Validation of UAV-based alfalfa biomass predictability using photogrammetry with fully automatic plot segmentation.

Area of Science:

• Agronomy
• Plant Breeding
• Remote Sensing

Background:

• Alfalfa is a globally significant forage legume, crucial for agriculture.
• Genetic improvements have focused on hardiness and disease resistance, but biomass traits remain challenging.
• Labor-intensive phenotyping hinders efficient biomass selection in alfalfa breeding.

Purpose of the Study:

• To develop and validate a high-throughput phenotyping method for alfalfa biomass selection using UAV imagery.
• To overcome the bottleneck of manual phenotyping in alfalfa breeding programs.

Main Methods:

• Utilized two alfalfa fields for model development and validation.
• Employed unmanned aerial vehicle (UAV) imagery with multispectral cameras for data acquisition.
• Developed an automated field plot segmentation method using GRID software and Normalized Difference Vegetation Index (NDVI).
• Extracted vegetative area and other spectral indices (Normalized Green-Red Difference Index, Normalized Difference Red Edge Index) for biomass prediction.

Main Results:

• The prediction model, using four UAV image features, explained 50-70% of biomass variation in the validation field.
• Automated segmentation and feature extraction from UAV data enabled high-throughput phenotyping.
• Vegetative area, plant height, and specific vegetation indices were key predictors of alfalfa biomass.

Conclusions:

• UAV-based high-throughput phenotyping offers a viable solution to accelerate biomass selection in alfalfa.
• This approach can significantly improve the efficiency and reduce the cost of alfalfa breeding programs.
• Integration of remote sensing technologies is crucial for advancing plant breeding.