Modeling Spatial Uncertainty for Locally Uncertain Data

The work discusses methods dealing with "soft" input data, where local uncertainty is represented by a variance. Modifications of ordinary kriging and sequential direct stochastic simulations based on such data are applied to a real hydrogeological case study and a synthetic environmental contamination study. The modification performed on direct simulation approach does not require any data transformation assumptions. The method is compared with Bayesian Maximum Entropy (BME) based stochastic simulations, which provide an alternative way of integrating "soft" information.