Improving Biomass and Grain Yield Prediction of Wheat Genotypes on Sodic Soil Using Integrated High-Resolution Multispectral, Hyperspectral, 3D Point Cloud, and Machine Learning Techniques

Sodic soils adversely affect crop production over extensive areas of rain-fed cropping worldwide, with particularly large in Australia. Crop phenotyping may assist identifying cultivars tolerant to soil sodicity. However, studies identify the most appropriate traits and reliable tools on sodic are limited. Hence, this study evaluated ability multispectral, hyperspectral, 3D point cloud, machine learning techniques improve estimation biomass grain yield wheat genotypes grown a moderately (MS) highly (HS) sites northeastern While number have reported using different remote sensing approaches quantify growth, stress, variation, limited combination these including genotypic yield, especially constrained environments. At close flowering, unmanned aerial vehicle (UAV) ground-based proximal was used obtain and/or data, while heights were also manually measured field. Grain machine-harvested at maturity. UAV sensing-derived spectral vegetation indices (VIs), such as normalized difference index, optimized adjusted enhanced index height closely corresponded yields. multispectral VIs more associated yields compared data. The red-green-blue (RGB) cloud technique effective determining height, which slightly better correlated than ground-measured These (VIs height) further simulated algorithms (multitarget linear regression, support vector Gaussian process artificial neural network) kernels yield. network predicted (R2 = 0.89; RMSE 34.8 g/m2 for MS R2 0.82; 26.4 HS site) 0.88; 11.8 0.74; 16.1 less error others. Wheat Mitch, Corack, Mace, Trojan, Lancer, Bremer identified constraints Emu Rock, Janz, Flanker, Gladius. improves our select accurate soils. This will farmers constraints.