On the modeling of soil fertility by stochastic simulation with uncertainty treatments

Abstract

This work described a methodology of fertility modeling that integrates chemical properties of the soil through the Monte Carlo method. The chemical attributes are spatialized, along with their estimation uncertainties, by a geostatistical procedure of random simulation. Uncertainties of the chemical properties are propagated to the resulting fertility model allowing the generation of fertility maps conditioned to preset risk levels. The method presented here is illustrated by a case study of fertility for soy culture, at the State of Santa Catarina, Brazil, considering the following chemical properties of the soil: exchangeable aluminum, potassium and cation exchange capacity.