Kriging is a interpolation technique with a minimum error variance widely used as a reliable estimation method. Among the used forms, ordinary kriging is the most usual and has features as recognition of anisotropies, unbiasedness and the computation of the minimized estimation variance also know as the kriging variance. Such features are not all avaible in other interpolation methods, as polynomial triangulation, inverse of weighted distance, minimum curvature, polynomial regression, etc. That means, kriging provides a set of statistical tools for incorporating the spatial coordinates of the observation in data processing, allowing for description and modeling of spatial patterns, prediction at unsampled locations, and assessment of the uncertainty attached to these predictions.

But for the appropriate application of kriging spatial data information must be known. In the geostatistical literature, spatial patterns are usually described in terms of correlation and/or covariance between observation as a function of the separation distance. Semivariogram provides tools for detecting and quantifying the major scales of spatial variability.

The main purpose of this paper is call attention when using the kriging technique without previous evaluations of the spatial dependence of the data and the chosen example was altitude data. Altitude models are usefully in geological and geomorphological for evaluate spatial distribution of topographical potential erosion as well as deposition. Besides that, kriging can be used for modelling orientation slopes of fluvial units and also direction and capacity of sediment transport.