The knowledge related to the spatial continuity of the soil moisture is essential for understanding and prediction of the hydrological processes. The geostatistical methodology has often been applied in such studies. In this context, the semi-variogram is used to model the spatial continuity of a specific variable for its mapping. This work aimed to model the spatial continuity of soil moisture in an experimental watershed located in Mantiqueira Range, MG, and then to generate map of soil moisture through the year. Exponential, spherical and gaussian semi-variogram models were fitted applying the Ordinary Least Square, Weighted Minimum Square and Maximum Likelihood methodologies, testing their adherence based on the Spatial Dependence Degree and database from Cross-Validation procedure. The exponential model presented better performance in 59% of the situations. The mean Spatial Dependence Degree was 88%, showing a strong dependence degree to the spatial continuity structure. The soil moisture mapping demonstrated behavior strongly associated to the land use and topographical conditions.
semi-variogram; cross validation; mapping; geostatistics