domain seismic data can be converted to depth domain for its use in structural modelling. In the talk, we will present modelling results, show uncertainty considerations and the usage of this regional velocity model to convert seismic data.

References:

Jaritz W., Best G., Hildebrand G., Jürgens U. (1991) Regionale Analyse der seismischen Geschwindigkeiten in Nordwestdeutschland. Geol Jahrb 45:23–57