Assessing field-scale rill erosion mitigation by cover crops in arable land using drone image analysis

Abstract

Cover crops (CC) effectively reduce soil erosion, a significant threat to farmers and the environment. Yet, there is lack of data quantifying their effect on rill erosion in the field scale. The major objective of this study was to use UAV-RGB images to estimate the effects of CC on rill erosion in the field scale and to characterize rill parameters in areas with and without CC. Images were collected from a 20-ha field in the "Model Farm for Sustainable Agriculture", consisting of plots with and without CC. Images were captured 33 days after CC sowing and following substantial rainfall events that formed three prominent rills. Following the elimination of vegetation pixels, structure from motion algorithm was used to generate a post-erosion digital surface model (DSM) and a baseline DSM simulating the pre-erosion soil surface (DSM _{reconstructed baseline)}. Change-detection analysis revealed that CC significantly reduced rill erosion. Average soil loss per m² was 48 %, 58 %, and 29 % lower in CC compared to bare soil plots in the three studied rills. Additionally, rill maximum depth was 74 %, 74 %, and 24 %, and cross-sectional surface area was 67 %, 87 %, and 43 % lower in CC, compared to bare soil plots. The findings highlight CC's effectiveness in mitigating field-scale rill erosion even in their early growth stages. However, creating a DSM _{reconstructed baseline} in CC plots is currently confined to partial CC vegetation coverage (leaving enough soil pixels visible), necessitating additional studies to determine the maximal coverage that won't compromise accuracy. Further assessments of the methods' quantitative accuracy require studies incorporating extensive ground truth data.