Moisture content estimation and senescence phenotyping of novel <i>Miscanthus</i> hybrids combining UAV‐based remote sensing and machine learning

Miscanthus is a leading perennial biomass crop that can produce high yields on marginal lands. Moisture content highly relevant quality trait with multiple impacts efficiencies of harvest, transport, and storage. The dynamics moisture during senescence overwinter ripening are determined by genotype × environment interactions. In this paper, unmanned aerial vehicle (UAV)-based remote sensing was used for high-throughput plant phenotyping (HTPP) the autumn winter 14 contrasting hybrid types (progeny M. sinensis x [M. sin sin, eight types] sacchariflorus sac, six types]). time series estimated using machine learning (ML) models range vegetation indices (VIs) derived from UAV-based sensing. most important VIs estimation were selected recursive feature elimination (RFE) algorithm BNDVI, GDVI, PSRI. ML model transferability only when above 30%. best accuracy achieved combining categorical variables (5.6% RMSE). This identifying stay-green (SG) hybrids generalized additive (GAM). Combining GAM modeling, applied to values UAV flights, proved be powerful tool HTPP.