I. Introduction
High-resolution lithological models are key to driving decision-making in subsurface earth resources applications such as groundwater [1]–[3], geothermal [4], [5], or oil and gas [6], [7]. Seismic data have been widely used to model the subsurface reservoir lithology in these applications [8]–[15]. As lithological variation represents a major component of geological heterogeneities, it determines the distribution of both petrophysical (e.g., porosity, permeability) and fluid properties (e.g., pressure, saturation). Quantifying the lithological uncertainty therefore plays an essential role in subsurface resource appraisal and development.