Controls on rockfall hazard and risk metrics at critical spots of access roads to Yosemite Valley (California, USA)

Yosemite National Park (California, USA) is characterized by high-relief granitic cliffs shaped by complex geological processes and forming iconic geomorphological features, including exfoliating granite and a steep glacially carved landscape. This setting results in frequent, often intense rockfall activity that poses a significant threat to humans, property and utilities along the road network accessing Yosemite Valley. Quantitative assessment of rockfall risk along these roads (El Portal, Big Oak Flat and Wawona) relies on synthesized metrics that integrate both hazard and exposure of elements at risk.

Rockfall hazard depends on release mechanisms and magnitude–frequency relationships. Slope topography and materials (e.g. fine vs coarse talus, shallow soil covering) also play a critical role influencing energy dissipation, trajectory dispersion or convergence. These factors ultimately determine the frequency, energy, and fly height of rockfalls reaching road segments. Exposure is mainly governed by traffic characteristics such as vehicle density, speed, and occupancy. Comparable risk values across different sites may be a result of different combinations of hazard and exposure factors, underscoring the need for site-specific mitigation strategies.

We assessed rockfall hazard and risk along Yosemite Valley access roads using high-resolution (1 m) 3D rockfall runout simulations performed with the Hy–STONE simulator combined with a historical rockfall inventory (1857–2023) and traffic data provided by the National Park Service. Hazard was quantified using a modified Rockfall Hazard Vector (RHV) method incorporating block kinetic energy, fly height, and a normalized annual frequency derived from both onset frequencies from inventory analyses and propagation frequencies from runout modeling. Although originally conceived as a susceptibility index, the modified RHV provides an effective proxy for quantitative hazard. Rockfall risk was computed by integrating hazard with exposure and vulnerability parameters, including vehicle speed, size, and traffic volumes. The road network was discretized into 10 m segments for each travel lane (inbound and outbound from Yosemite Valley), and risk was evaluated for different rockfall volume scenarios (0.01–100 m³) while accounting for model uncertainties. For each segment, the annual probability of loss of life (E(LOL)) was estimated under different traffic conditions.

The results identify several critical road sections where the distribution and magnitude of elevated risk arise from distinct combinations of hazard and exposure contributions. For example, in the Parkline sector, high risk conditions are dominated by high hazard concentrated within a narrow corridor and related to exfoliation sheet failures from a steep cliff directly above the road with risk further amplified by congested traffic patterns. At Windy Point, comparable risk levels are associated with lower hazard levels in an area with multiple small, structurally controlled sources, but with higher exposure to widespread rockfall trajectories and adverse traffic conditions. Conversely, at the junction between Big Oak Flat and El Portal Roads, high risk is dominated by exposure linked to traffic flow convergence despite moderate hazard levels.
These findings highlight the importance of disentangling the individual factors contributing to quantified risk metrics to design targeted and effective mitigation strategies for rockfall risk along park access roads and more widely to mountain roads.