Please use this identifier to cite or link to this item:
Title: Predição de volume em florestas plantadas utilizando escaneamento com laser aerotransportado e aprendizagem de máquinas
Other Titles: Volume prediction in planted forests using airborne laser scanning and machine learning
Authors: Carvalho, Luís Marcelo Tavares de
Carvalho, Luís Marcelo Tavares de
Ferreira, Maria Zélia
Calegário, Natalino
Keywords: Escaneamento aéreo com laser
Manejo florestal
Inventário florestal
Aprendizagem de máquinas
Airborne laser scanning
Forest management
Forest inventory
Machine learning
Issue Date: 28-Jan-2022
Publisher: Universidade Federal de Lavras
Citation: FERREIRA, F. C. Predição de volume em florestas plantadas utilizando escaneamento com laser aerotransportado e aprendizagem de máquinas. 2021. 49 p. Dissertação (Mestrado em Engenharia Florestal) – Universidade Federal de Lavras, Lavras, 2022.
Abstract: The management of planted forest have been increasing the necessity for more and fast information, in quantity and quality, to base its decision-making. Due to it, the present work aiming compare the volumetric estimative obtained by the forest inventory using the conventional method, based on the mean and standard deviation of the sampled units, and a method that uses data from airborne laser scanning (ALS) of the stands of the forest to generate maps with the volume of wood. In the ALS inventory, three different modeling methods were tested: linear model adjusted by ordinary least squares (OLS), Random Forest (RF) and Support Vector Machines (SVM). The models were trained and validated by the holdout method for 200 times, evaluating in the end the mean of the coefficient of determination (R²) and the root mean square error (RMSE). The model that obtained the best result was the linear model with R² of 53.74% and RMSE of 20,19 m³/ha. When comparing conventional and ALS inventory methods, both had low inventory errors. It can be concluded that the use of ALS inventories is promising for the forest sector.
Appears in Collections:Engenharia Florestal - Mestrado (Dissertações)

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.