Leek growth monitoring using multispectral UAV imagery

In recent decades, unmanned aerial vehicle (UAV)-based crop monitoring in arable crops has evolved into practical applications. However, UAV-based research on capital-intensive vegetable crops, such as leeks, has been limited. Therefore, this study aimed to estimate leek dry biomass from multispectral UAV images. To this end, 60 experimental plots spread across four fields were planted with leeks and subjected to different fertilization strategies. Next, these plots were monitored throughout the growing season by destructive plant sampling and multispectral UAV imaging. Robust regression models for estimating leek dry biomass were built through a leave-one-flight-out cross-validation. The results showed that a partial least squares regression model based on ten spectral vegetation indices in combination with plant height and green ground crop cover gave the best prediction of the dry biomass RMSEcv=5.6 g/plant, RRMSEcv=7.3%).