High Throughput Field Based Phenomics activities are conducted to collect or estimate physical measurements on plants across different treatments or cultivars in an automated way; typically using ground vehicles, unmanned aerial vehicles / systems (UAV, UAS, i.e. drones) or satellites coupled with remote sensing technologies. Field phenomics currently integrates many disparate disciplines in sensor development, data collection, handling and analysis (remote sensing, engineering, computer science, statistics, physics, etc.) with researchers interested in using the acquired data to understand the underlying biology (plant biology, plant physiology, etc.) and/or to make practical gains in yield, profit or sustainability in farmers fields (agronomy, plant breeding, etc.). Since phenomics science is where genomics was 30 years ago, a similar explosion of discoveries in phenomics can be expected with the right investment of knowledge and resources. While field phenomics applications to plant breeding are exciting they will require challenging transdisciplinary work to reach potential.

In Texas A&M maize (Zea mays, i.e. corn) breeding program target environments, plant height manually measured using a ruler is moderately correlated to grain yield (R² = 0.61), suggesting the plant height phenotype can be used as a proxy to select for yield. UAS estimates of plant height are highly correlated to manually measured plant height throughout growth (R² average ~0.80 ranging from 0.40 to 0.87 by day); importantly, objective variance component metrics show overall accuracy is similar between UAS and manually measured plant height but this can vary greatly depending on the flight. Currently we are using UAS height estimates and NDVI to generate temporal growth curves and test growth curves ability to predict grain yield. In the future we will be go beyond just replicating manual measurements to rapidly extract and test many image features (which may have no relation to any known or existing manual measurement) to find new predictors for grain yield. Moving research plot equipment to measure grain yield across the field locations is costly and dangerous. If yield can be accurately predicted through remote sensing this would allow us to dramatically increase the genotypes and environments used for screening, resulting in varieties with commensurate increases in farmers grain yield.