Determining leaf nutrient concentrations in citrus trees using UAV imagery and machine learning

Abstract Nutrient assessment of plants, a key aspect agricultural crop management and varietal development programs, traditionally is time demanding labor-intensive. This study proposes novel methodology to determine leaf nutrient concentrations citrus trees by using unmanned aerial vehicle (UAV) multispectral imagery artificial intelligence (AI). The was conducted in four different field trials, located Highlands County Polk County, Florida, USA. In each location, trials contained either ‘Hamlin’ or ‘Valencia’ sweet orange scion grafted on more than 30 rootstocks. Leaves were collected analyzed the laboratory macro- micronutrient concentration traditional chemical methods. Spectral data from tree canopies obtained five bands (red, green, blue, red edge near-infrared wavelengths) UAV equipped with camera. estimation model developed gradient boosting regression evaluated several metrics including mean absolute percentage error (MAPE), root square error, MAPE-coefficient variance (CV) ratio difference plot. determined macronutrients (nitrogen, phosphorus, potassium, magnesium, calcium sulfur) high precision (less 9% 17% average for respectively) micro-nutrients moderate 16% 30% respectively). Overall, this UAV- AI-based efficient generate maps commercial orchards could be applied other species.