Estimating Decadal Soil Erosion Rates in Minnesota's Agricultural Landscapes Based on Digital Terrain Attributes and 137Cs Profile Data.

In order to develop landscape-scale estimates of soil erosion in Minnesota’s agricultural landscapes, we conducted a broad survey study of ¹³⁷Cs in cultivated fields and uncultivated grassland reference sites located across the southern third of Minnesota. A ¹³⁷Cs conversion model was used to determine soil erosion rates for 107 locations in cultivated sites. Measured soil erosion rates ranged from 49 t ha^-1 yr^-1 (erosion) to -74 t ha^-1 yr^-1 (deposition). Based on these measured rates, regression models were developed with the goal of broadly predicting soil erosion rates based on topographic characteristics. Digital terrain attributes were calculated from LiDAR-derived (Light Detection And Ranging) digital elevation models and then used as predictor terms in regression model development. Resulting models showed that: (1) profile curvature, (2) planform curvature, and (3) slope steepness were significant model terms in predicting erosion rates for different Minnesota Major Land Resource Areas (MLRAs). When applied to cultivated landscapes, the regression models create maps of predicted long-term rates of soil erosion or deposition.