Search

In this study, surface runoff and sediment yields, simulated by the Soil and Water Assessment Tool (SWAT) for the Xiangxi catchment, China, are analysed at the spatial level of hydrologic response unit (HRU). This allows for a more precise targeting of best management practices than analysis at the coarser sub-basin level. It also provides the opportunity to validate simulated surface runoff and sediment yield by evaluating the plausibility of their spatial variation within the watershed. The results indicate that satisfactory model performance at a gauge does not guarantee plausible results at HRU level. Both surface runoff and sediment yields were found to vary reasonably with land use and soil types, but not with slope. The credibility of simulated surface runoff was increased by modifying selected SWAT algorithms and parameters. However, these modifications had only a marginal effect on sediment yield, which can be explained by the nonlinear relationship between sediment yield and HRU area in the modified Universal Soil Loss Equation. This study demonstrates that evaluation of model output at HRU level is highly recommended.Editor D. Koutsoyiannis; Guest editor V. Krysanova [ABSTRACT FROM AUTHOR]

Nutrient transportation from agricultural land to surface water has received much attention throughout the world since it plays an increasingly important role in affecting the water environment. Understanding the nutrient cycle in agricultural systems is essential to reduce non-point pollution for improving water quality. This paper describes the preliminary evaluation of the Xinanjiang-Phosphorus (XAJ-P) model for suspended sediment (SS) and phosphorus (P) estimation in the Xitiaoxi catchment in south-eastern China. It estimates the sediment yields and phosphorus loads on a cell-by-cell basis with a daily time step. The results show that an average of 127.4 ton/yr P was exported to the river in the catchment. Spatial distribution of P loads indicates that the non-point source load from arable land had a dominant contribution with an export rate of 1.63 to 4.92 kg/(ha·yr). It is also found that an annual average P load of 63.7 t was exported from point sources to the rivers. P budget analysis indicate that average P input and output were 71.3 kg/(ha·yr) and 46.2 kg/(ha·yr) respectively, while the total P utilization efficiency was 59.3%, leading to an average P surplus of 25.1 kg/(ha·yr) in the arable land of the Xitiaoxi catchment. The application of XAJ-P in the catchment shows that the developed model can be used as a major tool to simulate long-term daily P loads at the catchment scale. [ABSTRACT FROM AUTHOR]

• The key runoff variables influencing sediment yield were screened at event scale. • The flood events were classified into three regimes by clustering analysis method. • The runoff-sediment relationship in relation to the flood regimes was analyzed. • The impacts of soil and water conservation measures on regime-based sediment flow behavior were investigated. Understanding the changes in runoff-sediment relationships is a great help for implementing soil and water conservation measures, particularly in regions with severe erosion. We selected a typical coarse sandy catchment on the Loess Plateau to investigate the changes of the runoff-sediment relationships with a data set of 62 years. A change point occurred in 1979, dividing the runoff and sediment load series into a baseline period (1954–1979) and a changing period (1980–2015). A total of 342 flood events were classified into three regimes using hierarchical clustering method. Regime A (162 events) was characterized by the shortest duration, lowest flood crest, and the least flow depth. Regime B (165 events) was characterized by a medium runoff depth, medium flow variability, and medium duration. Regime C merely include fifteen events with longest flood duration, the highest runoff depth, and the largest peak discharge. The sediment yield of flood regime A, B, and C accounted 14.2 % (1.09 × 108 t), 51.8 % (3.99 × 108 t), and 34.0 % (2.62 × 108 t) of the total sediment yield, respectively. The Support vector machines method was applied to established models to predict event sediment yield. It is demonstrated that the performance of models are good for different flood regimes. [ABSTRACT FROM AUTHOR]