PREDICTING COTTON ABOVE GROUND BIOMASS USING GEOAI AND MULTISPECTRAL IMAGERY FROM UNCREWED AERIAL SYSTEMS

Monitoring the biomass of cotton allows agriculturalists to modify their management practices to optimize yield and address issues such as damage from storms and impacts of climate change. Across the 2018 and 2019 growing seasons, remote sensing images were acquired by the U.S. Department of Agriculture (USDA) Agricultural Research Service (ARS) via Uncrewed Aerial Systems (UAS). These images provided high spatial and temporal resolution, multispectral image data for the analysis of cotton at the Ashburn Cooperator Farm in Ashburn, Georgia and the Ty Ty Cooperator Farm in Ty Ty, Georgia. Ground measurements of cotton biomass collected by the USDA-ARS harvesting plants in representative plots were used to scale up to the field level by using geographic artificial intelligence (GeoAI) machine learning models, Random Forest and XGBoost. Results displayed models using the raw bands predicted cotton biomass within the designated range and outperformed models using vegetation indices.