Minnesota River TMDL Model: Representation of Sediment Sources and Controls

The Minnesota River HSPF model is a comprehensive watershed simulation tool initially developed for nutrient and DO analysis. The model covers an area of 12,200 mi2 and is currently being used by the Minnesota Pollution Control Agency (MPCA) to address turbidity TMDLs in multiple segments. Turbidity impairments are primarily due to inorganic sediment loads, which derive from a variety of sources, including sheet/rill erosion, ravines (gullying), collapse of banks and bluffs, and mobilization of sediment from the stream channel. A fundamental problem for any large scale watershed model is that many of the processes that determine sediment load operate at the field or reach scale. These can only be approximated in a large scale model, regardless of whether it is a lumped model (like HSPF or SWAT) or a distributed model. Even if individual fields/reaches were resolved in the model, data are not available to characterize conditions or practices at this scale.To provide a tool useful for TMDL implementation it is crucial that the model correctly attributes the loads from different source categories while allowing representation of a range of management practices. The first step in source attribution is ensuring that the model is rigorously calibrated; however, calibration to downstream measurements alone is not sufficient to determine a unique solution. Further constraints are needed. Model performance for sediment was improved by incorporating results from detailed hydraulic models, followed by a detailed reach-by-reach calibration. Additional constraints were provided by (1) use of radioisotope data which distinguishes recent (surface erosion) from older (subsurface) sediment sources, (2) information on rates of bluff collapse, (3) research and models on small watersheds, and (4) expert elicitation. The final model is able to meet calibration and validation criteria for prediction of sediment concentration and load while also conforming to source attribution constraints.The calibrated model is being used to simulate a variety of management options and implementation strategies. These include changes to land use (increase in CRP, shift to perennial crops), changes to cropping systems (increased crop residue, cover crops, elimination of surface tile inlets), drainage management (reduction in erosive power of flows from tile outlets, controlled drainage with storage of runoff), and bank stabilization to reduce channel scour rates.The source attribution work indicates that approximately one third of the sediment load is derived from upland sheet/rill erosion, with the remainder coming from ravine incision and bank/bluff collapse, all mediated by channel processes. Large reductions in load cannot be achieved by addressing only one component; however, there are also interactions among components. For instance, management that reduces upland load by controlling runoff peaks can also reduce the rate of channel scour. Finally, the source categories have important temporal differences; bluff collapse, while related to soil moisture conditions, tends to occur intermittently and is difficult to predict, while ravines and channel erosion loads are more concentrated in rare, low recurrence events than sheet/rill erosion. Interpretation of the temporal frequency and duration, in addition to the magnitude, of TMDL targets has direct bearing on the efficacy and feasibility of management strategies.The Minnesota River model provides a large-scale, organizing framework for addressing sediment (and nutrient) TMDLs in the basin that provides a credible description of both sources and potential management strategies. There will continue to be an important role, however, for field scale models and demonstration projects to refine strategies for achieving needed load reductions at the local level.