Sedecimtrees: An efficient data structure for interactive visualization of time-varying datasets

This research proposal outlines the development of a novel data structure for large time-varying datasets called the sedecimtree, a hierarchical subdivision scheme for 4D data related to quadtrees and octrees.

The goal of the sedecimtree is to provide a common, non-uniform subdivision of the spatial and temporal components for datasets.

This enables a more efficient data retrieval in interactive data visualization and analysis frameworks by providing a streamlined and uniform data access pattern, utilizing early data reduction.

The data structure is designed to handle large, distributed, out-of-core datasets which can be served from different physical locations over the internet.

Utilizing available rendering parameters, such as the camera location and transfer function, in the data retrieval step, the pressure on the transfer is reduced, thus reducing the impact of this bottleneck and enabling support for the interactive analysis of larger datasets.

The non-uniform subdivision in this 4D space will be guided through a novel concept called the Space-Time Signature graph, which provides an empirical connection between the spatial and temporal scales of a datasets, and a velocity-based interpolation method that is used as a compression method to reduce the amount of data required during the visualization step.

Utilizing this insight in the data processing provides further data transfer optimization opportunities and thus will enable the interactive analysis of larger datasets.