Datacubes: A Discrete Global Grid Systems Perspective

Datacubes are increasingly being implemented to manage big data workflows efficiently, particularly those for processing geospatial data. However, there is confusion in both the definition of the term "datacube" and the choices for how it is implemented. This and the conventional approach to managing spatial data (i.e., in map-projected data sets) have led to a restricted set of datacube implementations that are each tightly coupled to the spatial constraints of the data and how they are stored on disc - resulting in barriers to interoperability, particularly on global scales. This article discusses options and how it is possible to implement a datacube based on discrete global grid systems, while using the same topologies as conventional datacubes. These provide a flexible spatial data infrastructure that leverages the same topological advantages as conventional geospatial datacubes, while reducing barriers to data interoperability of both raster and vector data and providing additional functionality. Also, they potentially provide a very efficient approach to connecting to big data sources in order to extract datasets on demand prior to proceeding to multi-level intelligent big data processing, mining, machine learning, and visualizations.