SUMMARY
In Classical Geostatistics or Design Based Geostatistics, there is a great need for research that creates and/or explores methods of geospatial data sampling. In addition to this complex subject, some papers present solutions that use theoretical and practical mechanisms from different areas of scientific knowledge addressing specific demands of researches in the field. The purpose of this paper is to apply the Electrical Signal Information Theory, especially considering the Nyquist Rate Theorem, to determine an optimal size for georeferenced samples using a regular quadratic grid based on the spatial dependence Gaussian model. We expect to achieve a necessary sampling density to rebuild population maps of variables in which the following regularity conditions required in geostatistics were certified: 1st and 2nd order stationarity and/or stationarity of the variogram, absence of outliers and trends, and isotropic variogram. As a result, we can state that from the data set used, the greatest distance between points of the regular quadratic grid is nearly 30% of the practical range observed on the variogram of the first experimental sample.