<i>Cotton Yield Estimation based on Plant Height From UAV-based Imagery Data </i>

Abstract

Abstract. Accurate estimation of crop yield before harvest, especially in early growth stages, is important for farmers and researchers to optimize field management and evaluate crop performance. However, conventional methods of using ground sensing to estimate crop yield are not efficient. The goal of this research was to evaluate the potential of using a UAV-based remote sensing system with a low-cost RGB camera to estimate yield of cotton within season. The UAV system took images at 50 m above ground level over a cotton field at the growth stage of first flower. Waypoints and flight speed were selected to allow > 70% image overlap in both forward and side directions. Images were processed to develop a geo-referenced orthomosaic image and a digital elevation model (DEM) of the field, which was then used to map plant height by calculating the difference in elevation between the crop canopy and the bare soil surface. Twelve ground control points (calibration objects) with known GPS coordinates and height were deployed in the field and were used as check points for geo-referencing and height calibration. Geo-referenced yield data were registered with the plant height map row-by-row. Correlation analysis between yield and plant height was conducted row-by-row with row registration and without row registration respectively. Pearson correlation coefficients between yield and plant height for all individual rows were in the range of 66% to 96%, higher than those without row registration (54% to 95%). A non-parametric regression used for building a yield estimation model based on image-derived plant height was able to estimate yield with less than 10% error (root mean square error of 360.4 kg ha-1 and mean absolute error of 180.9 kg ha-1). The results indicated that the UAV-based remote sensing system equipped with a low-cost digital camera was able to estimate cotton yield with acceptable errors.