Mostrar registro simples

dc.contributor.advisorFigueiredo Filho, Afonso, 1951-pt_BR
dc.contributor.authorMôra, Rômulopt_BR
dc.contributor.otherMachado, Sebastião do Amaral, 1939-pt_BR
dc.contributor.otherUniversidade Federal do Paraná. Setor de Ciências Agrárias. Programa de Pós-Graduação em Engenharia Florestalpt_BR
dc.date.accessioned2020-02-12T14:11:12Z
dc.date.available2020-02-12T14:11:12Z
dc.date.issued2015pt_BR
dc.identifier.urihttps://hdl.handle.net/1884/40830
dc.descriptionOrientador : Prof. Dr. Afonso Figueiredo Filhopt_BR
dc.descriptionCoorientadores : Prof. Dr. Sebastião do Amaral Machadopt_BR
dc.descriptionTese (doutorado) - Universidade Federal do Paraná, Setor de Ciências Agrárias, Programa de Pós-Graduação em Engenharia Florestal. Defesa: Curitiba, 05/03/2015pt_BR
dc.descriptionInclui referências : f. 234-250pt_BR
dc.descriptionÁrea de concentração : Manejo florestalpt_BR
dc.description.abstractResumo: Neste estudo objetivou-se avaliar a eficiência de funções de afilamento de forma variável e os efeitos da aplicação de modelagem de efeitos mistos em fustes de Pinus taeda e Eucalyptus saligna. Os dados utilizados foram obtidos de plantios na região Centro-Oriental do estado do Paraná. Foram utilizados para o estudo 2252 fustes de Pinus taeda e 1027 fustes de Eucalyptus saligna, sendo uma parte para o ajuste e outra para validação dos modelos. A cubagem foi realizada, para ambas as espécies e consistiu em mensurar diâmetros nas alturas (h) absolutas 0,1; 0,3; 0,7 e 1,3 m e nas alturas relativas 5, 10, 15, 25, 35, 45, 55, 65, 75, 85 e 95% da altura total (H) em cada árvore. Para avaliar a acurácia das estimativas de diâmetro e volume, foram testados 43 modelos de afilamento, sendo 27 não segmentados (NS), 4 segmentados (S) e 12 de forma variável (FV), ajustados segundo técnica de regressão não linear (nls) em função da variável diâmetro (d). Os volumes foram calculados utilizando a fórmula de Smalian, somando-se os volumes parciais até o ponto de interesse. No primeiro momento seguiu-se a avaliação de todos os diâmetros do fuste em todas as alturas relativas e o volume total. Para avaliar melhor as estimativas das duas variáveis em estudo, o fuste foi dividido em três partes, de 0,1 m a 25% H (base), de 25 a 75% H (parte mediana) e de 75 a 95% H (ápice do fuste). A validação dos modelos foi feita pelo teste de Qui-quadrado com o uso dos coeficientes obtidos com os ajustes das funções. Os modelos com melhor performance NS, S e de FV, na avaliação de todos os diâmetros do fuste, foram selecionados para o ajuste segundo a técnica de modelagem de efeitos mistos não linear (nlme). Para avaliar diferentes cenários, utilizaram-se todas as combinações de 1, 2 ou 3 coeficientes aleatórios, estratificados por fuste para ambas espécies. As avaliações da acurácia das estimativas de diâmetro, sem estratificação e com estratificação e volume, total e parciais, ajustadas pela técnica nls e nlme foram feitas com as estatísticas R2(%), Syx(%) e análise gráfica de resíduos. Outros critérios estatísticos complementares foram utilizados, V, MD, DPD, MSE, PVE e AIC. Os resultados confirmam que as equações de FV foram melhores no ajuste em relação as equações S e as duas tiveram melhor performance que as equações NS para os diâmetros e os volumes. Na avaliação das estimativas de diâmetro sem estratificação, as equações com melhores desempenhos foram Demaerschalk(e) (1973) (NS), Clark et al.(1991) (S) e Bi (2000) (FV). Na avaliação das estimativas de volume total, as equações com melhores desempenhos foram Demaerschalk (1972) para Pinus e Ormerod (1973) para Eucalipto e Clark et al.(1991) e Bi (2000) para ambas. Nas avaliações das estimativas estratificadas e dos volumes parciais pelo ajuste nls, as equações mais acuradas nem sempre foram às obtidas nos ajustes de todos diâmetros e volumes totais. Assim, equações de afilamento nem sempre estimam com a mesma acurácia todos os diâmetros e volumes ao longo do fuste. A avaliação das estimativas de diâmetro e volume para Demaerschalk(e) (1973), Clark et al.(1991) e Bi (2000) pela técnica nlme mostram que o uso de apenas um coeficiente aleatório melhora as estimativas, principalmente na parte média e apical do fuste, tanto para as avaliações não estratificadas como estratificadas. Pelos resultados obtidos, os modelos de FV foram melhores que os NS e S em todos os estudos. O modelo de Bi (FV) foi o melhor na maioria dos estudos. Os modelos de FV são melhores e fáceis de ajustar que os S. A técnica nlme melhora as estimativas das variáveis, em relação a técnica nls, principalmente na porção mediana e apical. Palavras-chave: Modelos. Técnicas de Modelagem.Variáveis Dendrométricas.pt_BR
