The soil erodibility factor (K) is used in empirical erosion models based on the Universal Soil Loss Equation to account for soil susceptibility to detachment and transport by rainfall and runoff. Whilst soil erodibility is ideally measured from long-term standard plots, in catchment-scale modelling it is more often estimated by applying pedo-transfer functions. These are either based on soil properties reported in soil databases, or attributed by experts on the basis of soil characteristics. The aim of this study was to evaluate the impact of the soil erodibility factor on the amount and distribution of suspended sediment loads generated by hillslope erosion within the LaTrobe River catchment, in Victoria (south-east Australia). Two soil erodibility factor sets were developed for hydrologic soil groups in the LaTrobe catchment. The first ('local') set was based on a Victorian soil database; soil erodibility was attributed by an expert soil scientist on the basis of topsoil texture, soil structure, geology, hydrological properties of the profile, and local knowledge. The second ('global') set was derived from a global soil erodibility dataset using the probabilistic distribution of K based on climatic conditions, skeleton (i.e. fraction > 2mm), organic matter content, and topsoil texture. The K factor in the 'local' set ranged from 0.015 to 0.055 Mg ha h ha(-1) MJ(-1) mm(-1), whereas soil erodibility in the 'global' set had higher absolute values but a smaller range (0.044-0.067 Mg ha h ha-1 MJ-1 mm-1). Importantly, the two sets differed in ranking soils from the most to the least erodible. A catchment scale model based on CatchMODS was used to assess suspended sediment loads from three erosion processes: hillslope erosion (which depended on soil erodibility), gully, and streambank erosion. The model estimated deposition of suspended sediment on hillslopes, floodplains and in reservoirs. Hillslope deposition was calculated using a hillslope sediment delivery ratio (HSDR), which is a calibration parameter. The two model configurations (i.e. informed by the two erodibility factor sets) were calibrated independently using annual suspended sediment load estimates at ten water quality monitoring stations of the catchment for the period 1990-2005. The model performance was assessed in terms of model efficiency of specific sediment yield predictions. The calibration of HSDR did reduce the impact of absolute values of soil erodibility estimates on hillslope net erosion; with higher HSDR calibrated for the local K configuration. However, the two model configurations resulted in different contribution of hillslope net erosion to suspended sediment loads: in the local K dataset configuration, hillslope net erosion contribution was estimated at 3.6 kt/y in the local K configuration (11% of a total of 34 kt/y estimated to reach Lake Wellington). In the global dataset configuration, hillslope net erosion was estimated at 9 kt/y (23% of an estimated total of 40 kt/y at the lake). The spatial distribution of the soil erodibility factor (K) resulted in a measurable impact on model performance; the global K configuration better matched specific sediment load observations across the catchment (efficiency of 0.32). The main difference in the attribution of K by the two approaches was due to the influence of climatic conditions. Analysis of the global dataset indicated that, other conditions being equal, soil erodibility in warm climates is lower than in temperate climates (Salvador Sanchis et al., 2008). Apparently, the local dataset underestimated the climatic effect on soil erodibility, and resulted in an overall underestimation of net hillslope erosion in the study catchment. These exploratory results will need to be further explored in future research.
