Predicting hillslope sediment size distribution using remote sensing data, Inyo Creek, California

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. To address this knowledge gap, we map the steep granitic catchment of Inyo Creek, eastern Sierra Nevada, California and categorize geomorphic landscape units (GLUs) by overlaying basic GIS attributes to create polygons of similar inferred sediment production process regimes. Key attributes include elevation, slope, aspect, and land cover, which varies across 2 km of relief from bare bedrock cliffs at higher elevations to vegetated, regolith-covered slopes at lower elevations. We expect that factors that influence temperature and water residence time, and thus the intensity of chemical versus mechanical weathering, will correlate with resulting hillslope sediment size. For example, GLUs constructed from binned combinations of slope, elevation and aspect were used to predict three categories of sediment size on an ordinal scale. We used a map of predicted sediment size to guide field measurements, using point counts and photogrammetry to quantify hillslope surface size distributions. Areas predicted to have relatively large sediments were primarily covered in boulders and cobble-sized particles, whereas areas predicted to have small sediments were primarily covered in scree and gruss. Although hillslope sediment size at Inyo Creek correlates strongly with elevation, incorporation of slope and aspect creates a significantly better predictive model. We combine this result with supervised classification of aerial images using eCognition to estimate that more than half the catchment area produces boulder and cobble-rich sediment. Further analysis will include characterizing the local contributing areas to each field sampling point to account for sediment production on bare bedrock surfaces upslope. We are also exploring whether other landscape attributes, such as hillslope curvature, as well as estimates of spatial variation in erosion rate, can be used in refining catchment-wide predictions of hillslope sediment size.