The input data sets are generated before the tests are executed. They are based on simple equations to simulate temperatures and precipitation for different locations. The bias correction methods are then applied to those to to then validate if the methods improved the data.
Additionally some input data is saved as .zarr, so the dask compatibility can be tested as well.