Supporting the Construction and Use of Spatio-Temporal Domains InScientific Databases

Report ID: 
Terrence R. Smith, Jianwen Su, and Michael F. Worboys
1993-07-01 05:00:00


A modeling and database system (MDBS) provides computational support thatfacilitates the iterative development of scientific models in data-intensiveand computationally intensive applications in various areas of scientificapplication. We describe and exemplify core elements of an MDBS that is basedboth upon a simple characterization of scientific activity and upon the resultsof joint research with a group of EOS earth scientists. The MDBS that wedescribe is focused in particular on the representation and manipulation oflarge numbers of complex, spatio-temporal entities. The main contributions ofthe current paper are an advanced and comprehensive data model and a high-leveldatabase and modeling language (MDBL). The data model is based on a simplecharacterization of scientific activity that captures the essence of scientificmodeling and database activities in terms of four fundamental sets ofactivities. These activities are defined in relation to large lattices ofrepresentational domains (R-domains) that are employed to organize scientificknowledge, data and algorithms. The data model separates the specification andrepresentation of R-domains of elements, and allows multiple representationsthat are explicitly controlled by the user. Datasets and transformations areorganized in terms of an extensible lattice of abstract and concreteR-domains. In particular, the data model allows the easy definition andmanipulation of large numbers of domains of spatio-temporal elements. Wepresent many examples of such domains from our applications. MDBL is ahigh-level, largely declarative language that is also based on ourcharacterization of science and provides easy expression of the four classes ofscientific activities in relation to R-domains. MDBL permits the easyintegration of modules that support scientific modeling and databaseactivities. We are currently applying the data model and languages to severalEOS earth-science applications.