Abstract
Stochastic-threshold theory was developed using theoretical modeling and theorizes that river erosion occurs during discrete events rather than constantly through time. Many models assume relative homogeneity in natural systems over longer timescales. However climatic forcings on river systems vary greatly throughout the course of time and so these models are likely an oversimplification of the erosion patterns it aims to predict. While there is some real-world evidence to support the stochastic model, there is still a lack of direct tests in this matter. St. Anthony Falls initially formed approximately 12 ka when Glacial River Warren-- a high volume stream draining Lake Agassiz-- downcut a previously excavated river channel that had been filled in with glacial till near present-day St. Paul, Minnesota. The falls began incising what is now the Mississippi River gorge approximately 10 ka when they reached the confluence of the Mississippi and Minnesota Rivers. As the falls passed by the confluences of the tributary streams, local base level fell for the tributaries forming a series of knickpoints along this stretch of the Mississippi River. The models that are mentioned above offer predicted morphological differences between stochastically and continuously eroding streams. One suggested ideal case for observing such morphological conditions could be a river network that is responding to a rapid base level drop. Conditions at the knickpoints along the Mississippi River selected for this study provide such a platform. This study will utilize 3D imaging software to observe erosional patterns along the Mississippi River at knickpoints in selected tributaries below St. Anthony Falls and above the Mississippi River and Minnesota River confluence. We will collect and compile images from multiple locations over set time intervals to attempt to determine whether erosion is continuous or stochastic in nature. The underlying bedrock in the areas creates a distinct recessed profile. The concavity of the knickpoints encourages rock falls because a large portion of erosion occurs from the bottom. If stochastic erosion is occurring we might expect to see that the changes in the 3D models occur in distinct points on the knickpoints through time, as compared to continuous erosion, which should result in consistent changes across the whole knickpoint.