@cgrdn have you ever run into the N_HISTORY2 dimension before? I'm curious to know if it's a pattern whose use will be expanded or if it's an error of some kind that I can ignore.
Not something I have run into before. Since these look like a lot of BODC files, maybe @matdon17 could comment.
Here's my best guess: in a lot of Argo files, N_HISTORY is the unlimited dimension. In this traj file (and I'm assuming others, though perhaps I am wrong), N_MEASUREMENT is the unlimited dimension, which makes sense since traj files get added to as profiles come in. Since there can only be one unlimited dimension per netCDF file, adding to N_HISTORY could be a bit of a pain and creating a new dimension N_HISTORY2 could be a workaround.
But that's just a bit of a shot in the dark. Hopefully Matt can elaborate.
As a quick reply - these are almost certainly still v2 rather than v3 Argo NetCDFs traj files, so I wouldn't use these as an exemplar for whatever you are doing. We still have much work to do on his front, and we are unlikely to get to it anytime soon.
I wouldn't be surprised though if multiple history steps were being added as part of DMQC. I note though that at least some of these are very old floats (190083 was deployed in 2001!) and it will take me some time to provide any further insight.
Thanks to you both for this! I should have checked the actual values before flagging you...they're all jibberish values in the HISTORY_QCTEST variable as far as I can tell Thanks to @matdon17 I checked the version of the files and they're all 2.2. I also checked the deployment date and they're all pre-2013 except one. I appreciate the offer for further insight but I think there's enough here to suggest that I don't need to provide access to the N_HISTORY2 dimension here.
These files seem to have a
N_HISTORY2dimension of unknown utility:Created on 2021-04-13 by the reprex package (v2.0.0)
@cgrdn have you ever run into the N_HISTORY2 dimension before? I'm curious to know if it's a pattern whose use will be expanded or if it's an error of some kind that I can ignore.
Not something I have run into before. Since these look like a lot of BODC files, maybe @matdon17 could comment.
Here's my best guess: in a lot of Argo files, N_HISTORY is the unlimited dimension. In this traj file (and I'm assuming others, though perhaps I am wrong), N_MEASUREMENT is the unlimited dimension, which makes sense since traj files get added to as profiles come in. Since there can only be one unlimited dimension per netCDF file, adding to N_HISTORY could be a bit of a pain and creating a new dimension N_HISTORY2 could be a workaround.
But that's just a bit of a shot in the dark. Hopefully Matt can elaborate.
As a quick reply - these are almost certainly still v2 rather than v3 Argo NetCDFs traj files, so I wouldn't use these as an exemplar for whatever you are doing. We still have much work to do on his front, and we are unlikely to get to it anytime soon.
I wouldn't be surprised though if multiple history steps were being added as part of DMQC. I note though that at least some of these are very old floats (190083 was deployed in 2001!) and it will take me some time to provide any further insight.
Thanks to you both for this! I should have checked the actual values before flagging you...they're all jibberish values in the HISTORY_QCTEST variable as far as I can tell Thanks to @matdon17 I checked the version of the files and they're all 2.2. I also checked the deployment date and they're all pre-2013 except one. I appreciate the offer for further insight but I think there's enough here to suggest that I don't need to provide access to the N_HISTORY2 dimension here.