Estimativa dos parametros da floresta tropical a partir da analise da estrutura dendrologica e espacial das arvores e pelo metodo do vizinho mais proximo
Abstract
Resumo: A área do estudo situa-se na Flona Tapajós Santarém, PA, abrangendo 1000ha. A tese consiste na análise comparativa de processos de amostragem diversificados com inventário realizado com 100% de intensidade e mapeamento de todas árvores com dap acima de 55 cm. Os resultados mostram que quanto menor forem a intensidade de amostragem e a frequência de ocorrência das espécies, ocorre uma sistemática superestimativa das frequências e volumes. De uma forma geral as estimativas passam ter consistência a partir de uma intensidade de amostragem de 5% da área da população amostrada. Das técnicas de amostragem testadas as estimativas mais consistentes a nível de espécie e distribuição diamétrica foi a do vizinho mais próximo, sendo que os erros de estimativa podem ser definidos através de regressão linear. O processo do vizinho mais próximo permite efetuar após estratificação da floresta em termos horizontais e verticais, assim como avaliar sua potencialidade econômica. A metodologia adotada permite inferir que é possível desenvolver modelos matemáticos para definir a nível regional, intensidades de amostragem adequadas para inventários florestais, com vistas ao manejo de florestas tropicais. Abstract: The study area is located in the Tapajos Forest, Santarem, Para, with, 1000 ha. This thesis develops a comparative analysis of the sampling processes applied to a 100% inventory made with the located coordinates of all sampled trees with 55cm dbh and above. The results showed that, the smaller is the sample size taken, and the smaller is also the species frequency, the larger is the systematic superestimation of the frequencies and the volume in the forest. Generally, the estimates show consistence above a sample size over 5%, taken in terms of the sampled area. From the sampling techniques tested in this case, the more consistent estimate on the species basis and diameter distribuition, it was the "Closest neighbor method", where the standard errors could be estimated through linear regression techniques. The "Closest neighbor method" allows one to make a pos-stratification of the forest, considering horizontal and vertical variations, as well as its economical potentialities. The used methodology allows one to conclude that it is possible to develop mathematical models for better defining the sample size for inventories to be conducted for management purposes in the tropical forests.
Collections
- Teses & Dissertações [8698]