부산대학교 도서관

Debris flow runout poses a hazard to life and infrastructure. The expansion of human population into mountainous areas and onto alluvial fans increases the need to predict and mitigate debris flow runout hazards. Debris flows on unconfined alluvial fans can exhibit spontaneous self‐channelization through levee formation that reduces lateral spreading and extends runout distances compared to unchannelized flows. Here we modify the D‐Claw shallow flow model in two ways that are hypothesized to generate levees. We evaluate these modifications with observations from a large‐scale flume experiment. We investigate model performance when including the effect of two different friction sub‐models, as well as the inclusion of segregation effects on granular permeability. Results show that, for a wide range of plausible model input parameters, simulations including the effects of segregation promoted modeled levee formation, whereas simulations without the effects of segregation did not create levees. Further, using a forward predictive framework, simulations with the effects of segregation were more likely to better model the magnitude of debris flow depth and runout distance, whereas simulation timing of the debris flow was affected by the choice of friction sub‐model. Our results indicate that including the effects of segregation on granular permeability can improve the likelihood of better predictions of debris flow depth and runout prior to an event occurring. Plain Language Summary: Hillsides that receive heavy rainfall sometimes experience a debris flow, a flood‐like mix of water, soil, and rocks. As more people move into mountainous regions, the hazard that debris flows present increases. In this study, we use a computer simulation to model debris flow behavior. We look at different ways to model parts of the physics, specifically, friction and the effects of the debris material size, on the area that debris flow reaches. We find that capturing the effect of different sized debris is important to recreate the self‐forming debris flow channel that can extend the length a debris flow can travel. Including these effects improves the predictive capability of the computer simulations. Improving our understanding and ability to model debris flow runout is important to predicting locations where a potential hazard exists from debris flows. Key Points: Modeling different material properties through the effects of segregation in D‐Claw allows for spontaneous self‐channelizing levee formationIncluding the effects of segregation on granular permeability improves debris flow simulation performance for recreating a large scale experimental debris flow [ABSTRACT FROM AUTHOR]