The objectives of this work were to generate a representation of soil properties expressed by categorical attributes by kriging indicator; to assess the uncertainty in estimates, and to model the uncertainty propagation of map algebra by means of boolean procedures. The studied attributes were potassium (K) and aluminum (Al) exchangeable contents, sum of bases (SB), cationic exchange capacity (CEC), base saturation (V), texture (Tx) and relief classes (RC), effective soil depth, internal drainage, and stoniness and/or rockiness, extracted from 222 pedologic profiles and 219 extra samples from soils of the State of Santa Catarina, Brazil. The uncertainties evidenced the spatial variability of the data related to the samples' origin and attribute behavior. Attributes SB, V, K, and RC presented higher uncertainties than Tx and CEC, and there was an increase of uncertainty when categorical representations were integrated.
kriging indicator; spatial analysis; geostatistic; GIS
