Global estimation is the first and most important step in evaluating a coal deposit. Assessing accuracy for estimates constitutes the most difficult stage of this procedure. In this study, I introduce the ranked tile resampling method to assess uncertainty for the global means. It is based on producing a simulated version of the original field. The simulation is achieved by dividing the observation field into tiles, resampling them at random with replacement, and joining them together in a ranked order The method is applied to thickness and ash yield of a simulated coal deposit. The simulation results show that the ranking significantly improves the correlation structure of the resampled field, especially over short tile sizes.