Intelligent interpolation strategies in geological modelling for safe storage sites: comparison and uncertainty approach

@inproceedings{yang_intelligent_RM2023,
 abstract = {Geological models are an important foundation for many decisions in a variety of geoscientific areas. In this work, we specifically consider geological models in the context of nuclear waste disposal. One of the specific challenges this field is to determine the best location for long-term safety of storage and disposal systems, which requires the development of novel modeling and simulation approaches to ensure efficiency and safety. In addition, the decisions should be transparent and reproducible. A straight-forward way to ensure these aspects is the use of open-source tools. We utilized here the open-source Python geomodelling package GemPy, which is based on a universal co-kriging method for 3D implicit geological modeling. 
As introduced in the related GeoBlocks contribution (CARL ET AL., 2023A), a suite of generalized standard models was generated to represent generic geometric building blocks for rock salt, claystones and crystalline rocks, based on information picked from regional cross-sections and textbook examples. The standard models are here used as a benchmark to evaluate the performance of different interpolation methods. By applying cross-validation, the accuracy and reliability of each method are assessed, allowing for a comprehensive comparison. Additionally, spatial uncertainty is addressed on the geometric building blocks by updating a prior model with a likelihood function to produce the posterior model; where the prior is obtained by perturbing the geometric building block and the likelihood function is given by observed data.
Results indicate that the proposed building blocks method is suitable to address an important step in the modeling workflow: the selection of suitable geological modeling methods and parameterizations for specific rock types and expected geometries. Future work will aim to optimize the workflow of geological
modeling, construct adaptable geological structures, incorporate uncertainty analysis, and provide powerful stochastic modeling capabilities to obtain an accessible and transparent option for a robust site selection process.},
 author = {Yang, Jian and Colombo, Carlos and Carl, Friedrich and de los Angeles de Lucio, Gabriela and Achtziger, Peter and Kukla, Peter A and Wellmann, Florian},
 booktitle = {2023 {RING} meeting},
 language = {en},
 pages = {14},
 publisher = {ASGA},
 title = {Intelligent interpolation strategies in geological modelling for safe storage sites: comparison and uncertainty approach},
 year = {2023}
}