In-situ and remote sensors, as well as simulations, deliver exponentially growing multi-dimensional data volumes, such as 1-D time series, 2-D remote sensing imagery (x/y), 3-D remote sensing time series (x/y/t) and geophysics data (x/y/z), 4-D climate models (x/y/z/t), to name but a few. Data sizes frequently are extremely high, ranging into Terabyte and Petabyte volumes; with the proliferation of remote sensing and in-situ sensors, bio scanners, etc. additionally an increasingly high update rate can be observed. An emerging trend is to not just offer navigation and extraction facilities, but advanced server-side analysis and processing.

Jacobs University's raster data manager, rasdaman, offers an SQL-like multi-dimensional raster query language on the concepts of which the Web Coverage Processing Service (WCPS) emerging OGC standard grounds. The rasdaman architecture centers around tile-based raster storage, query evaluation, and optimization, using standard relational databases for storage. This works well for „google-like“ 2-D zoom & pan applications and beyond; efficiency needs to be improved, however, for advanced operations, such as on-the-fly spectral transformation.

The challenge on hand is to develop novel optimization techniques for complex queries on extremely large spatio-temporal earth science raster data sets. Tasks include: to establish application profiles across the earth sciences in close collaboration with ESSRS earth scientists; to review existing approaches in imaging, databases, super­computing, and other relevant domains and to come up with own original concepts; to implement and evaluate them in view of the previously established application profiles.

Related links:

rasdaman.eecs.jacobs-university.de, www.earthlook.org

International Collaborators:

Prof. Liu Chuang, Chinese Academy of Sciences, China

Prof. Claudia Bauzer Medeiros, University of Campinas, Brasil

various partners, Open GeoSpatial Consortium (OGC, www.opengeospatial.org)