Debris flows are concentrated slurries of water and sediment that shape the landscape and pose a major hazard to human life and infrastructure. Seismic ground motion‐based observations promise to provide new, remote constraints on debris flow physics, but the lack of data and a theoretical basis for interpreting them hinders progress. Here we present a new mechanistic physical model for the seismic ground motion of debris flows and apply this to the devastating debris flows in Montecito, California on 9 January 2018. The amplitude and frequency characteristics of the seismic data can distinguish debris flows from other seismic sources and enable the estimation of debris‐flow speed, width, boulder sizes, and location. Results suggest that present instrumentation could have provided 5 min of early warning over limited areas, whereas a seismic array designed for debris flows would have provided 10 min of warning for most of the city.

Steep, rocky landscapes often produce large sediment yields and debris flows following wildfire. Debris flows can initiate from landsliding or rilling in soil-mantled portions of the landscape, but there have been few direct observations of debris flow initiation in steep, rocky portions of the landscape that lack a thick, continuous soil mantle. We monitored a steep, first-order catchment that burned in the San Gabriel Mountains, California, USA. Following fire, but prior to rainfall, much of the hillslope soil mantle was removed by dry ravel, exposing bedrock and depositing ∼0.5 m of sandy sediment in the channel network. During a one-year recurrence rainstorm, debris flows initiated in the channel network, evacuating the accumulated dry ravel and underlying cobble bed, and scouring the channel to bedrock. The channel abuts a plowed terrace, which allowed a complete sediment budget, confirming that ∼95% of sediment deposited in a debris flow fan matched that evacuated from the channel, with a minor rainfall-driven hillslope contribution. Subsequent larger storms produced debris flows in higher-order channels but not in the first-order channel because of a sediment supply limitation. These observations are consistent with a model for post-fire ravel routing in steep, rocky landscapes where sediment was sourced by incineration of vegetation dams—following ∼30 years of hillslope soil production since the last fire—and transported downslope by dry processes, leading to a hillslope sediment-supply limitation and infilling of low-order channels with relatively fine sediment. Our observations of debris flow initiation are consistent with failure of the channel bed alluvium due to grain size reduction from dry ravel deposits that allowed high Shields numbers and mass failure even for moderate intensity rainstorms.

<p>Gravity moves dry grains or blocks downhill in rockslides and rockfall. These mass movements can cause large boulders to saltate and impact with huge energies. Boulder impacts into bedrock surfaces should cause significant bedrock erosion, likely shaping the topography even in the absence of water.  Examples of potential rockfall-driven bedrock landforms include bedrock gullies on steep hillslopes, so-called plinth surfaces on caprock-topped mesa escarpments, and steep impact-crater slopes on planetary surfaces. Although grain impact processes have been incorporated into mechanistic models for fluvial and debris-flow incision, similar models for dry rockfall erosion have yet to be developed.</p><p>To explore the potential for dry rockfall to erode bedrock and shape the topography, we set up a discrete, cellular D16 dry grain saltation trajectory model accounting for particle saltation dynamics and evolving topography. We calibrated the model variables (i.e., particle hop angles, distances and velocities) for different grain sizes and hillslope angles using laboratory experiments of dry gravel transport over a tilted foam bed that served as an erodible bedrock analogue. We then explored the calibrated model for a broad range of hillslope angles, grain sizes and bedrock erodibilities.</p><p>Both model and experiments predict significant erosion due to rockfall-driven impacts. As the topography develops, alcoves (shell-shaped hollows) form near the upslope end of the model domain. These alcoves eventually overdeepen and fill with talus, preventing further erosion. Farther downslope, topographic feedbacks drive rockfall into incipient channels, which cause those channels to incise resulting in gullies. Overall, our work suggests that dry rockfall can be a significant bedrock incision process, and can lead to gully formation, even for hillslope angles that are significantly less than the angle of repose.</p>

Highlights: Flow resistance increased with shallow flow, bedforms, and high sediment transport intensity  Downstream and upstream migrating alternate bars developed on bed slopes up to 20%  Transition to upper plane bed coincided with development of concentrated sheetflow Abstract: Quantifying flow resistance and sediment transport rates in steep streams is important for flood and debris flow prediction, habitat restoration, and predicting how mountainous landscapes evolve. However, most studies have focused on low gradient rivers and the application of this work is uncertain for steep mountain streams where surface flows are shallow and rough, subsurface flows are not negligible, and there is form-drag from bed-and channelforms that differs from those in low gradient rivers. To evaluate flow resistance relations and sediment transport rates for steep channel beds, experiments were conducted using a range of water discharges and sediment transport rates in a 12 m long recirculating flume with bed slopes of 10%, 20%, and 30%, and a bed of nearly uniform natural gravel. Flow resistance for planar beds and beds that developed bedforms match empirical models that account for bedloaddependent roughness. Some bedforms were atypical for natural rivers at these bed slopes, such as ACCEPTED MANUSCRIPTA C C E P T E D M A N U S C R I P T 2 stepped alternate bars and upstream migrating alternate bars. Total flow resistance increased with decreasing particle submergence and energetic sediment transport and drag on bedforms. Using linear stress partitioning to calculate bed stresses due to grain resistance alone, sediment flux relations developed for lower gradient rivers perform well overall, but they overestimate fluxes at 20% and 30% gradients. Based on previous theory, mass failure of the bed, which did not occur, was predicted for the highest Shields stresses investigated at 20% and 30% bed slopes; instead a concentrated layer, four to ten particle diameters deep, of highly concentrated granular sheetflow was observed.

Rockfall can erode rocky hillslopes even below the angle of repose • Grainsize has a dominant effect on impact abrasion; slope is of minor importance • Topographic steering of grains results in self-formed bedrock channels