Estimating specific surface area of fine stream bed sediments from geochemistry

Specific surface area (SSA) of headwater stream bed sediments is a fundamental property which determines the nature of sediment surface reactions and influences ecosystem-level, biological processes. Measurements of SSA – commonly undertaken by BET nitrogen adsorption – are relatively costly in terms of instrumentation and operator time. A novel approach is presented for estimating fine (<150 μm) stream bed sediment SSA from their geochemistry – after removal of organic matter – for agricultural headwater catchments across 15,400 km² of central England, UK. From a regional set of 1972 stream bed sediment sites with common characteristics for which geochemical data were available, 60 samples were selected – based on maximising their variation in Al concentrations – and their BET SSA measured by N₂ adsorption. After careful selection of potential regression predictors following a principal component analysis and removal of a subset of samples with the largest Mo concentrations (>2.5 mg kg⁻¹), four elements were identified as significant predictors of SSA (ordered by decreasing predictive power): V > Ca > Al > Rb. The optimum model from these four elements accounted for 73% of the variation in bed sediment SSA (range 6–46 m² g⁻¹) with a root mean squared error of prediction – based on leave-one-out cross-validation – of 6.3 m² g⁻¹. It is believed that V is the most significant predictor because its concentration is strongly correlated both with the quantity of Fe-oxides and clay minerals in the stream bed sediments, which dominate sediment SSA. Sample heterogeneity in SSA – based on triplicate measurements of sub-samples – was a substantial source of variation (standard error = 2.2 m² g⁻¹) which cannot be accounted for in the regression model.