Scattering Analysis and Ultra-Wideband Radar for High-Throughput Phenotyping of Wheat Canopies

Abstract

Rising the yield of wheat crops is essential to meet the projected demands of future consumption and it is expected that most yield increases will be associated to improvements in biomass accumulation. Cultivars with canopy architectures that focus the light interception where photosynthetic-capacity is greater achieve larger biomass accumulation rates. Identifying varieties with improved traits could be performed with modern breeding methods, such as genomic-selection, which depend on genotype-phenotype mappings. Developing a non-destructive sensor with the capability of efficiently phenotyping wheat-canopy architecture parameters, such as height and vertical distribution of projected-leaf-area-density, would be useful for developing architecture-related genotype-phenotype maps of wheat cultivars. In this dissertation, new scattering analysis tools and a new 2-18 GHz radar system are presented for efficiently phenotyping the architecture of wheat canopies. The radar system presented was designed with the objective to measure the RCS profile of wheat canopies at close range. The frequency range (2-18 GHz), topology (Frequency-modulated-continuous-wave) and other radar parameters were chosen to meet that goal. Phase noise of self-interference signals is the main source of coherent and incoherent noise in FMCW radars. A new comprehensive noise analysis is presented, which predicts the power-spectral-density of the noise at the output of FMCW radars, including those related to phase noise. The new 2-18 GHz chirp generator is based on a phase-locked-loop that was designed with large loop bandwidth to suppress the phase noise of the chirp. Additionally, the radar RF front-end was designed to achieve low levels of LO-leakage and antenna feed-through, which are the main self-interference signals of FMCW radars. In addition to the radar system, a new efficient radar simulator was developed to predict the RCS waveforms collected from wheat canopies over the 2-18 GHz frequency range. The coherent radar simulator is based on novel geometric and fully-polarimetric scattering models of wheat canopies. The scattering models of wheat canopies, leaves with arbitrary orientation and curvature, stems and heads were validated using a full-wave commercial simulator and measurements. The radar simulator was used to derive retrieval algorithms of canopy height and projected-leaf-area-density from RCS profiles, which were tested with field-collected measurements. The retrieved heights and projected-leaf-area densities compare well against ruler measurements and image-based retrievals, respectively.