摘要：Modeling the structure of spatial dependence by the geostatistical approach is fun- damental for the estimation of parameters that deˉne this structure. This is also used in the interpolation of values at unsampled locations by the kriging technique. It is well known the presence of outliers in sampled data can greatly a.ect the estimation of parameters. The goal of this work is to present diagnostic techniques using the local in- °uence method when there is perturbation in the matrix of explanatory variables of the Gaussian spatial linear model useful for geostatistical analysis. Studies with experimen- tal data have shown that the presence of outliers may cause variations in construction of maps by changing the structure of the spatial dependence. The application of tech- niques of local in°uence must be part of all geostatistical analysis to ensure that the information contained in thematic maps are of better quality.
关键词：Geostatistic ￠ Maximum likelihood ￠ Spatial variability.
