Avaliação de metodologia para inventário de uva-do-Japão (Hovenia dulcis Thunb.) utilizando imagens de veículos aéreos não tripulados (VANT) na UTFPR - campus Dois Vizinhos

Abstract

Due to the large logging in Paraná Southwest, the Araucaria’s Forest is suppressed in forest fragments. In the city of Dois Vizinhos, the forest fragment at UTFPR Campus Dois Vizinhos has the floristic composition, among other species, raisin tree (Hovenia dulcis Thunb.). Because it is an exotic invasive species, this specie can cause imbalance in flora and fauna and get nutrients and light. Its population should be controlled and its timber has economic potential and it’s used mainly for energy. Among the methods for estimating the total volume of wood in a population , there is the use of drone images. The study aimed to estimate the volume of wood of raisin tree (Hovenia dulcis Thunb) using drones images in the Federal Technological University of Paraná - Campus Dois Vizinhos. Individuals of raisin tree have been identified in the picture and it had their crown area measured by vectorization and by measuring the diameters in the north-south and east-west diretion in a GIS environment. Subsequently, the pre-selected individuals were found the field through the use of polar coordinates and it had the full height, diameter at breast height and canopy diameters in the north-south and east-west measured. Thus, it was made a Pearson correlation matrix between the variables of the image and field, where there was no correlation between dbh, total height and field volume with the image. Therefore, it was not possible to do modeling for individuals of raisin tree. Among the factors that may have influenced for that to happen is the spatial resolution of the drone image (0,15x0,18m), possible non-overlapping in the drone images, the time of data collection in the picture and data field, and errors in the assessment process parameter field and GIS. Thus, there is correlation between the variable total height, crown area and volume. Moreover, it is noted that between the two crown area of the image scaling methods, for vectorization and the diameters no statistical difference between them compared with the crown area measured field.