GeoBlocks: Building Blocks for Quantifying Uncertainties in Geological Models

Summary The GeoBlocks project develops an integrated, open-source workflow for quantifying and managing uncertainties in geological models to support the German site selection process for a repository for high-level radioactive waste. Geological decision-making is challenged by sparse, heterogeneous, and legacy subsurface data, particularly during early stages of the process. GeoBlocks addresses this challenge by treating geological models as compositions of modular building blocks, enabling transparent propagation of uncertainty from input data through modeling to decision-relevant outputs. The workflow combines implicit geological modeling, stochastic uncertainty quantification, and quantitative analysis of geometric uncertainty. A key innovation is a library of standard geometries representing typical configurations of claystone, salt, and crystalline host rocks, which supports conceptual model selection, method choice, and communication in data-poor settings. Benchmark studies demonstrate that interpolation performance strongly depends on geometry and data density, and that hybrid methods incorporating prior geological knowledge substantially improve model robustness under sparse data conditions. GeoBlocks further quantifies interpretational uncertainty in seismic and borehole data and supports iterative model updating across all stages of the German site selection process from identification of site regions through surface exploration to underground exploration, providing a transparent and reproducible basis for uncertainty-informed decision-making.