Associate Professor University of Nebraska-Lincoln
Disclosure: Disclosure information not submitted.
VIS-NIR-SWIR (Visible, Near Infrared, and Shortwave Infrared, 400 to 2500 nm) reflectance data have long been employed in remote sensing to estimate plant physiological and functional traits at leaf and canopy level. This technology can potentially be used for low-cost high-throughput analysis of leaf physiological and chemical traits. In this talk I will present the work and some findings we made in this line of research. Leaf-level VIS-NIR-SWIR reflectance data were collected from multiple crop species including maize, sorghum, and soybean; and a wide array of leaf properties were studied (including water content, chlorophyll content, nutrients, and cell wall compositions). Different modeling approaches including spectral indices, Partial Least Squares Regression, and 1-dimensional Convolutional Neural Network were compared to estimate the leaf traits from VIS-NIR-SWIR data. The influence of crop species, genotypes, and environments on VIS-NIR-SWIR model performance was also investigated. Finally, challenges and implications to scale up this technology to wider phenotyping user communities will be discussed.