Dados remotamente situados para estimativa de altura em remanescente florestal de Dois Vizinhos - PR

Abstract

This study evaluated the technical feasibility and accuracy of using Un-manned Aerial Vehicles (UAVs) and LiDAR sensors integrated with smartphones for collecting dendrometric data, comparing them to conventional forest inventory methodolgies.Tree heights were measured using the Vertex hypsometer (reference method), a smartphone LiDAR sensor via the Arboreal app, and the Canopy Height Model (CHM), derived from subtracting the Digital Terrain Model (DTM) from the Digital Surface Model (DSM) obtained through aerial photogrammetry. The data were subjected to nonparametric statistical analysis using the Friedman test (α = 0.05), due to thepaired nature of the samples and the distribution of the residuals. One individual was excluded due to a Vertex reading failure, resulting in a final sample of 15 individuals. The results indicated that there was no statistically significant difference between the measurement methodologies (n = 15; p = 0.105), however, the error metrics showed greater agreement between the mobile LiDAR and Vertex compared to the photogrammetric CHM. LiDAR presented an MAE of 0.5867 m and an RMSE of 0.7633 m, while CHM presented an MAE of 1.3953 m and an RMSE of 1.6981 m. It is concluded that, under the evaluated conditions, smartphone LiDAR sensor peformed better for individual height estimates, while UAV-derived CHM showed potential for spatial characterization of the canopy, albeit with greater variability at the ind vidual level.