Estimating yield-contributing physiological parameters of cotton using UAV-based imagery.

Area of Science:

• Agricultural Science
• Remote Sensing
• Crop Physiology

Background:

• Cotton lint yield is determined by light interception, radiation use efficiency, and harvest index.
• Traditional measurement methods for these parameters are labor-intensive, time-consuming, and destructive.
• There is a need for efficient, non-destructive techniques to monitor cotton's physiological status.

Purpose of the Study:

• To estimate fraction of photosynthetically active radiation intercepted by the canopy (IPARf), radiation use efficiency (RUE), and biomass using UAV-based imagery.
• To estimate lint yield via the cotton fiber index (CFI) derived from UAV imagery.
• To assess the potential of biomass and lint yield models for estimating cotton harvest index (HI).

Main Methods:

• Conducted field experiments with varying nitrogen treatments over two growing seasons.
• Collected bi-weekly UAV-based RGB and multispectral imagery, alongside ground measurements of light interception and biomass.
• Computed 20 vegetation indices (VIs) and developed generalized linear regression models using VIs and growing degree days (GDDs).

Main Results:

• IPARf models achieved R² values from 0.66 to 0.90, with RVI and RECI explaining 93% of variation.
• Cotton above-ground biomass was best estimated using MSAVI and OSAVI indices.
• Radiation use efficiency (RUE) prediction models explained 84% of variation; Cotton Fiber Index (CFI) showed a strong relationship (R² = 0.69) with machine-harvested lint yield.

Conclusions:

• UAV-based remote sensing provides accurate estimations for IPARf and biomass, crucial for cotton yield prediction.
• The developed models show potential for estimating cotton harvest index, though further refinement for lint yield is suggested.
• This study pioneers the use of UAV imagery for predicting cotton's functional yield drivers, paving the way for improved agricultural research and management.