Banca de DEFESA: ANTÔNIA GEICIANE SILVA DE OLIVEIRA DANTAS
Uma banca de DEFESA de MESTRADO foi cadastrada pelo programa.STUDENT : ANTÔNIA GEICIANE SILVA DE OLIVEIRA DANTAS
DATE: 21/05/2021
TIME: 14:30
LOCAL: Sala de reuniões do google
TITLE:
Estimating of hollow volume in trees of a dense rain forest in the eastern Brazilian Amazon
KEY WORDS:
Forest mensuration; tree defects; regression analysis; forest production
PAGES: 86
BIG AREA: Ciências Agrárias
AREA: Recursos Florestais e Engenharia Florestal
SUBÁREA: Manejo Florestal
SPECIALTY: Dendrometria e Inventário Florestal
SUMMARY:
The occurrence of hollows in trees in the Amazon rainforest, in many cases, can be expressive and has a negative impact on the production of timber in managed forests. In this context, the use of mathematical models to estimate the volume of hollows in standing tree boles in stock surveys, can help on estimating of the net standing volume. The aim of the study was to investigate and evaluate the feasibility of using mathematical models to estimate volume of tree hollows in a tropical rain forest in the western Pará State, Brazilian Amazon. Hollow volumes were calculated for 532 sample trees of 17 commercial species. To estimate hollow volumes, three linear and three linearized mathematical models commonly used to estimate volumes of tree boles were tested by the Ordinary Least Squares Method (OLS). The models were tested for the set of all species thus generating a general equation, and one specific equation for Manilkara elata. Two strategies were applied to test the models: the hollow volume was estimated having as independent variables the diameter D of the hollow of the hollow and its corresponding length (h) - strategy (A), and the DBH and the length of the hollow – strategy B. The criteria used for selecting the best models were the highest determination coefficient (R2), the lowest standard error of the estimate percent (Syx%) the distribution of the residues. The best fit was achieved for the linearized models, and the Spurr in linear form. Similar results were to obtained to Manilkara elata. For all the tested models, the distribution of residues was not homogeneous. The research findings indicated that the Spurr and the Schumacher & Hall mathematical models which use the length of the hollow as independent variables can only be used for used to estimate bole volumes of fallen trees. As the main focus of the research was to estimate the volume of hollows in standing trees and due to the impossibility of estimating the length of the hollow in the field, is recommended using the equations only for fallen boles. It is also recommended the continuity of the study with a more robust data set, having more sample trees in the smaller and in the larger hollow sizes as an attempt to improve the precision. The use of machine learning techniques such artificial neural networks could be a path to be followed.
BANKING MEMBERS:
Presidente - 142.060.489-91 - JOSE NATALINO MACEDO SILVA - Oxford
Externo ao Programa - 2295917 - ANGELO AUGUSTO EBLING
Externo ao Programa - 4143218 - JOAO OLEGARIO PEREIRA DE CARVALHO
Externa ao Programa - 1399943 - SINTIA VALERIO KOHLER
Externo à Instituição - EDSON MARCOS LEAL SOARES RAMOS - UFPA