Representation and propagation of soil data uncertainties.: II -: Numeric attributes

The objectives of this study were to use kriging indicators to generate a representation of soil properties expressed by numeric attributes, to assess the uncertainty in estimates, and to model the uncertainty propagation generated by the fuzzy procedures of map algebra. The studied attributes were exchangeable potassium (K) and aluminum (Al) contents, sum of bases (SB), cationic exchange capacity (CEC), base saturation (V), and total sand content (TST), extracted from 222 pedologic profiles and 219 extra samples, located in Santa Catarina State, Brazil. When the attributes were expressed infertility classes, the uncertainty of Al, SB, and V increased while the uncertainty of K and CEC decreased, for intervals of confidence of 95% probability. A larger number of numeric data for K, SB, and V provided a larger uncertainty for space inference, while the uncertainty degree decreased for the largest number of numeric data of TS and CEC. There was a decrease of uncertainty when different numeric representations were integrated by fuzzy procedures of map algebra.