Process recognition in multi-element soil and stream-sediment geochemical data

Stream-sediment and soil geochemical data from the Upper and Lower Coastal Plains of South Carolina (USA) were studied to determine relationships between soils and stream sediments. From multi-element associations, characteristic compositions were determined for both media. Primary associations of elements reflect mineralogy, including heavy minerals, carbonates and clays, and the effects of groundwater. The effects of groundwater on element concentrations are more evident in soils than stream sediments. A “winnowing index” was created using ratios of Th to Al that revealed differing erosional and depositional environments. Both soils and stream sediments from the Upper and Lower Coastal Plains show derivation from similar materials and subsequent similar multi-element relationships, but have some distinct differences. In the Lower Coastal Plain, soils have high values of elements concentrated in heavy minerals (Ce, Y, Th) that grade into high values of elements concentrated into finer-grain-size, lower-density materials, primarily comprised of carbonates and feldspar minerals (Mg, Ca, Na, K, Al). These gradational trends in mineralogy and geochemistry are inferred to reflect reworking of materials during marine transgressions and regressions. Upper Coastal Plain stream-sediment geochemistry shows a higher winnowing index relative to soil geochemistry. A comparison of the 4 media (Upper Coastal Plain soils and stream sediments and Lower Coastal Plain soils and stream sediments) shows that Upper Coastal Plain stream sediments have a higher winnowing index and a higher concentration of elements contained within heavy minerals, whereas Lower Coastal Plain stream sediments show a strong correlation between elements typically contained within clays. It is not possible to calculate a functional relationship between stream sediment–soil compositions for all elements due to the complex history of weathering, deposition, reworking and re-deposition. However, depending on the spatial separation of the stream-sediment and soil samples, some elements are more highly correlated than others.