Rockfall impacts on the road network of the Metropolitan Area of Rome: a risk prioritisation framework for decision-makers

Road networks in highly urbanised and geomorphologically complex areas are particularly vulnerable to landslide processes, which may severely affect mobility, emergency response and territorial resilience. In this study, a province-scale framework is proposed to assess rockfall susceptibility, runout and related level of risk affecting the road network of the Metropolitan Area of Rome (Italy). The methodology integrates machine-learning-based susceptibility modelling, semi-empirical runout simulation and network-oriented impact analysis to derive spatially explicit hazard and indexed-risk scenarios tailored to decision-making needs. Rockfall initiation susceptibility was modelled using a Random Forest classifier trained on an integrated landslide inventory and a set of lithological, morphometric, climatic, seismic and infrastructural predictors. Runout propagation allowed the estimation of impact probability, velocity and kinetic energy on elementary road segments. Hazard indices were combined with road vulnerability, intrinsic infrastructure value classes and network relevance metrics to produce a hierarchical risk classification of the road network at the provincial scale. Results highlight that high-risk road segments are spatially concentrated rather than uniformly distributed, and that a substantial share of cumulative network risk affects intermediate road classes due to their extensive distribution and functional connectivity role and generally lower protection levels. The resulting spatial scenarios are conceived as pre-defined operational configurations that can be activated under rainfall-threshold exceedance, supporting impact-based emergency planning, maintenance prioritisation, and strategic risk reduction under constrained financial and institutional conditions. • A province-scale framework is developed to assess rockfall susceptibility, runout and risk affecting road networks. • ML susceptibility modelling is integrated with semi-empirical runout simulation and network analysis. • Risk scenarios highlight the key role of intermediate road classes in cumulative infrastructure risk. • The approach supports impact-based emergency planning and evaluation of Limit Conditions for Emergency (CLE). • Spatial scenarios can be activated with weather forecast enabling nowcasting-oriented applications.