Discriminating between Brant`s oak (Quercus brantii Lindl.) and gall oak (Q. infectoria Oliv.) species using the UAV images

Today, tree species classification and mapping play an important role in decision making for sustainable forest management. The high spatial resolution of the UAV images makes them an effective tool for identifying tree species. The aim of this study was to evaluate the capability of unmanned aerial vehicle (UAV) imagery to detect Brant`s oak (Quercus brantii Lindl.) and gall oak (Q. infectoria Oliv.) species in the Kakasharaf area in the Lorestan Province, Iran. For this purpose, three stands were selected with areas of 3.6, 4.9 and 5.4 ha. The images were taken in May 2017 by a Phantom 4 UAV. Mosaic images were prepared using 10, 12 and 15 ground control points, respectively. Specifically, differentiation between two species was based on the classification of images by artificial neural network and spectral information. The reference data was prepared to evaluate the classification results by field survey, and classification was conducted by using 70% of the samples as training samples and the remaining 30% as test samples. Results showed that better performance achieved by neural network classification in all three stands with kappa coefficients of 0.77, 0.76 and 0.82 and overall accuracy of 84.03, 83.42 and 87.37 percent compared with the spectral classification method, which returned kappa coefficients of 0.7, 0.64 and 0.63 and overall accuracies 78.81, 73.4 and 72.19 percent, respectively. Conclusively, UAV data revealed to have a good ability to distinguish between Q. brantii and Q. infectoria in the study area, which suggests that those data can be used for discriminating between different tree species in similar forest areas.