We use each of the described eight features to design the information systems and its implementations to organize, manage, query, and compare water resources datasets and models. The use cases guide, prioritize, and demonstrate the design per the Jim Gray’s recommendation in designing databases (Szalay and Blakeley, 2009).
We use the information model (i.e., conceptual model in relational databases) to relate the key required and optional data and metadata elements (i.e., concepts or constructs) and their relationships to satisfy the first six design features. The information model is independent of any technology or programing language (Connolly and Begg, 2010; Fulton, 2005; Lee, 1999). We satisfy conditional data queries (Feature 7) by using the Relational Database Theory (also referred to as the Relational Model) to physically implement the information model concepts into interrelated tables (Chen, 1976; Codd, 1970).
Then we implement the data model in open-source physical database system to populate it with data (Feature #8). We iteratively revised these information, logical, and physical design models dozens of times over the course of four years to satisfy design requirements, use cases, and incorporate feedback from collaborators with the most parsimonious design that demonstrates the use cases (Hey et al., 2009). The following three sub-sections elaborate on the informational, logical, and physical implementations in the WaMDaM design.
We aim for a parsimonious design that efficiently minimizes the number of data and metadata entitles needed to effectively satisfy the eight design features and answer the use case questions (Hey et al., 2009). In a relational model implementation, a database normalization technique is used in an effort to eliminate redundancy and increase design generality and parsimony in metadata (Connolly and Begg, 2010).