Soil Surface Micro-Topography by Structure-from-Motion Photogrammetry for Monitoring Density and Erosion Dynamics

Soil erosion is a major threat to soil fertility, food security and water resources. Besides a quantitative assessment of soil loss, the dynamics of erosion-affected arable soil surfaces still poses challenges regarding field methods and predictions because of scale-dependent and soil management-related complex soil-crop-atmosphere processes. The objective was to test a photogrammetric Structure-from-Motion (SfM) technique for the mm-scale mapping of the soil surface micro-topography that allows the monitoring without special equipment and with widely available cameras. The test was carried out in May 2018 on three plots of 1.5 m(2) (upper-, middle-, and footslope) covering surface structural features (tractor wheel lane, seed rows) along a Maize-cultivated hillslope with a coarse-textured topsoil and a runoff monitoring station. The changes in mm-scaled surface micro-topography were derived from repeatedly photographed images of the same surface area during a 2-weeks period with two rain events. A freely available SfM-program (VisualSfM) and the QGIS software were used to generate 3D-models of the surface topography. Soil cores (100 cm(3)) were sampled to gravimetrically determine the topsoil bulk density. The micro-topographical changes resulting from rainfall-induced soil mass redistribution within the plots were determined from the differences in SfM maps before and after rain. The largest decrease in mean soil surface elevation and roughness was observed after rain for the middle slope plot and primarily in initially less compacted regions. The spatially-distributed intra-plot changes in soil mass at the mm-scale derived from the digital micro-topography models indicated that local depressions were filled with sediments from surrounding knolls during rainfall. The estimated mass loss determined with the SfM technique decreased, if core sample-based soil settlement was considered. The effect of changes in the soil bulk density could be described after calibration also with an empirical model suggested in the Root-Zone-Water-Quality-Model. Uncertainties in the presented plot-scale SfM-technique were due to geo-referencing and the numerical limitations in the freely available SfM-software. The photogrammetric technique provided valuable information on soil surface structure parameters such as surface roughness. The successful application of SfM with widely available cameras and freely available software might stimulate the monitoring of erosion in regions with limited accessibility.