Spectral fingerprinting: characterizing suspended sediment sources by the use of VNIR-SWIR spectral information

Abstract

as well as at several subcatchment outlets. All samples were dried and measured for spectral 50 properties in the laboratory using an ASD spectroradiometer. Colour parameters and physically based 51 features (e.g. organic carbon, iron oxide and clay content) were calculated from the spectra. Principal 52 component analyses (PCA) were applied to all three types of samples to determine natural clustering 53 of samples, and a mixing model was applied to determine source contributions. 54Results and discussion: We found that fine sediment stored in the river bed seems to be mainly 55 influenced by grain size and seasonal variability, while sampling location -and thus the effect of 56 individual tributaries or subcatchments -seem to be of minor importance. Suspended sediment 57 sources were found to vary between, as well as within, flood events; although badlands were always 58 the major source. Forests and grasslands contributed little (<10%) and other sources (not further 59 determinable) contributed up to 40%. The analyses further suggested that sediment sources differ 60 among the subcatchments and that subcatchments comprising relatively large proportions of 61 badlands contributed most to the four flood events analyzed. 62Conclusions: Spectral fingerprints provide a rapid and cost-efficient alternative to conventional 63 fingerprint properties. However, a combination of spectral and conventional fingerprint properties 64 could potentially permit discrimination of a larger number of source types. 81Knowledge of sediment provenance is a key factor in understanding sediment transport and delivery 82 processes and thus a first step in the design of sustainable watershed management strategies 88with little or no vegetation […] in an intensely dissected landscape" (Gallart et al. 2002). Lithology is a 89 major factor for badland development and though they are commonly considered characteristic of 90 dryland regions they also occur in more humid climates with high topographic gradients and intense 91 rainstorms (Gallart et al. 2002). Despite badlands being considered as major sediment sources, 92 significant changes in the colour of the suspended sediments have been observed between, and even 93 within, runoff events, suggesting the influence of varying sources. 94A direct approach to trace the origin of sediment is a method called fingerprinting. It is founded on the 95 principal assumptions that: (1) potential sediment sources can be discriminated based on a set of 96 characteristic properties ("fingerprints"); and (2) the comparison of these source characteristics with 97 those of (suspended) sediment allow for determination of relative source contribution (Collins and 100However, the adaptation of the technique as a management tool is hampered due to several reasons. 101Most importantly, the choice of successful fingerprint properties is highly site-specific and the lack of Poulenard et al. 2009Poulenard et al. , 2011 Martínez-Carreras 2010a, 2010b, 2010c Evrard et al. 2012; Legout et 108 al. 2013; Bros...

Water and sediment fluxes in Mediterranean mountainous regions: comprehensive dataset for hydro-sedimentological analyses and modelling in a mesoscale catchment (River Isábena, NE Spain)

“…3.3.1) and 152 soil samples from the main potential sediment source areas within the Isábena catchment (see Fig. 1 in Brosinsky et al, 2014a). Details on sampling and spectral measurement procedures can be found in the subsequent sections.…”

“…For each sample, four readings were taken and subsequently averaged, with the sample rotated 90 • after every reading to reduce illumination effects. For more details see Brosinsky et al (2014a).…”

“…Again, spectral measurements followed the laboratory protocol described above. Since we detected spectral differences between loose material and material on filters that are most likely caused by the alignment of sediment particles during vacuum filtration, the measurements should not be compared directly (Brosinsky et al, 2014a).…”

Water and sediment fluxes in Mediterranean mountainous regions: Comprehensive dataset for hydro-sedimentological analyses and modelling in a mesoscale catchment (River Isábena, NE Spain)

“…Details on sampling and spectral measurement procedures can be found in the subsequent sections. Further details on the spatial location of sampling sites as well as the use of these data in (spectral) fingerprinting studies to trace back the origin of suspended sediments are given in Brosinsky et al (2014aBrosinsky et al ( , 2014b.…”

“…Francke et al, 2014;López-Tarazón et al, 2012;López-Tarazón and Batalla, 2014). Sediment connectivity for two sub-catchments of the Isábena basin was analysed based 20 on digital elevation models obtained from airborne LiDAR data and surface cover fractions obtained from airborne hyperspectral imagery , while sediment origins were traced back by spectral fingerprinting based on measured spectra of suspended sediment samples and soil samples of potential sediment sources (Brosinsky et al, 2014a(Brosinsky et al, , 2014b. Furthermore, for the process-based, spatially semi-distributed model WASA-SED developed for water and sediment transport in dryland catchments (Mueller et al, 2010;Bronstert et al, 2014), a framework for studying the effect of model 25 enhancements by, e.g., a higher number of meteorological monitoring stations, improved water and sediment calibration data, and modifications in process formulations, was tested in the Isábena study area based on parts of the dataset presented here.…”