Aplicação de tecnicas multivariadas em mapeamento e interpretação de parametros do solo

AUTOR(ES)

Benedito de Freitas Bueno

DATA DE PUBLICAÇÃO

2001

RESUMO

The goal of this research was to investigate methodology for studying the spatial variability contained in a set of parameters collected in an experimental area of 4810 m2 in Piracicaba-SP (data published in a doctoral dissertation). For that, Geostatistics and other multivariate techniques were applied combined. Principal Components Analysis (PCA) was used for identifying variables which explain a major part of the variability in the whole set of parameters. Also, Cluster Analysis was performed on the parameters of the soil and/or plants, or on the principal components for defining homogeneous areas. Geostatístics was applied, fitting semivariogram models, followed by kriging interpolation of the data generated by the PCA. Afier that, maps of the components were built in order to interpret more easily the results. Finally, Multiple Regression models were fitted for estimating the productivity and compare the results using the original variables with using the principal components. It was found that the discrete maps showing the results from cluster analysis using the principal components presented similar patterns to those based on the cluster analysis using the original variables, suggesting that PCA reduces the dimension of the problem considerable, making it easier to interpret if the principal components provide some interpretation. When the proportion of the expl~ined variance of the first principal component is high, its interpolated map alone can already be similar to the one generated by the cluster analysis using a bigger number of principal components (normally explaining at least 90% of the variability). From the regression analysis in this research, the results suggested that the model using the principal components as independent variables presented estimated productivity further away trom the observed values, compared to the estimates given by the model using the original variables. However, we suggest further studies to investigate the behavior of the errors associated with the estimates

ASSUNTO(S)

mapeamento do solo analise multivariada

Documentos Relacionados

• Aplicação de técnicas multivariadas no estudo da divergência genética em cana-de-açúcar
• Aplicação de técnicas multivariadas em características de carcaça de ovinos da raça Morada Nova
• Aplicação de técnicas multivariadas na avaliação de linhas puras de feijão
• APLICAÇÃO DE TÉCNICAS ESTATÍSTICAS MULTIVARIADAS EM DADOS DE CERÂMICA VERMELHA PRODUZIDA NA REGIÃO CENTRAL DO RIO GRANDE DO SUL
• Avaliação da adequação de aplicação de técnicas multivariadas em estudos do comportamento do consumidor em teses e dissertações de duas instituições de ensino superior