Evaluation of an innovative, open-source and quantitative approach for the kinematic analysis of rock slopes based on UAV based Digital Outcrop Model: A case study from a railway tunnel portal (Finale Ligure, Italy)

Abstract

This paper presents a novel, cost-effective method for a rapid and quantitative characterization of discontinuities in fractured rock cliffs using Unmanned Aerial Vehicle-derived Digital Outcrop Models (DOMs). The proposed workflow combines manual extraction of discontinuity data from the DOM with a kinematic analysis using an automated algorithm (ROKA), which upgrades and renews the traditional Markland's test. The case study, a designed railway tunnel portal in NW Italy (Finale Ligure) demonstrates the advantages and limitations of this approach compared to current methods. Manual mapping is compared with (semi-)automatic extraction of discontinuity data from the DOM and validated with field-based scan lines. The results show that in complex geological settings, manual mapping provides more statistically robust and geologically reliable measurements of discontinuity orientation, validated by the user's expertise. Traditional stereographic approaches in kinematic analysis are limited in characterizing spatially variable slopes and discontinuity networks, providing only generic estimations of unstable slope surfaces and discontinuity intersections. This limitation is overcome by new algorithms that incorporate 3D spatial relationships of both the discontinuity network and the rock slope, resulting in a more accurate and site-specific kinematic characterization of the rock cliff, emphasized by 3D visualizations of the critical planes and potential rock failure volumes. The new approach significantly impacts the time, effort, and cost of the engineering projects, allowing to quantitatively define the potential unstable areas with a fast analysis. The Authors stress the importance of integrating digital workflows with comprehensive field-based characterizations of the local litho-stratigraphic and structural setting to effectively utilize large datasets and partially automated procedures.