Background The water-level-fluctuation zone (WLFZ) area of the Three Gorges Reservoir (TGR) is approximately 349 km², where sedimentation process of sediment has been rapidly accelerated since 2010 when the reservoir was filled to an elevation of 175 meters, inducing potential risk to the aquatic ecological environment. In this study, sediment source fingerprinting was firstly introduced to quantify the source of sediment in the WLFZ for a secondary tributary of Yangtze River in the TGR.

Methods Sediment source apportionment for a specific watershed includes collecting of sediment of the WLFZ during low water-level period of the TGR, qualitative classification of source types (e.g. land uses and soil types) and sampling, screening of composite fingerprint factors, and quantifying source contribution using multivariate linear mixing model based on the least square method. About 71 source samples were collected from four land uses, and 12 sediment samples were collected from the WLFZ at both sides of 6 sections in the mainstream of the secondary tributary. Empirically, TOC, TN, TP, ¹³⁷Cs, ²¹⁰Pb_ex, ²²⁶Ra, K, Mg, As, Cd, Co, Cr, Cu, Mn, Ni, Pb, Ti, and Zn, were selected as tracing fingerprints. Properties including particle size composition, concentrations/specific activities for all fingerprints were tested and analyzed.

Results As for sediments, particle size is generally fine, enrichment rate of clay particles (less than 0.002 mm) reaches 1.60, specific surface area of which is 41% higher than that of source samples. The content of soil organic nutrients, soil alkali metals, most heavy metals and the specific activity of atmospheric deposition nuclides were positively correlated with the content of fine particles. Five fingerprint factors including TP, TOC, TN, As, and Co showed significant differences among the four land use samples, with a cumulative correct discrimination rate in classification about 87.3% for the four land use samples. Other six fingerprint factors (K, Mg, Cu, Ni, Mn, and As) showed significant differences between two different soil type samples, and 97.2% of the 71 samples from two geological zones were discriminated correctly. Multi-variate linear mixing model was used for quantifying sediment source, and the average relative contribution of forest-grassland, orchard (citrus orchard), slope cropland (dry land) and paddy field were 43%, 39%, 16% and 2%, respectively. As for the geological and geomorphic distribution of this Tributary basin, the average relative contribution of purple soil area is 51%, and that of lime soil area is 49%. Sediment source variation in terms of land use revealed comprehensive effects of differentiation of sediment yield under managements, sorting during slope-basin sediment transport process, disturbance of bank erosion from the WLFZ and interference of external sediment input from the reservoir water. The sediment source apportionment also suggested differencing in soil erodibility related to two typical parent materials.

Conclusions An attempt has been made to discriminate sediment sources in the TGR, that the feasibility of quantitative analysis of sediment sources in the WLFZ using sediment source fingerprinting was preliminarily verified. Sediment source fingerprinting provides vital decision-making support for the precise management of soil and water conservation for river basins in the TGR area.