Mapping geomorphic process domains to predict hillslope sediment size distribution using remotely-sensed data and field sampling, Inyo Creek, California

Tuesday, 16 December 2014
Shirin Leclere, Leonard S Sklar and Jennifer Rose Genetti, San Francisco State University, San Francisco, CA, United States
The size distribution of sediments produced on hillslopes and supplied to channels depends on the geomorphic processes that weather, detach and transport rock fragments down slopes. Little in the way of theory or data is available to predict patterns in hillslope size distributions at the catchment scale from topographic and geologic maps. Here we use aerial imagery and a variety of remote sensing techniques to map and categorize geomorphic landscape units (GLUs) by inferred sediment production process regime, across the steep mountain catchment of Inyo Creek, eastern Sierra Nevada, California. We also use field measurements of particle size and local geomorphic attributes to test and refine GLU determinations. Across the 2 km of relief in this catchment, landcover varies from bare bedrock cliffs at higher elevations to vegetated, regolith-covered convex slopes at lower elevations. Hillslope gradient could provide a simple index of sediment production process, from rock spallation and landsliding at highest slopes, to tree-throw and other disturbance-driven soil production processes at lowest slopes. However, many other attributes are needed for a more robust predictive model, including elevation, curvature, aspect, drainage area, and color. We combine tools from ArcGIS, ERDAS Imagine and Envi with groundtruthing field work to find an optimal combination of attributes for defining sediment production GLUs. Key challenges include distinguishing: weathered from freshly eroded bedrock, boulders from intact bedrock, and landslide deposits from talus slopes. We take advantage of emerging technologies that provide new ways of conducting fieldwork and comparing field data to mapping solutions. In particular, cellphone GPS is approaching the accuracy of dedicated GPS systems and the ability to geo-reference photos simplifies field notes and increases accuracy of later map creation. However, the predictive power of the GLU mapping approach is limited by inherent uncertainty in remotely sensed data and aerial imagery. This work is a contribution toward the long-term goal of reliable and automated mapping of hillslope sediment size distributions for use in sediment budgets and hazard delineation, and for understanding the feedbacks between climate, erosion and topography that drive sediment production.