Increased size of Seasonal forecast subdaily data on pressure levels

Datasets and Usage C3S - Datasets and Usage
1

Dear all,

I have downloaded Seasonal forecast subdaily data on pressure levels for a single year, on a specific pressure level and for a single initialization month keeping all the lead time hours. Compared to the previous version from the old c3s, the size has nearly doubled (from 3.6 GB to 6.2 GB on average). Looking at the variables, it looks like that each dataset contains much more information about the data. I was wondering if anyone had noticed it and if there is a way to reduce the size before the download.

Thanks in advance!

2

Dear @Giacomo_Giuliani

I would say this behaviour is unexpected, but it would be worth trying to get to the bottom of it.
In order to do so, it would be helpful to clarify a couple of items of your specific requests:

  • Are you downloading the data in GRIB or in netCDF format?
  • If the latter, as there have been changes in the conversion tool used by the CDS (GRIB to netCDF conversion on new CDS and ADS systems - Copernicus Knowledge Base - ECMWF Confluence Wiki), even though differences this big shouldnā€™t be expected, I would say it might happen. In this case (netCDF retrievals) have you tried downloading specifying ā€œnetcdf_legacyā€ as the format? (this should be equivalent to what you had in the legacy CDS)
  • Finally, providing a sample request (API request code) could be helpful to investigate further any other possible causes.

I hope that sounds useful!

3

Thank you for your detailed answer. Investigating in depth, I have also noted that in the newer version the type of the variables have changed. For example, lon and lat are now double, while previously they were float. Similar changes have occurred for the other variables. I think this might be the reason behind the increased size. Moreover, there are many more metadata, i.e. information about each variable, especially for physical variables. For instance, I downloaded the wind component v and a lot of v:GRIB_ info appear, even though I selected the netcdf format.

By the way, I will try to use ā€œnetcdf_legacyā€ as the format to download the data and let you know.

Edit: using the ā€œnetcdf_legacyā€ format, the data size is the same as the old one.

Thank you so much again for your time and the answer!

1 Like