Using rare earth elements to monitor sediment sources from a miniature model of a small watershed in the Three Gorges area of China

Understanding soil erosion processes at different landscape positions is important in order to predict and control watershed soil losses. Rare earth elements (REEs) can be used to trace eroded soil sources but their efficacy may be soil dependent. We constructed a miniature watershed model of a small watershed located in the Three Gorges Area of China, and used oxides of eight REEs to trace the erosion of a purple soil. The miniature watershed was divided into eight regions containing a different landform type as a potential sediment source. A different REE was applied in each region. Redistributions of the REEs under three successive simulated rainfall events with intensities of 1.0, 1.5 or 2.0mmmin−1 were examined. The percentage contribution from each region to the total soil loss from the watershed fluctuated relative to landform type during the three rainstorms. Contributions from the lower main gully decreased before stabilizing, while those from the upper main gully increased before decreasing, and those from other sources all increased before stabilizing. Overall, the contribution of the gully system, comprising main and branch gullies, was greater than that of the slopes. Contributions from the gully system tended to decrease with increases in rainfall intensity and rainstorm duration while those from the slopes increased. A comparison of the calculated and actual soil loss masses indicated that the accuracy of the REE tracing method was less for the coarse textured purple soil than those previously found for fine textured soils. The increased errors, likely due to the assumption used in the calculation that there is no particle size selectivity during erosion, needs to be addressed. This pilot study provided a technical reference for the use of REEs in monitoring sediment sources from a natural watershed, and a theoretical basis for soil conservation in the Three Gorges Area.