Airborne Spectral Indices Applied As a Breeding Selection Tool.

Airborne Spectral Indices Applied as a Breeding Selection Tool.

Maria Tattaris, Matthew Reynolds.

Remote sensing measurements offer a non-intrusive approach to obtain information relating to physiological processes and the status of plants. Rapid advances in genotyping are increasingly observed, however phenotyping techniques are lagging behind. Low level aerial remote sensing can overcome this bottleneck; providing efficient, high-throughput data collection over large areas whilst maintaining enough spatial resolution to distinguish between experimental wheat plots.

Plant physiological traits, particularly NDVI and canopy temperature (CT) have been shown to have a strong association with yield and biomass. This work investigates the potential use of NDVI and CT derived from airborne imagery to be used as a selection tool for breeders. Specifically, the focus is on time series analysis of these airborne measurements against yield and biomass, in order to determine the optimum phenological stage for airborne data collection for yield and biomass prediction, under stressed and non-stressed environments.

Sampling was performed with an eight rotor unmanned aerial vehicle (UAV). Instruments mounted on the UAV alternate between a three channel multispectral imaging spectrometer and a thermal camera. Multiple flights were conducted during the 2013 cycle over experimental wheat trials in Northern Mexico under three different environments; optimal irrigated, drought stress and hot-irrigated. Measurements were made at different stages of the growth cycle for each environment. Thermal and multispectral aerial images were corrected and processed to determine a data point for each plot within trials and subsequently used to calculate NDVI and a thermal index, relating to CT, at plot level.

The relationship between the airborne derived thermal and NDVI indices and yield and biomass was investigated at different measurement times within the plant growth cycle. Airborne indices were also validated by equivalent indices collected at ground level. Correlations between airborne data and yield/biomass at plot level proved to be similar or even better to the equivalent correlations using data collected from instruments on the ground. Results give confidence to the application of such airborne remote sensing techniques for high throughput phenotyping, in particular the ability to evaluate the level of stress and performance of thousands of genetic resources under extreme heat and drought conditions.