Pairing soil sampling with very-high resolution UAV imagery: An examination of drivers of soil and nutrient movement and agricultural productivity in southern Ontario

• Published in: Geoderma 2020-12, Vol.379, p.114630, Article 114630
• Main Authors: Menzies Pluer, E.G., Robinson, D.T., Meinen, B.U., Macrae, M.L.
• Format: Article
• Language: English
• ISSN: 0016-7061; 1872-6259
• DOI: 10.1016/j.geoderma.2020.114630

•UAV-based digital terrain model is used to map topographic variation and erosion flow pathways.•Water content and organic matter are primary controlling factors on crop yield.•Soil and nutrients move in erosional pathways in the lower portion of catch basins.•Novel opportunities for integrating UAV data and soil sampling for time-series analysis. Soil erosion from agricultural lands continues to be a global societal problem. The movement of soils is often accompanied by nitrogen and phosphorus that are crucial to crop growth, but their redistribution from farm fields to waterways can reduce crop yields and degrade water quality. While within-field sediment and nutrient movement has been quantified using small plots and edge-of-field monitoring, these approaches fail to capture their spatial distribution. The pairing of soil sampling with unmanned aerial vehicle (UAV) data offers a novel and low-cost approach to map the spatial distribution of soil characteristics and nutrient concentrations within a farm field. UAV data are used to generate a digital terrain model and subsequently map within-field topographic variation and erosional flow pathways. Topographic variation is discretized into landform elements (flat, shoulder, backslope, footslope) that capture within-field heterogeneity and have potential for scaling out soil sampling to larger spatial extents. Our results show the controlling factor of water content and organic matter on crop yield, as represented by normalized difference vegetation index (NDVI). Significant differences in water content and organic matter were found across landform elements with increases in both parameters downslope. Upslope landform elements contained more sand content (9–20%) and had lower NDVI values than downslope elements. Complementing these findings, significant differences in organic matter, soluble nitrogen, and soluble reactive phosphorus occurred along erosional flow pathways. Our within-field results have implications for farmers, as our analysis of soil characteristics indicated that NDVI was positively correlated with water content (0.05), organic matter (0.15), silt (0.36), and clay (0.17) content and negatively correlated with soluble nitrogen (−0.47) and phosphorus (−0.30) concentrations. In addition to discussing the challenges and opportunities for expanding upon the presented research, we use a simple proof-of-concept hydrological model to demonstrate the potential role of hydrological connectivity and