Mining soil databases for landscape-scale patterns in the abundance and size distribution of hillslope rock fragments

摘要

Landscape-scale variation in rock fragments on soil-mantled hillslopes is poorly understood, despite the potential importance of rock fragments in soil weathering and coarse sediment supply to river networks. We explored the utility of soil survey databases for data mining, with the goals of identifying landscape-scale patterns in the abundance and size distribution of rock fragments (diameter D>2mm) and potential controls on grain size production. We focus on data from three regions: the Hawaiian Islands, and the Sierra Nevada and Cascade Mountains, where elevation transects span a range of environmental conditions. We selected pedons from pits dug on hillslopes with active soil production and transport. For the 27 pedons selected, we constructed depth-averaged grain size distributions and calculated the mass fraction of rock fragments (F _(RF)) and the median rock fragment grain size (D _(50RF)). We also categorized as bimodal, size distributions with a clear 'breakpoint' between fine and coarse modes. Several strong patterns emerge from the data. We find rock fragments in 85% of the pedons, primarily in distinct coarse modes within bimodal size distributions. Values of F _(RF) and D _(50RF) are strongly correlated, although the best-fit power law scaling between F _(RF) and D _(50RF) differs between the warmer Hawaiian, and colder Sierra Nevada and Cascade Mountain sites. We also find a regional contrast in the variation in F _(RF) with elevation; F _(RF) declines with elevation in Hawaii, but increases in the mainland sites. Although this contrast could be an artifact of variable lithology, precipitation may influence many patterns in the data. Lower mean-annual precipitation correlates with higher F _(RF), dominantly bimodal distributions and surface enrichment in the vertical distribution of rock fragments. These observations may be useful in refining models of coarse sediment supply to rivers, and suggest opportunities for future work to test mechanistic hypotheses for rock fragment production on soil-mantled hillslopes.