Estudo das distribuições diametricas da floresta do planalto Tapajos-Para

• Autor(a) principal: Barros, Paulo Luiz Contente de, 1952-
• Data de Publicação: 2013
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da UFPR
• Texto Completo: http://hdl.handle.net/1884/25616
• topic: Florestas - Amazônia (BR); Levantamentos florestais; Florestas - Medição - Amazônia (BR); Mapeamento florestal; Silvicultura - Amazônia (BR); Teses

Abstract: The objectives of this research were to test several mathematical models which express the diameter distribution, as well as to verify the influence of the diameter class interval in the fitness of the tested models. The models we re applied individually for: all species in the area, for the most frequent and the commercial species. The data for this research carne from the National Forest of Tapajós, located in the county of Santarem, state of Para, Brazil. A total number of li.i73 trees with diameter breast height above 15 centimeters, and totalling 167 species were measured. The following mathematical models were tested to describe the diameter distribution of the area: MODEL 1: Exponential, type I yi = A . e -BXi MODEL 2: Exponential, type II yi = A . e -BX²i MODEL 3: Power function, MERVART yi = A . e Xi-B MODEL 4: Hyperbola, PIERLOT yi = [(xi - x0) / (10A+BXi - 1)] + y0 MODELO 5: Polynomial, Goff & West log y = b0 + b1Xi + b2Xi2 + … + bnXin MODELO 6: Weibull function F (X) = 1 - exp {-(X/b)C} MODELO 7: Beta function F (X) = c (X-a)a . (b-X)? The sum of squares of residuals was used as a criterium to select the equations by group of species, and by class intervals. The probability associated with the number of runs of residuals, which gives the probability that the observed number of signs changes in the plot of residuals occurred randomly, was also used. Following these criteria, among the seven tested models, to represent the relationship of the number of trees by 1OO ha on diameter classes, the Beta function, the polynomial and the exponential models presented the best results in fitting such relationship. The Weibull function also seems to estimate very well the percentages of the number of trees smaller or equal to a specified diameter of the stand. In comparing class intervals the models in general presented the best results for the class interval of 1O cm, the precisson of the equations decreased with the decreased sing of the class intervals.