Multi-temporal UAV data for automatic measurement of rill and interrill erosion on loess soil

• Published in: Earth surface processes and landforms 2015-05, Vol.40 (6), p.741-755
• Main Authors: Eltner, Anette, Baumgart, Philipp, Maas, Hans-Gerd, Faust, Dominik
• Format: Article
• Language: English
• Subjects: automatic rill extraction; Erosion; Erosion mechanisms; field scale; Mathematical models; rill-interrill; Seasons; Soil (material); Soil erosion; UAV; Unmanned aerial vehicles; Widening
• ISSN: 0197-9337; 1096-9837
• DOI: 10.1002/esp.3673

The fragile landscape of the north European loess belt is prone to soil erosion due to soil properties and intense land use of the fertile region. Exact measurement of surface changes with high temporal and spatial resolution over large areas is necessary to quantify and understand rill and interrill erosion processes. High resolution aerial imagery, acquired by an unmanned aerial vehicle (UAV), is used to automatically generate precise digital surface models (DSMs) of high spatial resolution by applying structure‐from‐motion image processing tools. During an investigation period of ten months, a 600 m2 field plot is observed during four field campaigns. A stable reference system is established for multi‐temporal comparison. The overall accuracy of the DSMs generated from UAV images is less than 1 cm, verified by comparison with terrestrial laser scanner (TLS) data. Furthermore, a method for automatic rill extraction and rill parameter calculation is developed, which enables objective rill description with cm‐accuracy and ‐resolution. Soil surface roughness and rill development as well as volumetric quantifications are analysed for multi‐temporal change detection. Surface changes during winter season are controlled by soil consolidation, crusting and sheet erosion. During rainy spring season sheet erosion and rill incision occur. Two thunderstorms in summer season cause dominant rill erosion. Erosion rills are more dominantly deepening than widening (from to 2 to 4 cm depth and from 17 to 23 cm width), resulting in average per rill erosion values of 0.03 and 0.07 m3 respectively. An orientation dependent lateral rill shift is revealed, implying rill widening in eastern direction due to dominant winds from the West. Volumetric quantifications indicate high erosion volumes, reaching up to 121 tha‐1 during the summer events. Highest erosion volumes are due to rill erosion rather than interrill erosion. Copyright © 2014 John Wiley & Sons, Ltd.