Feasibility study of directly georeferenced images from low-cost unmanned aerial vehicles for monitoring sward height in a long-term experiment on grassland

摘要

Monitoring sward height is an important step in the assessment of overall grassland quality and quantity. A common approach involves measuring compressed sward height with a rising plate meter (RPM), which has a limited spatial coverage. The concept of multi-temporal Crop Surface Models (CSM) based on unmanned aerial vehicles (UAV)-imaging offers non-invasive near real-time analysis of vegetation with high spatial and temporal resolution. CSMs provide similar information as RPM measurements, but require precisely measured ground control. The rapidly evolving technology of UAVs, however, provides low-cost systems with direct georeferencing of images during flight with a moderate accuracy. Therefore in this study we evaluated the accuracy of sward height derived from directly georeferenced images of a low-cost UAV in an experimental grassland field in Germany. The good correlation of RPM based sward height and UAV-based sward height (R2 = 0.86) indicate that an automated monitoring approach for swardheight is possible.