Growth analysis of cotton using UAS derived multi temporal canopy features.

Unoccupied aerial systems (UAS) are becoming a regular tool in agriculture to obtain phenotypic information of plant growth and development. In this study, we collected red, green, and blue (RGB) images using UAS multiple times throughout the growing season from a cotton field experiment conducted in 2016 and 2021. Collected images were processed to obtain digital surface models (DSMs) from which canopy height (CH) measurements were extracted. Crop growth curve was obtained by fitting several non-linear growth functions on the multi-temporal CH measurements. The five-parameter logistic function performed best with highest R2 (0.98) and lowest RMSE (6.41). The first and second order derivative of the five-parameter logistic function was performed to obtain several canopy growth parameters. These parameters were used to evaluate the maturity of cotton genotypes and correlated with yield. The maximum growth rate was correlated with yield (R2 = 0.46 in 2016 and R2 = 0.68 in 2021). Additionally, the time of onset of steady phase was used to rate maturity of the genotypes with 80% accuracy. This study demonstrated an approach to summarize high-resolution multi-temporal data obtained by UAS to better understand crop growth and development with a potential to be used for assessing the maturity of the genotypes, yield estimations, and management decisions of plant growth regulators.