The use of visible and near-infrared reflectance measurements for identifying the source of suspended sediment in rivers and comparison with geochemical fingerprinting

Purpose Visible and near-infrared (Vis-NIR) reflectance measurementsmay be an alternative technique to identifysuspended sediment sources in streams of headwater catchments.In this study, we examined if Vis-NIR reflectancemeasurements are capable of estimating sediment source contributionsto sediment yield and compared this technique witha more conventional (i.e. geochemical) technique.Materials and methods Two headwater catchments inEthiopia, Unta (2,052 ha) and Desera (1,657 ha), wereanalysed with the same techniques in order to find similaritiesand differences in the results obtained. The first techniqueused Vis-NIR spectral analysis as a fingerprint, using a partialleast squares regression model. The second technique was aquantitative composite fingerprinting technique using geochemicalanalysis of source materials and suspended sedimentsamples. As a comparison, the partial least squares model wasalso used on the geochemical data. In August and September2009, 30 soil samples of three different land uses (landslides,croplands, and grazing lands) and 21 suspended sedimentsamples at the catchment outlet were collected. Source sampleswere sieved to <63 μm. Geochemical analyses consistedof total element concentrations, percentage carbon, percentagenitrogen, and atom percentage 15N and δ13C. Reflectancemeasurements were taken on dried source samples with aspectrometer.Results and discussion Neither technique was able to predictthe contributions of the three land use types; they could onlydistinguish between landslide and topsoil material. The agreementbetween the results of both techniques was significantfor the Unta catchment (R2=0.80) but not for the Deseracatchment (R2=0.39). The uncertainty of the technique usingVis-NIR reflectance measurements was slightly higher thanwith the geochemical approach. Both techniques revealed thattopsoil erosion played an important role during storm runoffdischarges. Using the partial least squares model for the geochemicaldata revealed that uncertainty can differ greatlywhen using other statistical techniques.Conclusions The quantitative composite fingerprinting techniqueusing spectral signatures from both source andsuspended sediment samples was able to quantify the contributionof two source materials (landslides and topsoil). Itprovided a faster and more cost effective alternative to theconventional geochemical procedure.status: publishe