Abstract
<jats:p>This study investigates the effectiveness of combining UAV multispectral imagery, biomass-chlorophyll indices, and an unsupervised machine learning classifier (UMLC) to map carob tree dieback caused by the Barbary macaque. The data were rigorously preprocessed, twelve biomass-chlorophyll indices were derived and UMLC was implemented, analyzed, and validated. The results show that among the tested indices, TDVI and CIG exhibited similar and superior performance with excellent dynamic range, and the map generated from the TDVI threshold showed a clear spatial distribution of carob classes. Similarly, UMLC proved its ability to classify individual trees according to their degree of dieback in a complex environment, achieving an excellent overall classification accuracy of 93%. However, since the tested approaches provide almost alike results, the method based solely on spectral indices is recommended for its simplicity, facility to transfer between sensors, and it constitute an accurate alternative compared to the sophisticated and complex methods such UMLC.</jats:p>