Detecção e modelagem de padrão espacial em dados binários e de contagem / Detection and modelling of space pattern in binary and counting data
AUTOR(ES)
Denise Nunes Viola
DATA DE PUBLICAÇÃO
2007
RESUMO
The spatial distribution of insects and diseases in commercial fields is important for the efficient application of pesticides. However, in the past this has not been considered in crop management recommendations, experiment planning and sampling plans. The papers presented in this thesis were motivated by two different situation, one envolving count data and the other binary data. The two models used differ in relation to the strategies of the description of the spatial dependence structure. In the the first paper the response variable is a count. In order to characterize the spatial distribution pattern of the onion thrips a survey was carried out to record the number of insects in each development phase on onion plant leaves, on different dates and sample locations, in four rural properties with neighboring farms with different infestation levels and planting methods. The Mantel randomization test was used to test for spatial correlation, and when detected this was modelled by a mixed spatial Poisson model with a geostatistic random component. This model has allowed a spatial pattern characterization as well as the production of prediction maps of susceptibility to levels of infestation in the area. In the second paper the response variable is binary. In this paper a simulation study on pseudo-likelihood estimators of autologistic parameters to verify the effect of different covariate and neighbouring structures is described, with three pest infestation levels and five different spatial correlation coefficient values. An application of the methodology is presented using a bell pepper data set. It is shown that the pseudo-likelihood method can be used when a researcher is interested in the effect of covariates, but should not be used for the estimation of the spatial correlation. A study with different percentages of missing data is made to verify the influency on parameter estimation.
ASSUNTO(S)
geostatistics simulação (estatística) poisson distribution simulation (statistics) cebola spatial distribution thrips distribuição de poisson geoestatística tripes distribuição espacial onion
