This section explores the process of building metadata bases compliant with the models presented above and lists the problems faced by a metadata producer. Examples refer to the ISO19115 standard.
It is not an easy task to understand the meaning of the metadata standards and how they should be used. For instance, the element MD_Metadata.hierarchyLevel has several possible values, among which are “series” and “model,” but the difference between both is not always obvious; are the specifications of a data product a model or a series? For instance also, the element MD_Metadata.parentIdentifier is to refer to metadata of which this metadata is a subset. It is not easy to understand why its maximum occurrence is 1.
Besides, the metadata producer often has to specialize or extend the standard. For instance, in the ISO19115 standard, the definition of the browse graphic element is very fuzzy and needs specializing for the implementation. For instance also, the standards need extending elements for grid data and for services.
Discovery metadata will be queried in an iterative process to select resources of possible relevance to the user need. Defining a storage model that is adapted to the complex metadata model and that supports iterative queries is a difficult task. Most applications nowadays mix relational database techniques and XML storage.
Last but not least, acquiring metadata and maintaining a metadata base are difficult. Metadata information often exists in a nonstandard and possibly nonformal format. It can be, for instance, a text file storing the spatial extents of raster maps in a series. The person working on this file may change its structure over the year. A similar file may be stored on the next PC if his colleague is working on the same product. There may be no reference on either of these files (author, date, ...). Moreover, people holding such files may be unwilling to share “their” files and will probably be reluctant to commit themselves to maintaining them.
In the United States, it is mandatory for geographical data producers to document their data with the FGDC CSDGM. There is no such obligation in Europe. Yet, other initiatives favor the implementation of metadata such as the Spatial Data Infrastructures (SDI). SDI are presented in the last chapters of the present book. Regarding the production of metadata bases, SDI are useful to share the implementation efforts enumerated above, as well as to obtain a homogeneous implementation across various data producers.