MULTI-Scale Analysis of FIELD-Grown Sorghum Using RGB, Lidar, and Hyperspectral Imaging

Bottlenecks in our ability to collect accurate, high-resolution, phenotype data on crops like sorghum limit how efficiently we can combine plant characteristics with genomic information in developing superior cultivars. Most field crop phenotyping studies for aboveground traits occurs at the plant and canopy scale of biological organization. Phenotypes are therefore typically measured, analyzed, and/or expressed at the row, plot, strip, management zone, and/or field level. We are developing remote sensing platforms that can be used for high-throughput field phenotyping to enhance the efficiency of crop improvement. New sensors and sensor platforms, novel georeferencing techniques, and sophisticated image and data analysis methods (e.g., feature extraction, image segmentation) are being implemented to quantify variation in plant- and plot-level traits. These measurements provide insights into research plot and field quality, field equipment performance, genotype stress tolerance, physiological plasticity, and spatial variability. Such information contributes to field crop breeding and management, precision agriculture, and equipment manufacturing communities. Challenges from phenotyping at the individual plant level using traditional and emerging techniques will be discussed. Agronomic performance and remote sensing data matched with genotypic data enables trait dissection and optimization of sorghum for biomass and energy yield for transportation fuel.

Christopher Boomsma¹, Javier Ribera¹, Yuhao Chen¹, Fangning He¹, Weifeng Xiong¹, Zhou Zhang¹, Addie Thompson¹, Karthikeyan Natesan Ramamurthy², Aurelie Lozano², Naoki Abe², Ayman Habib¹, Edward Delp¹, Larry Biehl¹, Michael Leasure¹, Keith Cherkauer¹, Clifford Weil¹, and Melba Crawford^{1 1}Purdue University, West Lafayette, IN

²IBM Research, Yorktown Heights, NY