Reducing the information load in map animations as a tool for exploratory analysis

This dissertation investigates the information load that animated maps cause to their viewers, and presents two novel visualisation methods to support the exploratory visual analysis of the animations. Information load consists of the information content of the map and its presentation. The number of objects and their attributes are the unavoidable content, but the visualisation of the objects, the background map, and display settings of an animation have an effect on the information load and experienced complexity of the presentation. Information overload causes a load on human beings' cognitive capacity and working memory. Our brain is capable of processing only a few pieces of information at a time, and if new information is provided faster than it can be moved into the long-term memory, some of the information is lost. Luckily, the information can be segmented into meaningful wholes, mental chunks, to increase the processing capacity. The aim was to find out, what factors increase the information load in animated maps, how this load could be reduced, and how the formation of mental chunks of an animation can be supported. The most important factors affecting the information load were recognised to be the temporal extent of the dataset, geometry complexity and illogical movement of the data presented, and the user's task. When the temporal extent of the information grew too big for the users, the amount of information was automatically reduced from another aspect. The combination of two datasets, that was designed based on previous knowledge about combination colours, was experienced as being too complex because of the graininess of the dataset and the unexpected behaviour of the phenomena. The novel visualisation methods presented, temporal equal density transformation and temporal classification, aim to reduce the information load without any loss of the information content; a feature that is important in exploratory analysis where the task is unknown. In the user tests, they were proved to be particularly useful to reveal such spatio-temporal patterns that would have been left unnoticed with traditional animations. They seem to be able to reduce the information load by spreading the information flow equally over the whole period and by segmenting the animation into easily adoptable chunks. As a conclusion, it can be argued that the designing of map animation sufficient for exploratory analysis should take into account the characteristics of both the spatial and temporal structureof the data, since a task-based visualisation is not possible to define in exploratory use.