Landslide Inventories and Grain Size Trends in Debris Flow-Dominated Channels

Lyons, Nathan

ABSTRACT: Recent research in the field of geomorphology has approximated the location in mountainous headwater streams where the transition from debris flow to fluvial-dominated channel erosional processes occurs. The dominant channel process controls channels’ sediment transport capacity and geomorphology, which affects streambed grain size and hillslope response. The transition has often been observed at a scaling break in a channel’s plot of log drainage area versus log slope (DS Plot) that can be created from a digital elevation model (DEM). Little research has investigated channel process transitions or explore the wide-ranging implications and applications of the dominant channel process type throughout geomorphic realms in the southern Appalachians, an area with a long history of debris flow activity, and complex sediment transport and hillslope response dynamics. The importance of the dominant channel process relative to the ability to inventory landslides and evaluate streambed grain size distribution in the Great Smoky Mountains of the southern Appalachians is investigated here. Researchers continue to improve inventory accuracy with new technology and techniques, though rarely is the dominant channel process incorporated into the inventory process despite the ability to confine debris flow activity to areas upstream from the scaling break. In this investigation of debris flow-dominated channels, we test an inventory technique that limits a land surface segmentation and classification algorithm to areas that are upstream from the scaling break in DS plots of five mountainous streams. 79% of surfaces were correctly classified when the algorithm incorporated the process-based analyses that limited the study area to areas upstream of the DS Plot scaling. 61% was correctly classified when the entire study area was included. This method allows a more focused approach to statistically characterize the land surface which resulted in a more accurate inventory in this study area. In the second investigation, a streambed grain size model was created with a modified form of the Shields equation in order to determine grain size trends of mountainous streams relative to the DS Plot scaling breaks. Typically, grain size becomes progressively finer downstream despite local factors that disrupt this trend. This observation has mostly been made in alluvial reaches far from headwaters where debris flows occur. In situ grain size measurements in debris flow-dominated streams in the Smoky Mountains, however, coarsen downstream. Stream wide grain size trends modeled with DEM inputs parameterized with field data also coarsen downstream where the maximum grain size occurs at approximately the DS plot scaling break. This rarely studied and observed trend may have important effects on geomorphology and stream ecosystems where it arises. A case study to illustrate a potential impact of downstream coarsening, the control of grain size in competitive advantages amongst trout species, is also presented here. Encroachment of exotic trout species are confining brook trout (Salvelinus fontinalis), the only salmonid native to the southern Appalachians, to headwater streams. Previous region-specific studies have investigated the impact of hydrologic, biologic, and land use controls on salmonid distributions. Streambed grain size availability may be an additional control on species fecundity since brook trout use a significantly smaller grain size than the invasive salmonids to construct redds, the streambed mounds in which salmonids deposit their eggs. The type of analyses required to conduct these investigation require datasets that spans multiple scales and methods of data collection. A considerable part of this work required reconciliation of data surveyed in the field and derived from remotely obtained topographic and spectral data. Problems common in geospatial data are addressed. Moderate correlations between field surveyed and modeled channel metrics were observed that provided further insight into the methodology of this study and dominant channel processes.