Joel Jaeschke

Answering my own question in part: by changing the chunksize of the time domain to 1, everything suddenly works as expected. My guess is that since the chunksize of 31...

Hey @martindurant Thanks a lot for your input! I guess now that I am thinking about it, the limitation does make sense to some extent as that was likely not...

I played around with this a little out of curiosity. I think it is possible to expand on the idea of @thodson-usgs a bit, considering how the quantization is done....

It does make sense what you are suggesting and have played around with this some more, using some ERA5 wind speed data. When running `get_keepbits` on the dequantized field, I...