Mostrar registro simples

dc.contributor.advisorArce, Julio Eduardo, 1968-pt_BR
dc.contributor.otherPellico Netto, Sylvio, 1941-pt_BR
dc.contributor.otherUniversidade Federal do Paraná. Setor de Ciências Agrárias. Programa de Pós-Graduação em Engenharia Florestalpt_BR
dc.creatorWeber, Saulo Henriquept_BR
dc.date.accessioned2025-09-22T15:29:57Z
dc.date.available2025-09-22T15:29:57Z
dc.date.issued2006pt_BR
dc.identifier.urihttps://hdl.handle.net/1884/12720
dc.descriptionOrientador : Julio Eduardo Arcept_BR
dc.descriptionCoorientador : Sylvio Pélico Nettopt_BR
dc.descriptionDissertação (mestrado) - Universidade Federal do Paraná, Setor de Ciências Agrárias, Programa de Pós-Graduação em Engenharia Florestal. Defesa: Curitiba, 2006pt_BR
dc.descriptionInclui bibliografia e anexospt_BR
dc.descriptionÁrea de concentração : Manejo florestalpt_BR
dc.description.abstractAo pesquisar diversas espécies arbóreas contidas nos fragmentos de floresta nativa remanescentes na Fazenda Experimental Gralha Azul da Pontifícia Universidade Católica do Paraná - PUCPR, localizada no município de Fazenda Rio Grande, PR, verificou-se que as distribuições probabilísticas aplicadas atualmente não são suficientemente flexíveis para representar a freqüência de sobreviventes da regeneração natural. A partir da análise dos dados coletados, verificou-se que a quantidade de sementes que germinam no interior da floresta é grande, porém, a porcentagem de plantas que se tornam adultas (DAP maior que 10 cm), em relação à quantidade de sementes que germinam, é relativamente baixa. Essa situação não é bem representada por nenhuma das distribuições clássicas, como a Normal, a Weibull, a Gamma, a Beta e a Exponencial, que são comumente utilizadas para esses casos. Fundamenta-se este trabalho na identificação de um modelo matemático que atenda às seguintes condições: ser suficientemente flexível para representar a distribuição de freqüências das alturas de regeneração natural de espécies nativas; e que possa ser convertida em uma função densidade de probabilidade, atendendo às características e propriedades atinentes a ela. Este trabalho teve como objetivo principal testar modelos matemáticos, que foram ajustados por meio de regressão não linear, segundo o procedimento de Marquadt, pelo método de mínimos quadrados e com múltiplas iterações. A característica principal dessa função é a flexibilidade para representar a grande quantidade de indivíduos contidos na primeira classe, seguido de um decréscimo abrupto nas classes subseqüentes. Esse modelo resultou em um melhor ajuste à distribuição de alturas de regeneração natural, em comparação com os modelos clássicos. Foram desenvolvidas as fórmulas para estimar os parâmetros média, variância e moda, bem como para a obtenção dos pontos de inflexão do modelo. As formas que a função pode assumir foram simuladas de acordo com a variação dos coeficientes. Ao comparar as principais distribuições utilizadas no meio florestal e a desenvolvida neste trabalho, verificou-se que essa representou melhor os dados, resultou em bons valores nos testes de aderência, podendo inclusive ser utilizada como função preditiva. A primeira parte do presente trabalho consistiu em revisar a literatura pertinente para aprofundar conhecimentos sobre as distribuições clássicas utilizadas para avaliar mortalidade de regeneração natural. A seguir, manipulando dados de regeneração natural de Sassafrás (Ocotea odorifera (Vell.) Rowher), encontrou-se que os modelos probabilísticos clássicos (Normal, Weibull, Gamma, Beta e Exponencial) não resultaram em um bom ajuste a esses dados. Em um terceiro momento, foi tentado conseguir novas alternativas para resolver este problema, buscando várias possibilidades aplicativas de outras funções matemáticas que pudessem representar bem a dispersão de dados. Após definido o novo modelo procedeu-se às operações para adequá-lo aos requisitos de uma Função Densidade de Probabilidade (FDP). O quinto passo foi derivar as fórmulas para calcular a média e a variância, usando-se o método de expectativa matemática. Na seqüência, foram derivadas as fórmulas para a obtenção da moda e dos pontos de inflexão, bem como foram simuladas as formas das curvas, usando-se variações dos coeficientes da função. Finalmente, os resultados das funções clássicas ajustadas e os do modelo proposto na presente pesquisa foram comparados, usando-se os testes de aderência. O modelo proposto no presente trabalho mostrou-se ser suficientemente flexível para o ajuste aos dados de altura da regeneração natural do Sassafrás. A média aritmética, o desvio padrão e a moda, calculados usando as fórmulas derivadas do modelo proposto, resultaram em 2,29 m, 2,71 m e 0,8 m, respectivamente, os quais tiveram muito boas aproximações com os valores estimados da amostra (2,29 m, 2,66 m e 1,04 m).pt_BR
