Monitoring winter wheat growth at different heights using aerial imagery

Abstract

AbstractDrones (unmanned aerial vehicles) provide another system to mount sensors for measuring plant characteristics. For winter wheat (Triticum aestivum) this can include evaluating stands and making nitrogen (N) recommendations. Timing these flights and adequate camera resolution (based on flying height), must be known before applying tasks. This study observed six winter wheat planting populations (222, 297, 371, 445, 494, and 544 seeds m–2) at three different heights above ground level (30, 60, and 120 m) over three growing seasons. Plant populations could be separated at all growth stages and heights flown but were easier to separate right after emergence (GS11). In the spring, additional tillering caused the higher populations (371–544 seeds m–2) to have similar normalized difference vegetative index (NDVI), much like the final yields. Comparing changes in NDVI between flights was also successful in separating high and low planting populations, with inverse relationships in the fall and spring. A random forest classification of tiller counts by NDVI measurements ranked change in NDVI between stages as the most important compared to single flights. As separation and classification was successful at the lowest camera resolution (120 m), it can be possible for scouts to collect whole field imagery for analyses prior to deciding on split N applications.