A model built with LUE can run on parallel hardware and on distributed memory systems. Reading and writing data from and to datasets happens as if such a model runs on a single CPU core using a single process, like in the old days.
But, in case of a model that is not capable of scaling over parallel hardware and distributed memory systems, a common way of handling the I/O is to use multiple processes, each performing calculations for a certain spatial extent, and use a set of datasets per process. Afterwards, these datasets are merged together again for further analysis of the model output.
In this example we are illustrating how model outputs for running a model for different spatial extents can be merged in a single LUE dataset, formatted according to the LUE data model.
A LUE dataset is a standard portable HDF5 file in which all data related to a model can be stored. In this example we are only storing a single temporal stack of rasters (a "data cube"), but the example can be expanded to store multiple of those, and static rasters, and mobile point data, for example.
