Title: Probabilistic Geomodeling: Recent Developments and Relationship to Reality
Florian Wellmann (1,2,3), Miguel de la Varga (1,3), Jan von Harten (1,3), Alexander Schaaf (4), Elisa Heim (1,5), Fabian Stamm (3), Zhouji Liang (1,2), Stefan Crummenerl (1), Alexander Jüstel (1,6) & Nilgün Güdük (1)
Computational Geoscience and Reservoir Engineering, RWTH Aachen University, Aachen, Germany (1); Aachen Institute of Computational Engineering Sciences, RWTH Aachen University, Aachen, Germany (2) M Terranigma Solutions GmbH, Aachen, Germany (3); Geology and Petroleum Geology, School of Geosciences, University of Aberdeen,UK (4); Applied Geophysics and Geothermal Energy, RWTH Aachen, Germany (5); Institute of Geology, RWTH Aachen University, Aachen, Germany (6)
Event: Abstract GeoUtrecht2020
Geological models, as 3-D representations of subsurface structures and property distributions, are used in many economic, scientific, and societal decision processes. These models are built on prior assumptions and imperfect information, and this aspect results in uncertainties about the predicted structures and property distributions, which will affect the subsequent decision process. We examine here uncertainties at different steps in the model construction process and discuss recent approaches in the consideration of these uncertainties.
Specifically, we will present an integrated probabilistic geomodelling approach, based on the open-source geomodelling package GemPy, that enables the consideration of uncertainties in geological interface observation points and orientation data, and the combination with random property fields. This workflow is based on an implicit geometry representation and a cokriging interpolation of point and orientation information. Subsequently, property kriging can be performed inside deformed domains. We also show how additional information, e.g. geophysical measurements, or information about geological topology, can directly be integrated in the workflow, and how the resulting model ensemble can be used to visualize and communicate uncertainties, and further combined with a framework to estimate optimal decisions under uncertainty.
Finally, a probabilistic geomodelling workflow can only capture specific aspects of uncertainty, and the final results have to be interpreted in light of this limitation. These limitations are aspects of active research, and we show promising results of successful applications.