Quantifying suspended sediment sources during runoff events in headwater catchments using spectrocolorimetry

Understanding hydro-sedimentary dynamics at the catchment scale requires high temporal resolution data on suspended sediments such as their origin, in addition to the common measurements of sediment concentrations and discharges. Some rapid and low-cost fingerprinting methods based on spectroscopy have recently been developed. We investigated how visible spectra could be used to predict the proportion of various source materials in suspended sediment samples, paying particular attention to the potential alteration of spectrocolorimetric signatures between soils and suspended sediments during transport. The 22-km(2) Galabre catchment, France, is composed of black marls, limestones, molasses, undifferentiated deposits and gypsum. Forty-eight source materials were sampled and 328 suspended sediment samples were collected at the outlet during 23 runoff events. Measurements were taken with a diffuse reflectance spectrophotometer on dried samples. As the erosion processes are particle size selective, five particle size fractions of source material were measured in order to assess the potential alteration of the fingerprint signatures. As the biogeochemical processes occurring in the river could also affect the signatures, source materials were immersed in the river for durations ranging from 1 to 63 days and subsequently measured. Finally, partial least-squares regression models were constructed on 81 artificial laboratory mixtures to predict the proportions of source materials. The spectrocolorimetric measurements discriminated the primary source materials but not the Quaternary deposits. As the gypsum was not conservative, only the black marls, molasses and limestones were used in the fingerprinting procedure. The construction of the partial least-squares regression models led to a median absolute error of 1.1%. This error increased to 3.9% when the models were applied to source samples with: (1) different particle sizes; (2) different durations spent in the river; or (3) different origins than those used for their construction. The effect of particle size on the fingerprinting procedure was larger than the effect of biogeochemical reactions or the spatial variability of the spectrocolorimetric signatures. Half of the 23 runoff events analysed exhibited huge variations in the source proportions from one sediment sample to another. The spectrocolorimetric fingerprinting approach was able to quantify routinely the proportion of primary source materials in all suspended sediment samples collected during runoff events. The high temporal resolution of the predicted proportions revealed that only analysing three or four suspended sediment samples during a runoff event could lead to a misunderstanding of the hydro-sedimentary processes for more than half of the investigated runoff events.