ABSTRACT
Knowledge of sediment movement throughout a catchment environment is essential due to its influence on the character and form of our landscape relating to agricultural productivity and ecological health. Sediment fingerprinting is a well-used tool for evaluating sediment sources within a fluvial catchment but still faces areas of uncertainty for applications to large catchments that have a complex arrangement of sources. Sediment fingerprinting was applied to the Manawatu River Catchment to differentiate 8 geological and geomorphological sources. The source categories were Mudstone, Hill Subsurface, Hill Surface, Channel Bank, Mountain Range, Gravel Terrace, Loess and Limestone. Geochemical analysis was conducted using XRF and LA-ICP-MS. Geochemical concentrations were analysed using Discriminant Function Analysis and sediment un-mixing models. Two mixing models were used in conjunction with GRG non-linear and Evolutionary optimization methods for comparison. Discriminant Function Analysis required 16 variables to correctly classify 92.6% of sediment sources. Geological explanations were achieved for some of the variables selected, although there is a need for mineralogical information to confirm causes for the geochemical signatures. Consistent source estimates were achieved between models with optimization techniques providing globally optimal solutions for sediment quantification. Sediment sources was attributed primarily to Mudstone, ≈38-46%; followed by the Mountain Range, ≈15-18%; Hill Surface, ≈12-16%; Hill Subsurface, ≈9-11%; Loess, ≈9-15%; Gravel Terrace, ≈0-4%; Channel Bank, ≈0-5%; and Limestone, ≈0%. Sediment source apportionment fits with the conceptual understanding of the catchment which has recognized soft sedimentary mudstone to be highly susceptible to erosion. Inference of the processes responsible for sediment generation can be made for processes where there is a clear relationship with the geomorphology, but is problematic for processes which occur within multiple terrains.
