A 3D geological model for the Ruiz-Tolima Volcanic Massif (Colombia): Assessment of geological uncertainty using a stochastic approach based on Bézier curve design *

• A method based on implicit 3D modelling techniques combined with a stochastic approach toassess geological uncertainty is presented. The method uses Bézier curves on 2D crosssections to represent geological features.
• The modelling results reflect the state of information of the study area as well as its uncertainty.Alternative geological hypotheses are considered and modelled. Results are presented asprobabilistic models, with diffuse boundaries based on fuzzy logic.
• The resulting geological model is intended to synthesise information from diverse sources intoa spatially and conceptually coherent frame. These should be taken as a first generation of 3Dgeological models for Nevado del Ruiz Volcano.

The Ruiz-Tolima Volcanic Massif (RTVM) is an active volcanic complex in the Northern Andes, and understanding its geological structure is critical for hazard mitigation and guiding future geothermal exploration. However, the sparsity of data available to constrain the interpretation of this volcanic system hinders the application of standard 3D modelling techniques. Furthermore, some features related to the volcanic system are not entirely understood, such as the connectivity between the plutons present in its basement (i.e. Manizales Stock, El Bosque Batholith). We have developed a methodology where two independent working hypotheses were formulated and modelled independently (i.e. a case where both plutons constitute distinct bodies, and an alternative case where they form one single batholith). A Monte Carlo approach was used to characterise the geological uncertainty in each case. Bézier curve design was used to represent geological contacts on input cross sections. Systematic variations in the control points of these curves allows us to generate multiple realisations of geological interfaces, resulting in stochastic models that were grouped into suites used to apply quantitative estimators of uncertainty. This process results in a geological representation based on fuzzy logic and in maps of model uncertainty distribution. The results are consistent with expected regions of high uncertainty near under-constrained geological contacts, while the non-unique nature of the conceptual model indicates that the dominant source of uncertainty in the area is the nature of the batholith structure.