esmvalgroup / dummydata Goto Github PK

Currently there are only two options for generating data:

1. random numbers from a uniform distribution between 0 and 1
2. constant

Option 1) is almost useless, because physical quantities are typically defined at some other scale than between 0 and 1.
Option 2) does not allow creating e.g. contour plots and it is probably also very insensitive to changes in the output, .e.g you couldn't detect a change in regridding using this.

It would probably be good to add more interesting ways of generating data.

Also, the current implementation is very inflexible. The only way to get other data in, is to either override the private method _get_variable_data or to first generate the file with some useless data, and then open it again to change the values. A more flexible approach could be to make the Model2 or Model3 object available as a context manager (just like the netCDF4.Dataset object that it inherits from).

According to the CF-convention, standard_name should be precipitation_flux instead of precipitation for variable 'pr'.

python setup.py test does not run any tests

As mentioned by @bouweandela in #3 it would be better to get the information in the Metadata class from an external source, e.g. autogenerate from the cf-convention xml document, or import from another Python package, instead of hardcoding.