Open Journal Systems

USO DE MODELOS ESTATÍSTICOS PARA CLASSIFICAÇÃO DE SÍTIOS EM POVOAMENTOS DE Pinus taeda NA REGIÃO DE CAÇADOR – SC

Saulo Jorge Téo, Diego Ricardo Bressan, Reinaldo Hoinacki da Costa

Resumo


Este trabalho teve como objetivo testar diferentes modelos estatísticos para ajuste de curvas de índice de sítio e verificar se as curvas anamórficas geradas foram satisfatórias para classificação de sítios em povoamentos de Pinus taeda L. na região de Caçador, SC. Os dados de altura dominante (hdom) utilizados nesta pesquisa foram obtidos de parcelas permanentes de área fixa e formato retangular, com áreas de 500 e 625 m². As parcelas foram distribuídas de forma aleatória nos povoamentos, a fim de abranger a maior variabilidade possível de produtividade. A seleção do melhor modelo estatístico ajustado foi feita por meio das seguintes estatísticas de ajuste e precisão: coeficiente de determinação ajustado (R²aj), erro padrão da estimativa (Syx), erro padrão da estimativa em porcentagem (Syx%) e distribuição de resíduos em porcentagem. O modelo monomolecular apresentou os melhores resultados para os critérios de seleção, portanto foi escolhido para a construção das curvas de índice de sítio pelo método da curva guia. Depois de se proceder à classificação de sítios, foram realizados testes de anamorfismo, os quais indicaram o padrão anamórfico das curvas geradas pelo modelo estatístico ajustado para classificação de sítios dos povoamentos de Pinus taeda na região de Caçador, SC.

Palavras-chave:  Modelos estatísticos; teste de anamorfismo; curvas de índice de sítio; altura dominante.

 

Abstract

Use of statistical modells for site classification of Pinus taeda plantation in the region of Caçador, Santa Catarina State, Brazil. This study aimed to test different statistical models for fitting site index curves and check if the anamorphic curves generated were satisfactory to classify the loblolly pine (Pinus taeda L.) plantations in the region of Caçador, Santa Catarina State, Brazil. The data of dominant height (hdom) used in this study were obtained from permanent plots of fixed area and rectangular shape, with areas of 500 and 625 square meters. The plots were distributed randomly in the stands to cover the largest possible variability of productivity. The selection of the best equation was made by the following fitting and precision statistics: adjusted coefficient of determination (R²aj), standard error of estimation (Syx), standard error of estimate in percentage (Syx%) and graphical distribution of the residuals in percentage. The monomolecular model showed the best performance, so it was chosen for the construction of the site index curves by guide curve method. After construction of the site index curves, the stability of the curves was also tested, which indicated the anamorphic pattern of the curves generated by the equation for site classification of loblolly pine plantations in the region of Caçador, Santa Catarina State, Brazil.

Keywords:  Statistical models; test of anamorphism; site index curves; dominant height.


Palavras-chave


Modelos estatísticos; teste de anamorfismo; curvas de índice de sítio; altura dominante.

Texto completo:

PDF


DOI: http://dx.doi.org/10.5380/rf.v41i1.21201