Quantifying the effects of soil texture and weather on cotton development and yield using UAV imagery

Quantification of interactions of soil conditions, plant available water and weather conditions on crop development and production is the key for optimizing field management to achieve optimal production. The goal of this study was to quantify the effects of soil and weather conditions on cotton development and production using temporal aerial imagery data, weather and soil apparent electrical conductivity (EC a ) of the field. Soil texture, i.e., percent of sand and clay content, was calculated from EC a to estimate three soil quality indicators, including field capacity, wilting point and total available water. A water stress coefficient K s was calculated using soil texture and weather data. Image features of canopy size and vegetation indices (VIs) were extracted from unmanned aerial vehicle (UAV)-based multispectral images at three growth stages of cotton in 2018 and 2019. Pearson correlation ( r ), analysis of variance (ANOVA) and eXtreme Gradient Boosting (XGBoost) were used to quantify the relationships between crop response derived from UAV images and environments (soil texture and weather). Results showed that soil clay content in shallower layers (0–0.4 m) affected crop development in earlier growth stages (June and July) while those in deeper layers (0.4–0.7 m) affected the later-season growth stages (August and September). Soil clay content at 0.4–0.7 m had a higher impact on crop development when water inputs were not sufficient, while K s features had a higher contribution to the prediction of crop growth when irrigation was applied and water stress was less.