Scalable growth models for time‐series multispectral images

Vegetation indices (VIs) are produced as a combination of different reflectance bands that captured by multispectral images (MSIs). These indices, such normalized difference vegetation index (NDVI), reported to be proxy indicators photosynthetic activity, plant canopy biomass, and leaf area index. To determine the utility using VI derived from MSI model growth, random regression (RR) models with linear splines orders Legendre polynomials were applied data collected (years 2019 2020) part Genome-to-Fields initiative. Growth curves maize (Zea mays L.) hybrids modeled both NDVI cumulative (cNDVI) phenotypes. Due in recording dates, sparse overlap between years, all analyses nested within year. Results indicate RR provide robust scalable method for modeling growth phenotypes extracted MSI; however, showed inconsistent convergence. estimated cNDVI low-to-moderate heritability (0.11–0.44) range genetic correlations (−0.15 0.97) grain yield. This study demonstrates trends, best results obtained when cNDVI.