dc.description.abstractAbstract: This study aimed to evaluate the efficiency of variable-form taper functions and effects of the application of mixed-effects modeling to stems of Pinus taeda and Eucalyptus saligna. Data used were obtained from stand in the East-Central region of the Parana state. They were used in this research 2252 stems of Pinus taeda and 1027 stems of Eucalyptus saligna, being a part for the fitting and another one for model validations. The scaling was made, for both species, and consisted of measuring diameters at the absolute height (h): 0.1; 0.3; 0.7; 1.3 m and at relative heights: 5, 10, 15, 25, 35, 45, 55, 65, 75, 85, and 95% of the total height (H) of each tree. To evaluate the accuracy of the estimates of diameter and volume they were tested 43 taper models, being 27 non-segmented (NS), 4 segmented (S) and 12 variable-form (FV), adjusted by non-linear regression (nls) as a function of the variable diameter (d). The volumes were calculated using Smalian formula adding the partial volumes up to the point of interest. Initially were evaluated all stem diameters at all relative height and the total volume. To better evaluate the estimates of the two variables studied, the stem was divided into three parts: from 0.1 m to 25% H (base), from 25% to 75% H (middle part) and from 75% to 95% H (stem top). The model validation was performed using the Chi-square test with the use of the coefficients obtained through on the fitting of the functions. The better performance model, NS, S and FV, in the evaluation of all stem diameters, were selected to the fitting according to the modeling technique of nonlinear mixed-effects (nlme). To evaluate different scenarios, were used all combinations of 1, 2 or 3 random coefficients, stratified by stem for both species. The evaluations of accuracy of the estimates of diameter without stratification and with stratification, and volume, total and partial, adjusted by the techniques nls and nlme, were made by the statistics R2(%), Syx (%) and graphical analysis of residuals. Other complementary statistical criteria were used, V, MD, DPD, MSE, PVE and AIC. The results confirm that FV equations were better in relation to the S equations and the two had better performance than NS equations for diameters and volumes. In evaluating the diameter estimates without stratification, the better performance equations were Demaerschalk(e) (1973) (NS), Clark et al. (1991) (S) and Bi (2000) (FV). In evaluating the total volume estimates, the better performance equations were Demaerschalk (1972) to Pinus and Ormerod (1973) to Eucalipto and Clark et al. (1991) and Bi (2000) to both. In the evaluations of stratified estimates and partial volumes by nls fitting, the most accurate equations were not always the same obtained in the fitting of all diameters and total volumes. Thus, taper functions not always estimate with the same efficiency all diameters and partial volumes along the stem. The evaluation of the estimative of the diameter and volume by Demaerschalk(e) (1973), Clark et al. (1991) and Bi (2000) nlme shows that the technique using only one random component improves the estimates, principally in middle and apical part of the stem, with and without stratification. The results show that the FV models were better than the NS and S in all studies. The Bi (FV) model was the best in most studies. The FV models are better and easy to fit than the S ones. The nlme technique improves the estimates of the studied variables, in relation to nls technique, especially in the middle part and the apical parts. Keywords: Models. Modeling Techniques. Dendrometric Variables.pt_BR
dc.format.extent274 f. : il., grafs., tabs.pt_BR
dc.format.mimetypeapplication/pdfpt_BR
dc.languagePortuguêspt_BR
dc.relationDisponível em formato digitalpt_BR
dc.subjectRecursos Florestais e Engenharia Florestalpt_BR
dc.subjectDendrometriapt_BR
dc.subjectArvores - Mediçãopt_BR
dc.subjectPinus taeda - Mediçãopt_BR
dc.subjectEucalipto - Mediçãopt_BR
dc.titleFunções de afilamento de forma variável e modelagem de efeitos mistos em fustes de Pinus taeda e Eucalyptus salignapt_BR
dc.typeTesept_BR


Arquivos deste item

Thumbnail

Este item aparece na(s) seguinte(s) coleção(s)

Mostrar registro simples