Predicting suspended sediment loads at a catchment scale: A comparison between models

Owing to financial and other constraints on natural resource management, comprehensive data on the physical processes that determine instream water quality are rare. Consequently, managers are looking to models as decision-making tools. One important application is to identify areas in greatest need of erosion management. There are many ways that suspended sediment loads can be assessed. The main constraints in the determination of loads are the data available and the scale at which this information can be applied. This study has used two models to estimate the suspended sediment load in a number of catchments in the mid-region of the Murrumbidgee River catchment. The load derived using an empirical estimation technique is compared to the load predicted from a semidistributed, lumped conceptual model (SedNet). Both these models can predict suspended sediment loads at a catchment scale with minimal data requirements compared to more complex physics-based models. Even so, it is important to recognise that these models predict loads at different time scales and include different assumptions, which can result in significant differences in the derived suspended sediment load. Nevertheless, managers rely on such models because of the reduction in available data to directly measure suspended sediment load and the increasing pressure for effective resource allocation. Hence, it is necessary to ensure consistency in the models used to assess water quality. Three sub-catchments in the Mid-Murrumbidgee River catchment - Tarcutta, Muttama and Jugiong creek catchments - were used for the suspended sediment load comparison. The comparison presented in Table 1 indicates that the long-term steady-state prediction from SedNet is within the statistical uncertainty for two of the catchments (Muttama and Tarcutta creek catchments) on a per area basis. However, it seems that there is a large discrepancy between the empirical model estimate and SedNet prediction for the Jugiong creek catchment. (Table Presented) The large discrepancy for Jugiong Creek may indicate that there are serious inaccuracies in the models used to predict suspended sediment load. Possible reasons may include the limitation that suspended sediment concentration (SSC) measurements used to estimate the load for the Jugiong creek catchment are not necessarily representative of the long term load. Another possibility considered is that SedNet may overpredict the amount of suspended sediment from its sediment sources. Supporting this theory is the high gully erosion density in Jugiong compared to the other two catchments. An over-prediction in suspended sediment from gully erosion may explain the contradiction between modelled load estimates for Jugiong. Owing to the uncertainty in both models, it is unlikely that either is correct. The use of these models as a tool to support policy and management prioritization is jeopardized when differences in model outputs contradict the ranking of catchments. Tarcutta creek catchment is the highest contributor of suspended sediment to the Murrumbidgee River using the empirical sediment rating curve model but the Jugiong creek catchment is highest when using the SedNet model. This research illustrates that differences between models will greatly affect the types of decisions that can be made based on the output of these models. It is clear that for effective natural resource management, a better understanding of the uncertainty and limitations of such models is needed.