Tracking Spatiotemporal Parameters for in-Season Nitrogen Management and Yield Prediction in Cotton

Nitrogen (N) is an essential nutrient that controls plant development processes and has impact on the rate of maturity and yield in cotton. Noticeable differences in the physiology of cotton plants, based on N uptake, are observed in the canopy chlorophyll content, canopy  expansion, onset of senescence, and fiber quality and quantity. While N deficit conditions can lead to stress induced early maturity; high amounts of N in the soil can delay maturity. Early harvest before the plants reached full maturity or late harvest can each result is loss of yield. Hence, it is important to optimize the time of harvest. While manually tracking the growth and development of plants in large fields can be cumbersome, UAV-based data collection can capture spatio-temporal variations across large areas quite efficiently.

Multi-spectral aerial images were used in this study to track the developmental stages of cotton throughout the season. The onset of maturity and boll development were tracked by counting the number of flowers observed from the aerial images. Tracking the time of appearance of the first opened boll and its count over the maturity phase are critical in determining the time of harvest and yield. It was observed that the spectral indices (NDVI, NDRE, ExGR) and canopy morphology (canopy height and volume) showed changes in trends during the maturity phase of cotton. However, the determination of maturity based on flowering and boll formation can give more precise estimates of yield and time for scheduling the harvest. The availability of various open source image processing libraries and GIS platforms are to be utilized for feature extraction. This presentation will focus on the feature extraction methodology for counting and tracking changes in flowering and boll opening for the purpose of yield prediction and harvest time estimation. 

(Image and statistical analyses are in progress)