Examining Ground and Aerial Remote Sensing Measurements for NUE and Variety Selection in Wheat.

One way to reduce production cost with Nitrogen (N) fertilization in wheat (Triticum aestivum L.) is to improve the nitrogen use efficiency (NUE) of the cultivars themselves. However, the selection tools currently available to breeding programs are often slow and inefficient. The objective of this research was to determine if aerial measurements, which can be taken in a matter of minutes, can be used instead of time-consuming ground measurements of normalized difference vegetation index (NDVI) and canopy temperature for NUE estimation. Twelve wheat varieties were examined at two locations in Virginia, the Tidewater AREC in Suffolk and the Eastern Virginia AREC in Warsaw in 2015-16 and 2016-17. Varieties were subjected to two N rates corresponding to a low spring application of 67 kg N ha^-1 and a recommended normal rate of 134 kg N ha^-1.

Measurements were taken at several growth stages with both ground and aerial sensors with an unmanned aerial vehicle (UAV). Ground taken measurements included NDVI, red-green-blue (RGB) images leaf area index, and canopy temperature depression. The UAV measurements were taken with three sensors individually mounted on the UAV: a RGB digital camera, a near-infrared (NIR) camera, and an infrared (IR) camera. Aerial images taken with the RGB camera were used to compute color space characteristics such as hue angle, intensity, saturation, and RGB-derived vegetation indices Green Area (GA) and Greener Area (GGA). The images from the NIR camera were used to derive aerial NDVI, while the IR images were used to obtain canopy temperature.

At each growth stage, ground and aerial measurements were compared with each other. Measurements were also correlated with final grain yield. Measurements taken with the UAV were more time effective than measurements taken from the ground. Both the GA and GGA were correlated (R² = 0.7) with the NDVI taken from the ground, and the IR camera was successful in discriminating varieties with high and low NUE. Aerial RGB indices (R²=0.72) were also a better predictor of yield than ground-collected NDVI (R²=0.72).