Using canopy height model derived from UAV images to tree height estimation in Sisangan forest

Recent advances in unmanned aerial vehicles (UAVs) technology, as well as the development of lightweight sensors, offers a great possibility for the measurement of different tree features with relatively low costs compared to traditional methods. In this research, the precision and accuracy of tree height measurement and estimation using imagery by a low-cost UAV were studied. For this aim, 854 images with an altitude of 100 m above the ground were taken and the images were processed and dense point cloud was extracted by applying Structure from Motion (SFM) algorithm. The study was conducted in 34.79 ha of Sisangan forest park and 28 sample plots (30 × 30 m) were located in the field and tree heights were measured. Also, tree height was measured using the canopy height model. Linear regression was applied to estimate the actual tree heights based on CHM derived tree eights. The accuracy and precision of the estimates were assessed using relative bias and relative root mean square error. The differences between the field measured and CHM derived tree heights were statistically significant. Based on the results, the relative root means the square error of the height estimation of Buxus hyrcana, Carpinus betulus, Parrotia persica, and other species was 20.39, 20.39, 20.57 and 20.52 percent, respectively. The results showed that tree height measurement based on UAV images and methods that were applied in this research, is biased and the estimations are highly uncertain.