Assessment of different colour parameters for discriminating potential suspended sediment sources and provenance: A multi-scale study in Luxembourg

Abstract

Sediment colour has recently been used successfully to estimate suspended sediment sources in small catchments using the fingerprinting approach. The methodology offers opportunity for further research since it provides a rapid and cheap means for investigating sediment sources. However, the colour-based fingerprinting approach has not yet been tested in medium and large catchments. This paper aims to test whether colour parameters are capable of discriminating sediment sources in a nested system of seven sub-catchments ranging from 0.7 to 247 km 2 of the Attert River catchment, NW Luxembourg. Time-integrated suspended sediment samples and samples of potential sediment sources (land-use types and channel banks) were collected in all catchments. Sediment colour was then computed from diffuse reflectance spectrometry measurements (ASD FieldSpec-II spectrometer, 0.4–2.5 µm) taken over the visible wavelength range. Twenty-four colour parameters were derived from several colour space models (CIE xyY, CIE XYZ, RGB, Munsell HVC, Helmholtz chromaticity, CIELUV and CIELAB) and their ability to discriminate potential suspended sediment sources and provenance was evaluated and compared. Results demonstrated that time-integrated suspended sediment samples collected in the study catchments had statistically different colour values. Moreover, these values always represented a mixture of the colour values measured on potential suspended sediment sources in the catchment. Inter-source colour contrasts (land-use types and channel banks) were observed in all catchments (Kruskal–Wallis H-test). However, although colour is able to distinguish potential sediment sources in small catchments, the level of source discrimination decreases as the catchment size increases, probably due to heterogeneous geology and pedology, intra-source variability and to source overlap. Nevertheless, in the studied medium-sized catchments (ranging from 19.4 to 247 km 2), colour could differentiate between topsoil and sub-surface (i.e. channel bank) material and/or up to three sources types. No single colour model had discrimination power across catchments, instead in each catchment a different combination of colour parameters gathered from different colour space models produced optimal discrimination of potential sediment sources. Furthermore, a colour-based fingerprinting approach did not possess potential for integrating spatial provenance and source type information because colour parameters could not discriminate between contrasting geological sub-areas. In summary, although colour parameters were not capable of discriminating the range of land-use type and channel banks as potential suspended sediment sources in medium-sized catchments, they afford substantial information and could be integrated into the classical fingerprinting approach together with other constituents (e.g. geochemistry, radionuclides and/or organic compounds).