Granularity

Following the data model, in principle the granularity is defined as one to many data sets per event. The definition of the granularity depends on the scientific data type and the size of one entity (data set). The size should allow a download over the Internet in an acceptable time. A set might consist of e.g. all parameters belonging to one analysis or method or category.

Data entities are structered in a hierarchical tree. Starting from a
 * 1) single analytical value, data are mostly organized in sequences with an (increasing) geocode in space (e.g. depth) or relative/absolute time. Such a sequence of one to many data points forms a
 * 2) data series, described by one parameter with relations to event, method, PI and an optional comment. One to several data series are forming a
 * 3) data set which is the major data granule in the archive; data are imported in data sets and are provided on the Internet in a similar configuration. Data sets have relations to authors (staff) and references and are citable. One to many data sets may be grouped to a
 * 4) parent set to form an overall entity. The parent set has an official bibliographic citation and is added to the TIB library catalog GetInfo. An abstract is mandatory in parents.

Examples of typical data set granules are:
 * Physical oceanography from one cruise
 * Grain size distribution of one sediment core
 * Photos from one sea-bed ROV station
 * Mineralogy of surface sediments from several samples
 * Isotopes measured in a biological object
 * Oxygen lander profile through the water/sediment interface
 * Time series with oceanographic measurements over one year
 * Seismic profile