Identifying Erosional Hotspots in Streams Along the North Shore of Lake Superior, Minnesota using High-Resolution Elevation and Soils Data.

摘要

Many streams on the North Shore of Lake Superior, Minnesota, USA, are impaired for turbidity driven by excess fine sediment loading. The goal of this project was to develop a GIS-based model using new, openly-available, high-resolution remote datasets to predict erosional hotspots at a reach scale, based on three study watersheds: Amity Creek, the Talmadge River, and the French River. The ability to identify erosional hotspots, or locations that are highly susceptible to erosion, using remote data would be helpful for watershed managers in implementing practices to reduce turbidity in these streams.;Erosion in streams is a balance between driving forces, largely controlled by topography; and resisting forces, controlled by the materials that make up a channel's bed and banks. New high-resolution topography and soils datasets for the North Shore provide the opportunity to extract these driving and resisting forces from remote datasets and possibly predict erosion potential and identify erosional hotspots. We used 3-meter LiDAR-derived DEMs to calculate a stream power-based erosion index, to identify stream reaches with high radius of curvature, and to identify stream reaches proximal to high bluffs. We used the Soil Survey Geographic (SSURGO) Database to investigate changes in erodibility along the channel. Because bedrock exposure significantly limits erodibility, we investigated bedrock exposure using bedrock outcrop maps made available by the Minnesota Geological Survey (MGS, Hobbs, 2002; Hobbs, 2009), and by using a feature extraction tool to remotely map bedrock exposure using high-resolution air photos and LiDAR data.;Predictions based on remote data were compared with two datasets. Bank Erosion Hazard Index surveys, which are surveys designed to evaluate erosion susceptibility of banks, were collected along the three streams. In addition, a 500-year flood event during our field season gave us the opportunity to collect erosion data after a major event and validate our erosion hotspot predictions.;Regressions between predictors and field datasets indicate that the most significant variables are bedrock exposure, the stream power-based erosion index, and bluff proximity. A logistic model developed using the three successful predictors for Amity Creek watershed was largely unsuccessful. A threshold-based model including the three successful predictors (stream power-based erosion index, bluff proximity, and bedrock exposure) was 70% accurate for predicting erosion hotspots along Amity Creek. The limited predictive power of the models stemmed in part from differences in locations of erosion hotspots in a single large-scale flood event and long-term erosion hotspots. The inability to predict site-specific characteristics like large woody debris or vegetation patterns makes predicting erosion hotspots in a given event very difficult. A field dataset including long-term erosion data may improve the model significantly. This model also requires high resolution bedrock exposure data which may limit its application to other North Shore streams.