Detecção de árvores individuais por meio do lidar para a gestão da arborização do Campus-Dois Irmãos da UFRPE

Abstract

Urban forests are elements that promote the balance between urban and rural environments, presenting important social, political, economic and architectural character, besides providing an improvement in the population's quality of life. LiDAR is a technology that has been used in forest applications with certain frequency growth, mainly due to the fast availability and access to data and its accuracy. Forest inventory is an essential activity for the qualitative and quantitative knowledge and characterization of forest areas, but it is very time consuming, so many companies and professionals are using remote sensing techniques to optimize the performance of the activity, in terms of time, cost and efficiency. With the constant improvement of technologies, remote sensing became part of the techniques and methodologies used in the monitoring and management of green areas. Automatic detection of individual trees is a fundamental procedure for studies that aim to extract dendrometric data at tree level, height, for example, is a measure that in the field can be difficult to obtain in the traditional way. These data from individual trees allow us to represent canopy and / or canopy characteristics, as well as future analyzes and structural and floristic characterizations of the vegetation. Given the above, the objective of the present work was to explore the potential of airborne LiDAR data for detecting individual trees in an urban forest area at the UFRPE headquarters campus, using Fusion, the technology has as its principle a laser scanning of a terrain emitting pulses at a high frequency and calculating the return time of these pulses to the sensor, forming a cloud of dots mapping the terrain below. Using the Fusion software, soil and surface separation was performed using the Groundfilter command, followed by the creation of digital terrain and surface models using the GridsurfaceCreate command. Still in a Fusion environment, a digital canopy model (CHM) was created using the Canopymodel command. After processing, data analysis was performed in RStudio, using packages and functions specific to this type of processing. In the FindTreesDetection function, three filter window sizes were used, since the study refers to the vegetation area without standard spacing. The results showed that in the study area, as calculated by RStudio, based on the 3x3, 5x5 and 7x7 filter windows, there are, respectively, 6,263, 2,356 and 1,367 trees identified during processing.