The classification of tree species and their spatial location display are difficulties in forest ecological investigation and are of great significance for forest management and biodiversity conservation This study explores the application value of the combination of GF-2 multispectral image data and LiDAR point cloud data from unmanned aerial vehicles in the classification of tree species in mountainous bamboo forests Taking the mountainous areas in central Fujian as an example we extracted spectral and texture features of GF-2 imagery as well as height features intensity features terrain factors and canopy height model CHM features from LiDAR data Random forest RF was used to select feature variables and the application potential of single GF-2 image single LiDAR data and the combination of GF-2 image and LiDAR data in tree species classification was compared Three distinct feature combination schemes were designed and two machine learning classifiers-Random Forest RF and Support Vector Machine SVM -were employed for the classification and recognition of tree species within the distribution area of bamboo forests The research results indicate that the 25% importance of all echo intensities in the LiDAR feature set is the highest and the mean blue light band 11_B1_Mean and green normalized vegetation index GNDVI of window 11×11 in GF-2 texture features perform well Compared with different data sources GF-2 combined with LiDAR data has better classification accuracy than single source data Scheme F3 combines terrain factors and canopy height features to improve the accuracy of tree species classification was improved from 78 47% to 92 55% with an overall accuracy of 92 55% and a kappa coefficient of 0 91 With the addition of auxiliary features the classification accuracy is significantly improved and the recognition accuracy of bamboo forest tree species is improved