Search for: All records

Abstract Debris flows pose persistent hazards and shape high‐relief landscapes in diverse physiographic settings, but predicting the spatiotemporal occurrence of debris flows in postglacial topography remains challenging. To evaluate the debris flow process in high‐relief postglacial terrain, we conducted a geomorphic investigation to characterize geologic, glacial, volcanic, and land use contributions to landslide initiation across Southeast Alaska. To evaluate controls on landslide (esp. debris flow) occurrence in Sitka, we used field observation, geomorphic mapping, landslide characteristics as documented in the Tongass National Forest inventory, and a novel application of the shallow landslide model SHALSTAB to postglacial terrain. A complex geomorphic history of glaciation and volcanic activity provides a template for spatially heterogeneous landslide occurrence. Landslide density across the region is highly variable, but debris flow density is high on south‐ or southeast‐facing hillslopes where volcanic tephra soils are present and/or where timber harvest has occurred since 1900. High landslide density along the western coast of Baranof and Kruzof islands coincides with deposition of glacial sediment and thick tephra and exposure to extreme rainfall from atmospheric rivers on south‐facing aspects but the relative contributions of these controls are unclear. Timber harvest has also been identified as an important control on landslide occurrence in the region. Focusing on a subset of geo‐referenced landslides near Sitka, we used the SHALSTAB shallow landslide initiation model, which has been frequently applied in non‐glacial terrain, to identify areas of high landslide potential in steep, convergent terrain. In a validation against mapped landslide polygons, the model significantly outperformed random guessing, with area under the curve (AUC) = 0.709 on a performance classification curve of true positives vs. false positives. This successful application of SHALSTAB demonstrates practical utility for hazards analysis in postglacial landscapes to mitigate risk to people and infrastructure. 

Abstract Empirical thresholds for landslide warning systems have benefitted from the incorporation of soil‐hydrologic monitoring data, but the mechanistic basis for their predictive capabilities is limited. Although physically based hydrologic models can accurately simulate changes in soil moisture and pore pressure that promote landslides, their utility is restricted by high computational costs and nonunique parameterization issues. We construct a deep learning model using soil moisture, pore pressure, and rainfall monitoring data acquired from landslide‐prone hillslopes in Oregon, USA, to predict the timing and magnitude of hydrologic response at multiple soil depths for 36‐hr intervals. We find that observation records as short as 6 months are sufficient for accurate predictions, and our model captures hydrologic response for high‐intensity rainfall events even when those storm types are excluded from model training. We conclude that machine learning can provide an accurate and computationally efficient alternative to empirical methods or physical modeling for landslide hazard warning.