Deriving Temporal Height Information for Maize Breeding

Phenotypic data such as height provide useful information to crop breeders to better understand their field experiments and associated field variability. However, the measurement of crop height in many breeding programs is done manually which demands significant effort and time and does not scale well when large field experiments are involved. Through structure from motion (SfM) techniques, small unmanned aerial vehicles (sUAV) or drones offer tremendous potential for generating crop height data and other morphological data such as canopy area and biomass in cost-effective and efficient way. We present results of an on-going UAV application project aimed at generating temporal height metrics for maize breeding at the Texas A&M AgriLife Research farm in Burleson County, Texas. We outline the activities involved from the drone aerial surveys, image processing and generation of crop height metrics. The experimental period ran from April (planting) through August (harvest) 2016 and involved 36 maize hybrids replicated over 288 plots ( 1.7 Ha). During the time, crop heights were manually measured per plot at weekly intervals. Corresponding aerial flights were carried out using a DJI Phantom 3 Professional UAV at each interval and images captured processed into point clouds and image mosaics using Pix4D (Pix4D SA; Lausanne, Switzerland) software. LiDAR data was also captured at two intervals (05/06 and 07/29) to provide another source of height information. To obtain height data per plot from SfM point clouds and LiDAR data, percentile height metrics were then generated using FUSION software. Results of the comparison between SfM and field measurement height show high correlation (R2 > 0.7), showing that use of sUAV can replace laborious manual height measurement and enhance plant breeding programs. Similar results were also obtained from the comparison of SfM and LiDAR heights. Outputs of this project are helping plant breeders at Texas A&M automate routine height measurements in maize and quickly make actionable decisions and discover new hybrids.