The identification of forest species is an important task in urban management. Traditional methods to do this can be expensive and take a lot of time. By evaluating UAV RGB and multispectral point clouds from the El Paraiso Park (Cuenca – Ecuador), we were able to segment individual trees and extract relevant data. Combining spectral and geometric features of trees, we assessed the potential of using RGB data along with geometric characteristics such as tree height and crown height for species identification. The results demonstrated higher accuracy for species like Sauce (F1-score = 0.88) and Fresno_Cholan (F1-score = 0.73) compared to the use of multispectral cloud points. This work offers an approximation for urban forest species identification based on RGB and geometric features.