Despite an ever-increasing number of spaceborne, airborne, and ground-based data acquisition platforms, remote sensing data are still often spatially incomplete or temporally irregular. While deterministic interpolation techniques are often used, they tend to create unrealistic spatial patterns and generally do not provide uncertainty quantification. Geostatistical simulation models are effective in generating an ensemble of realistic and equally probable realizations of an unmeasured phenomenon, allowing data uncertainty to be propagated. These models are commonly used in several fields of earth science, and in recent years, they have been applied widely to remotely sensed data. This study provides the first review of the applications of geostatistical simulation to remote sensing data. We review recent geostatistical simulation models relevant to satellite remote sensing data and discuss the characteristics and advantages of each approach. Finally, the applications of each geostatistical simulation model are categorized in different domains of natural sciences, including soil, vegetation, topography, and atmospheric science.