In #245 we implemented a mechanism to cache the intermediate, "case study" subsets for EWB analyses. By default, these write to Zarr archives. However, in some cases the data may be very small and it would be easier to write as single NetCDF outputs to facilitate sharing with other researchers. We propose extending the user configuration to allow users to specify between NetCDF or Zarr for writing these cached outputs.