dc.description.abstractDoing a research in remaining fragments of the Araucaria Forest in the Gralha Azul Experimental Farm of the Pontifical Catholic University of Parana – PUCPR, located at the municipality of Fazenda Rio Grande, PR, it was observed various forest species, where it was verified that the probability distributions currently used does not satisfy the flexibility required to well represent the frequency of natural regeneration survivors. From an analysis of data collected, it was verified that the amount of germinated seeds was high, however, the percentage of these plants that became mature (DBH bigger than 10 cm), in relation to the amount of germinated seeds, was relatively low. This situation is not represented satisfactorily by any of the classic probabilistic distributions as Normal, Weibull, Gamma, Beta and Exponential, which are commonly used for these cases. This work was based on the necessity of identifying a mathematical model that includes the following conditions: presents enough flexibility to represent the frequency distribution of natural regeneration; and that can be converted into a probability density function, considering its characteristics and properties. This work had as main objective to test mathematical models that was adjusted by means of non-linear regression, Marquadt procedure, by the least squares method, with multiple iterations. The main characteristic of this function was to attain enough flexibility to represent the great amount of individuals in the first class, followed by an abrupt decrease in the following classes. This model resulted in a better adjustment for the variable height of natural regeneration data, in comparison to the application of the classical models. The frequency distribution was better represented by the proposed model in this work and resulted in good approximation of the estimated mean and variance values. The expressions were developed to calculate these parameters, as well as those to calculate the mode and the inflection points. The shapes that the function can assume were simulated in accordance with the parameters variations. When comparing the main distributions used in the forest science and the one developed in this work, it was verified that it showed a better adjustment to the height data, also viable to be used as predictive function, and presented non-significance when the application of the statistical adjustment tests. The first part of the present work consisted of literature revision to be acquainted with the application of the classical distributions to evaluate mortality of natural regeneration. After manipulating natural regeneration data of Sassafras (Ocotea odorifera (Vell.) Rowher), it was verified that the classical probabilistic functions (Normal, Weibull, Gamma, Beta and Exponential) did not well adjusted to mortality and survival distribution data, because the shape of the observed data required a better flexibility of the mathematical model. At the third moment, it was tried to get new alternatives to solve this problem, by searching various possible applications of other mathematical functions that could well represent the data dispersion. After defined the new proposed model, it was proceeded operations to adequate it to the requirements of a Probability Density Function (PDF). The fifth step was to derive the formulas to calculate the mean and the variance, using the method of mathematical expectation. After, the mode and the inflection points formulas were derived, as well as the shape curves was simulated, using variation of the parameters values. Finally, the results of the adjusted classical functions and the proposed in the present research were compared by means of goodness of fit tests. The model proposed in this research work showed to be enough flexible for the adjustment of Sassafras natural regeneration height data. The mean, the standard deviation and the mode calculated by the formulas derived from the proposed distribution model resulted in 2.29 m, 2.71 m and 0,8m, respectively, which had very good approximation with the estimated sample values (2.29 m, 2.66 m and 1,04m).pt_BR
dc.format.extentxiii, 74 f. : il., grafs.pt_BR
dc.format.mimetypeapplication/pdfpt_BR
dc.languagePortuguêspt_BR
dc.relationDisponível em formato digitalpt_BR
dc.subjectFlorestas - Reprodução - Paranápt_BR
dc.subjectDistribuição (Teoria da probabilidade)pt_BR
dc.subjectRecursos florestais e engenharia florestalpt_BR
dc.titleDesenvolvimento de nova função densidade de probabilidade para avaliação de regeneração naturalpt_BR
dc.typeDissertaçãopt_BR


Arquivos deste item

Thumbnail

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

Mostrar registro simples