Validating UAV-based Crop Architecture Traits with Non-Destructive Ground Truth Data

Remote sensing technologies have transformed agricultural monitoring practices by enabling rapid evaluation of agronomically-important traits across spatial and temporal scales. These remote sensing approaches offer substantial improvements in efficiency, requiring significantly less time and labor compared to traditional manual measurements. Unoccupied aerial vehicles (UAVs) is a remote sensing tool rising in popularity within the field of phenomics because of its desirable input-to-output ratio. However, UAV derived measurements must be validated against ground truth data to ensure reliability in phenotypic trait estimation. This study tests the accuracy of UAV derived estimates of leaf area index (LAI) and plant height against ground-truth measurements. The Genomes to Fields (G2F) initiative provide maize hybrids, at a site in Aurora, NY. LAI and plant height were the two key phenotypic traits extracted from multispectral images. Plant height, from digital surface models (DSM) constructed from georeferenced orthomosaics. LAI, from transformed normalized difference vegetation index values (NDVI). Regression analyses compare aerial and ground observations with anticipated coefficients of determination (R²) of ≥0.70 for both traits. Time and operational costs for each method were documented to quantify throughput gains. It is anticipated that UAV-based trait extraction will prove more efficient. A functional and reproducible pipeline for UAV-based trait analysis was also established. By validating UAV derived estimates of canopy architecture traits such as LAI and plant height, this study provides reliable ground truth data that supports their broader integration into high throughput phenotyping pipelines and process-based crop growth models, ultimately expanding trait collection and improving the accuracy of yield predictions within G2F.

Participating in the REU program at the Boyce Thompson Institute was a pivotal experience in my development as a researcher. My primary goal was to deepen my research skills and make a more substantial intellectual contribution to my work. Under the mentorship of Dr. Luke Gregory and guidance of Dr. Michael Gore, I was able to achieve that and more. Prior to this experience, my understanding of remote sensing and phenomics was present but fragmented. Through this program, I was able to integrate these concepts into a cohesive framework, strengthening both my technical proficiency and conceptual grasp of the field. This experience has significantly boosted my confidence in my readiness for graduate school and affirmed my commitment to a research-driven career. Beyond the lab, I valued becoming part of the plant science community in Ithaca. The connections I made during my time here have been both meaningful and inspiring. This fall, I plan to apply to Ph.D. programs, with the goal of continuing research and contributing to the advancement of plant science.