I agree. A nested data format similar to hdf5 is inevitable if we want to use it for all IS2 products. I think a major motivation for people to use pandas and xarray are their easy-to-use functions for data handling. Instead of converting hdf5 data to other formats to take advantage of the xarray/pandas functionality, can we build on the hdf5 format, and provide some popular functionality that are most frequently used in pandas and xarray? For example, how about we inherit the hdf5 class and add methods such as sel , loc , where , concat , etc., to provide easier data subset and aggregation. Just as Fernando mentioned, users should still be able to use the underlying data directly by calling the values method.Xarray is designed to handle gridded data and pandas is designed to handle tabular data. For nested data such as hdf5, there is still not an advanced package to provide similar high-level methods yet (are there?). Do you guys think this is something we should pursue or is it too far off-focus from icepyx?