Teaching 3D Geological Mapping and Modeling Sensitivity Using Visual KARSYS

Spatial understanding of complex geological structures is a fundamental component of geoscience expertise. However, when teaching geological mapping, students often face challenges in grasping 3D concepts through ‘traditional’ 2D or projected teaching materials, such as geological maps and cross-sections. In this perspective, we developed a specialized course for Master's students in the Geosciences program at Lyon (https://lyongeologie.fr/m1-geosciences/), using the Visual KARSYS app (https://www.visualkarsys.com/).

Visual KARSYS is a free web-based platform that enables users to construct geological and hydrogeological 3D models by integrating various data types, including those inferred from drillhole logs, field survey observations, structural data (e.g., bedding geometry, unit contacts, fault planes), and geophysical imaging products. Geological models are computed using the GmLib library (implicit approach based on potential-field method), developed by the French Geological Survey (BRGM).

As part of the course, over fifty students were tasked with building a 3D geological model of the Mont-d’Or Lyonnais massif (MOL, northwest of Lyon, France). The MOL massif comprises a monoclinal sedimentary series ranging from Triassic sandstones to Upper Jurassic carbonates, underlain by a Variscan orthogneiss basement and overlain by Quaternary units to the east. Most students had previously visited the MOL during a field trip and were familiar with its geology, providing a solid foundation for their modeling work. The course is structured around three primary learning objectives:

(i) Building a 3D Geological Model: Students learn to construct a geological model using Visual KARSYS, developing an understanding of the fundamental principles of the potential-field method employed by the GmLib library. This involves working with different types of data (e.g., hard data such as drillholes and field measurements, and soft data such as inferred contacts and fault geometries) and defining the lithostratigraphic framework of the model.

(ii) Comparing Model-Driven and Non-Model-Driven Approaches: Students are asked to compare two geological cross-sections with identical endpoints—one derived from the 3D geological model (supported by hard data only) and another hand-drawn by each student using a geological map. Students quantitatively assess the similarities and differences between the two products, evaluating which approach yields a more realistic representation. Interestingly, most students tend to favor their manually drawn cross-sections, often influenced by preconceived notions of the local geology.

(iii) Exploring Model Sensitivity: Students build an initial 3D model using only hard geological data (e.g., boreholes, field observations) and subsequently enhance the model by progressively incorporating ‘softer’ data (e.g., lithological contacts or fault structures extracted from geological maps, geophysical interpretations). With each iteration, students analyze the impact of additional data on the model, gaining insights into its sensitivity and the implications of integrating interpretative datasets.

This teaching approach provides students with practical experience in 3D geological modeling while fostering critical thinking about data integration, model-driven approaches, and geological model sensitivity.