This are data representing a particular feature of the real world or the result of a model run. For example, the hydrology data contains a rasterized grid where each cell represents the elevation of the water suface at that point. This type of information is what is ment by actual data.
This is information about the character of a data set. It includes things such as the spatial extent of the data, the resolution at which the data is stored, or the period of time represented by the data. It also includes information about the configuration of the data file. What kind of data structures are used: array, linked lists, etc... What type of fundemental types are involved: doubles ,floats, ints, etc. In the case of the fundemental types, information detailing how many bytes are used to express these type in recorded. The metadata may also include information about who created the data file, how they data was created, what messures have been taken to assure the accuracy of the data, what errors are involved, when the data was created, and what errors are known to exsist or may exsist in the data. At this point in time the modeling components of ATLSS use a subset of these types of information. For a detailed description see the Metadata documentation for the ATLSS project.