Comparison of Modelling Strategies to Estimate Phenotypic Values from an Unmanned Aerial Vehicle with Spectral and Temporal Vegetation Indexes

Aboveground dry weight (AGDW) and leaf area index (LAI) are indicators of crop growth status grain yield as affected by interactions genotype, environment, management. Unmanned aerial vehicle (UAV) based remote sensing provides cost-effective non-destructive methods for the high-throughput phenotyping traits (e.g., AGDW LAI) through integration UAV-derived vegetation indexes (VIs) with statistical models. However, effects different modelling strategies that use dataset compositions explanatory variables (i.e., combinations sources temporal VI datasets) on estimates LAI have rarely been evaluated. In this study, we evaluated three VIs (visible, spectral, combined) types datasets (mono-, multi-, full-temporal) LAI. The were derived from visible (RGB) multi-spectral imageries, which acquired a UAV-based platform over wheat trial at five sampling dates before flowering. Partial least squares regression models built to estimate each prediction date. results showed mono-temporal obtained similar performances estimating (RRMSE = 11.86% 15.80% visible, 10.25% 16.70% combined VIs) 13.30% 22.56% 12.04% 22.85% 13.45% across dates. Mono-temporal outperformed other two in general. Models generally better than multi- full-temporal VIs. suggested can be an alternative in-season combination used self-calibration method demonstrated potential normally less 15%) breeding or agronomy trials.