Quantifying catchment scale sediment source using composite fingerprinting technique

Abstract: Severe soil and water loss and land destruction, diffused sediment pollution caused by the excessive sediment impairs water quality and plays a key roil on the transfer and fate of nutrients and contaminants. Researches on catchments (river) sediment provenance, ascertaining sediment flux into the pond, river and reservoir and apportioning catchments sediment sources all are very essential and instructive. Accordingly, this study makes an attempt to: 1) assess the potential for using composite fingerprint technique in tracing pond sediment from a agricultural catchment (10.7 hm2) in the Three Gorge Reservoir Region; 2) and to give an exploration of the sediment sources defined in term of land uses. Based on detailed field investigation and well arrangement for the sampling campaign, the small agricultural catchment was qualitatively divided into three main sources (namely woodland and pasture, slope cropland and paddy field). A total of 15 potential source samples were collected from those three land use areas (with sample sizes of 7, 4, 4 for slope cropland, paddy field and woodland and pasture, respectively). Each source sample comprises 5-7 scrapes of the surface materials (c. 2 cm) retrieved along the slope; and three cores (upper 5 cm of the pond sediment) were extracted from the pond and followed a homogeneous mixing so as to represent contemporary sedimentation during the latest 10 year. 16 geochemical properties, including radionuclides (137Cs, 210Pbex, 226Ra), organic and inorganic constituents (total C, total N, and total P), base cation (K, Mg), trace metal (Mn), heavy metals (Cd, Co, Cr, Cu, Ni, Pb and Zn) as well as grain size composition, were measured and statistically analyzed in order to determine the optimized composite fingerprints. 5 tracers (cf. total C, 137Cs, 226Ra, K and Zn) were selected constituting the integrated fingerprints through a two steps statistical analysis involving Kruskal-Wallis H test (K-W H test) and discrimination function analysis (DFA), which all together are capable of discriminating 87% of the source samples correctly. Followed with an application of a multivariate linear mixing model, relative contributions of those three sources were apportioned. Results showed that approximately 84 % of the sediment reserved in the pond was originated from slope cropland, 14% of the sediment was from paddy field, wood land and pasture makes a 2% contribution to the pond sediment. The study catchment is a typical small agricultural catchment with no drain system where runoff went down the slope through plots and converged in the lower paddy field in which temporary sedimentation was supposed to occur. While there may exist some extent of erosion, especially during early summer when the paddy field was deeply ploughed and evenly harrowed for rice, thus might suffer heavy storm events at that time. The case study demonstrates that the composite fingerprint technology provides an alternative for elucidating the sediment sources of the agricultural catchments in the Three Gorges Reservoir Region.