Optimal ordering of realizations for visualization and presentation

In geostatistical simulation, a realization represents one possible outcome of the spatial uncertainty model. Tens to hundreds of realizations are generated in order to understand response property variation. There are ways to summarize local uncertainty, but visualizing all realizations is important to understand joint uncertainty between multiple locations. There is no straightforward manner to visualize all realizations at the same time or in sequence. This paper presents a new method to sequentially display multiple geostatistical realizations. The proposed algorithm performs an ordering of the visible portion of the realizations (images), according to the distance between realizations. The concept of distance corresponds to the differences computed cell by cell for every realization pair or to the differences computed from a moving window filtering applied to each realization. To define an optimal sequence of realizations, the shortest path route through the realizations is established by a simulated annealing technique. The gradual transition between realizations is enhanced by an image morphing technique where intermediate images are introduced between the original images. The final result consists of an animation that shows the sequence of realizations and allows better understanding of the uncertainty model.