HDF5 files are organized in a directory-like structure. Files created by the XMDF library are not required to have a specific organization. Figure 6 is a screenshot of an XMDF file created by SMS visualized inside NCSA’s HDFView browser. An attribute is written to files written by XMDF within each group that identifies the type of data stored in the group. Generic groups have no specific type of data associated with them and are used to organize the file. Special group types include mesh, grid, multi-datasets, scalar dataset, vector dataset, and property groups. Multi-datasets groups are special groups that hold datasets that all must belong to a corresponding spatial data object (mesh or grid). The other group types should be self-explanatory. Models relying upon XMS to create XMDF files for geometry input should not expect a specific file organization because the file format may change without warning. Instead, the model developer should work with the EMRL to have the XMS package write out paths to required objects to an external file or use command line arguments. For models without a specific XMS interface, the necessary information can be determined for an individual file using an HDF5 browser.