Home » Projects » Debris Flows

Debris Flows

Three men stand on a large log among large woody debris and sediment deposited by a debris flow.

Photo by Gordon Grant

In many mountainous regions, debris flows can be an
important factor determining stream-habitat conditions.
Human activities can influence the frequency, magnitude,
and content of debris flows. Thus, understanding about
debris-flow processes and effects are important to stream
conservation and sustainable forest management.

Research Description:

This integrated research project strives to develop tools for modeling and mapping debris- flow effects on salmon and trout habitats. Such tools are needed to sustainably manage and conserve these commercially, recreationally, and culturally important fish. The project has three primary objectives: 1) building a foundation for research and management by developing data on landslides and stream networks over large areas; 2) modeling the probability of landslides and debris flows to affect fish habitats; and 3) evaluating the potential consequences of forest management alternatives on debris-flow behavior and effects.

Key Findings:

Example of modeled streams.

Figure 1. Clarke et al. 2008

Characterizing Landslides and Streams

This research quantifies how differences in accuracy, spatial resolution, and spatial extent of digital terrain, landslide inventories, and stream maps may affect broad-scale assessments relevant to salmonid habitats (Clarke and Burnett 2003; Vance-Borland et al. 2008; Miller and Burnett 2007). For example, after correcting for both topographic variability and air photo bias, sub-sampling an air photo–based landslide inventory demonstrated that older forests, when sampled over small areas as in field studies, commonly exhibited higher landslide densities than other forest cover classes but when sampled over larger areas always exhibited the lowest densities (Miller and Burnett 2007). This highlights the need to develop regionally applicable forest-cover specific estimates of landsliding from data collected over large areas (>500km2).

Figure 1. Tools were developed that capitalize on strengths in available field and regional digital data to model stream networks and stream attributes at a high resolution over large areas. Example of streams that were modeled from the 10-m DEMs for 96,000 km of streams in Oregon Coastal Province (Clarke et al. 2008). Outset shows the portion of the delineated stream network identified at a 70% probability of perennial flow overlaid with hydrography from 1:24,000-scale USGS topographic maps.

Figure 2. Burnett and Miller 2007 model outputs.

Fig 2. Burnett and Miller 2007

Modeling Debris Flows

An approach was developed to model probabilities of debris-flow initiation and runout from regionally available digital elevation and land-cover data and to calibrate these probabilities with field data (Miller and Burnett 2007; Miller and Burnett 2008).

Figure 2. Outputs of the coupled debris-flow initiation and delivery model (Miller and Burnett 2007; Miller and Burnett 2008) illustrated for a portion of the Knowles Creek basin in the central Oregon Coastal Province, USA. Modeled: (A) landslide density; (B) probability of debris-flow delivery to a fish-bearing channel; (C) delivery-weighted landslide density expressed for each pixel as the product of the landslide density and the probability of debris-flow delivery to a fish-bearing channel; and (D) probability that a pixel is traversed by a debris flow that traveled to a fish-bearing channel. This probability is presented over a hill-shade view of the underlying DEM.

Figure 3. Burnett and Miller 2007, modeled debris flow initiation and traversal zones.

Fig 3. Burnett and Miller 2007

Evaluating Potential Consequences of Forest Management Alternatives

Probabilities were ranked for hillslopes and headwater streams to initiate or transport debris flows that run out to fish-bearing streams, and the ranks were applied to delineate alternative streamside management zones for headwater streams (Burnett and Miller 2007). Sites with the highest probabilities of initiating or transporting debris flows were contained in a relatively small percentage of the study area.

Figure 3. Initiation and traversal zones illustrated for the Knowles Creek basin. The modeled fish-bearing channel network is shown in white.
(A) Initiation zones (top) are in dark gray and include (from left to right) 25%, 50%, and 75% of initiation sites for landslides that delivered to fish-bearing channels.
(B) Traversal zones (bottom) are in black and include (left to right) 25%, 50%, and 75% of pixels traversed by debris flows that delivered to fish-bearing channels.
Gray polygons include 35-m extensions for traversal zones along all nonfish-bearing streams.