The use of UAS-based high throughput phenotyping (HTP) to assess sugarcane yield

Abstract

A sensor mounted on an unmanned aerial system (UAS) may enable breeding programs’ selection efficiency. The objectives of this study were to investigate the use of UAS with red, green, and blue (RGB) camera to assess sugarcane yield and to investigate the direct and indirect influences of canopy features on cane yield of sugarcane. A trial was conducted from 2019 to 2021 at the Texas A&M AgriLife Research and Extension Center in Weslaco, Texas, and arranged in a complete randomized block design, consisting of 7 genotypes with 4 replications. Seven UAS image acquisitions were performed at the plant cane stage using an RGB sensor. At the first ratoon stage, five flights were conducted using RGB and multispectral sensors. Ground measurements were obtained, including pol, tons of sugar per hectare (TSH), and tons of cane per hectare (TCH). The results showed that the correlation of canopy features with pol at the elongation phase was poor, compared to those acquired at the maturity phase. Likewise, the most significant correlations between canopy features with TSH and TCH were found in the late elongation to maturity phase. In addition, a good agreement of the validation dataset between the observed and predicted pol (R2 0.58, RMSE 0.77%), TSH (R2 0.68, RMSE 2.24 Mg ha−1), and TCH (R2 0.66, RMSE 13.64 Mg ha−1) was observed. The canopy height model (CHM) and Normalized Difference Vegetation Index (NDVI) had the highest direct effect on TCH. Therefore, assessing sugarcane yield using UAS-based imagery looks promising for High Throughput Phenotyping (HTP) in sugarcane breeding programs.

Keywords

• Predictive model
• Remote sensing
• Vegetation indices
• Path coefficient