Evaluating the perception of uncertainty in alternative visualization strategies

Although a number of methods have been proposed for representing uncertainty in spatial information, there has been relatively little testing of how users interpret uncertainty from those methods. Using results of a fuzzy classification of satellite imagery, this study looks at users' perceptions of a number of different methods for visualization of uncertainty. Both relatively expert and novice users are highly successful at determining classification uncertainty among pixels when shown in grey-scale images and histograms, as well as having moderate success with random animations. Viewers considered that although random animation is a good method of showing overall classification uncertainty, grey-scale images are a better way to extract specific information. Serial animation was not widely appreciated. In general, the accuracy of answers to specific questions was greatly improved when linked views and ancillary graphic information were available. All the methods examined in this study, in short, were relatively successful, with the exception of serial animation.