Effects of wood on debris flow runout in small mountain watersheds

Abstract

[1] Debris flows have typically been viewed as two-phase mixtures of sediment and water, but in forested mountain landscapes, wood can represent a sizable fraction of total flow volume. The effects of this third phase on flow behavior are poorly understood. To evaluate whether wood can have a significant effect on debris flow runout in small mountainous watersheds, we used a landscape-scale model combining empirical, stochastic, and physical submodels of storms, fires, forest growth, tree fall, wood decay, soil production and diffusion, landslide initiation, debris flow runout, and fluvial sediment transport. We examined changes in the cumulative distribution function of debris flow runout lengths in a small (2 km 2 ) watershed in the Oregon Coast Range due to presence or absence of two hypothesized effects of wood: (1) velocity reduction due to entrainment of wood in the runout path and (2) velocity reduction due to changes in flow direction angle. The model was calibrated such that the distribution for simulations including both effects was similar to that measured in the study basin, and amounts of wood in the simulation and the field, both fallen in small valleys and incorporated by debris flows, were comparable. Removal of either effect, or both, significantly shifted runout length distributions to longer lengths. Simulations and field observations indicate that with wood, fluvial transport is a significant source of sediment output, few debris flows reach the outlet, and debris flow deposits are widely distributed throughout the network. Simulations indicate that without wood, basin sediment yield greatly increases, that yield is dominated by longer-runout debris flows, and that debris flow deposits are concentrated in the low-gradient reach near the outlet.

Debris dams and the relief of headwater streams

“…In general, the component of gradient due to debris dams, S d , changes in a complicated way downstream, likely because the probability of debris-flow deposition, the primary dam-forming agent, changes in a complicated way downstream and depends on, e.g., the sizes of and spacings between tributaries (e.g., Benda and Cundy, 1990;Lancaster et al, 2001Lancaster et al, , 2003Benda et al, 2003). In the leaf-shaped Hoffman Creek basin (figure 1), the trend of fewer debris-flowdelivering tributaries downstream (figure 5) likely contributes not only to the strong relationship between h d /λ and contributing area (figure 6) but also to the nearly power-law decline in stream gradients with increasing contributing area (figure 3).…”

“…That maximum is near Cedar Creek's outlet, near the middle of the Hoffman Creek profile, and further upstream for Bear Creek. Visual inspection of the basin maps (figure 1) and figures 4 and 5 suggests that locations of relief-fraction maxima are related to "proximal" upstream network structure with a high bifurcation ratio (i.e., degree of branching), although that proximity is likely measured in terms of high (but not too high) gradients and low tributary junction angles rather than short horizontal distances (Benda and Cundy, 1990;Lancaster et al, 2001Lancaster et al, , 2003Daniel J. Miller and Kelly Burnett, personal communication, 2003). In the steepest, upstream-most parts of the profiles, accumulation of relief in debris dams is relatively small because debris flows often scour and rarely deposit there.…”

“…Measuring that relief in the field is infeasible because attributing alluviation in particular locations with particular dams is difficult, especially where alluviated reaches overlap or are superimposed-debris flows frequently deposit on older deposits (Lancaster et al, 2003). We therefore seek an analytical expression for alluvial gradient, S a , that, when coupled with observed distributions of debris dams, will allow estimation of impoundments' effect on profile relief.…”

“…(2) Dam formation is stochastic such that, at any point along a given reach, the ratio of dam height to spacing is a random variable with a well defined mean. The inherent variability of debris-dam formation is recognized (Benda and Dunne, 1997;Lancaster et al, 2001), as is the controlling influence of local network structure (Benda and Cundy, 1990;Benda and Dunne, 1997;Lancaster et al, 2001Lancaster et al, , 2003Benda et al, 2003). Dams…”

“…Montgomery et al (2003a) surmised that the relative immobility of large wood, its impounding of sediment, and the concomitant shielding of the underlying bedrock decrease the effective bedrock erodibility and, hence, steepen the longitudinal channel profile. These effects are likely greater further upstream because (a) debris flows richer in wood deposit further upstream (May, 2002;Lancaster et al, 2003) and (b) large wood has lower mobility in smaller channels (Marcus et al, 2002). Boulders, which are typically deposited by debris flows but essentially immovable by fluvial processes, have similar effects (figure 2).…”