Figure 14.19: Ordinary kriging (and most other estimation methods) tends to average or smooth the data to achieve a best guess of reality. Bayesian kriging with conditional simulation provides a means for modeling the variability observed in nature, while still honoring the field data. Conditional simulation does not produce a best estimate of reality, but it yields equiprobable models with characteristics similar to reality. Unfortunately, when only hard data is used, the range of values for the conditional simulation is limited to the range of the hard data. Because it is not reasonable to expect the exploration program to identify the full data range, by incorporating soft data, these bounds can be exceeded. When multiple simulations are made their average values will approximate the smoothed, best fit curve.