Periodically, we see requests for support for scale+offset representation of variables. It belatedly occurs to me that this is a form of compression. Since netcdf-4 will support hdf5 dynamically loaded compression filters, we could use that mechanism to support scale+offset representations. Is there a description of the proper way to convert a number (float/double?) to scale+offset and the inverse? There are CF conventions for this, I believe. =Dennis Heimbigner Unidata
netcdfgroup
archives: