Estimating Defoliation of Two Distinct Cotton Types Using Reflectance Data

Cotton defoliation is an important part of cotton harvest preparation. Visual estimates of cotton defoliation allow producers to monitor harvest readiness and make further defoliation decisions as necessary, but visual estimates are subjective and may differ from one reviewer to the next. In this manuscript, a spectrometric method for quantifying cotton defoliation is proposed. In 2003 and 2004, leaf area index (LAI) was monitored in multiple environments on 0.91 m sections of row to quantify percentage defoliation. Reflectance over each plot was measured using a narrow-band spectrometer, and normalized difference vegetation index (NDVI) models composed of reflectance at all wavelengths were regressed against LAI to determine which wavelengths most accurately estimated changes in LAI. Both linear and quadratic models were tested for their usefulness in estimating LAI. Quadratic models more accurately estimated LAI in the red spectral region than linear models, but reached a maximum LAI value of about 1.5. Therefore, the quadratic models were of limited usefulness. At the red edge (about 705 nm to 720 nm), the quadratic and linear models had similar coefficients of determination, which were higher than those derived from linear models in other wavelengths. These results suggest that reflectance indices based on red edge measurements can offer accurate, consistent defoliation estimates, and could potentially increase defoliation efficiency and decrease costs.