Bad delayed mode data

[jklymak]

• dfo-walle652-20190718 . Suspect the vertical profiles are not being chosen properly. Also need to not interpolate across big gaps in pressure.
• https://cproof.uvic.ca/gliderdata/deployments/dfo-rosie713/dfo-rosie713-20220531/L0-gridfiles/ and https://cproof.uvic.ca/gliderdata/deployments/dfo-k999/dfo-k999-20220803/L0-gridfiles/dfo-k999-20220803_grid.nc delayed grids listed as existing, however, they do not...

[jklymak]

Actually most of the delayed mode data is not usable on Line P except for dfo-walle652-20191209?

I think we should be more careful about posting clearly broken data.

[jklymak]

• Fixed dfo-walle652-20210121
• Good: dfo-walle652-20191209
• Bad: dfo-walle652-20190718: We were doing single dives for some of this deployment, so grid file isn't taking these into account properly...
  • Fixed...
• dfo-walle652-20210902: fixed
• dfo-walle652-20200616: not broken.
• dfo-k999-20220803: not broken , but delayed mode not processed
• dfo-maria997-20220603: very short
• dfo-rosie713-20220531: delayed mode not processed.

[jklymak]

So a few things

1. delayed mode data does not need such a long filter length for the profile finding. 20 s versus 400 s is fine and gives much better profiles
2. Some of the delayed mode data was being gridded from the non-delayed mode files in a reprocessing. That lead to a huge number of gaps as the realtime data wasn't sampled as often.
3. Naming files - the old code would just make dirname/dfo-rosie713-20220531.nc. But @hvdosser had a naming convention of dirname/dfo-rosie713-20220531_delayed.nc. Not sure how that file was bing made? manual copy?
   • Added new optional argument to ncprocess.py:make_gridfiles fnamesuffix. This defaults to empty, but could be fnamesuffix='_delayed' to add that suffix.
   • OTOH, why are these not just stored in a different subdirectory? L0-timeseries->L0-timeseries-delayed seems easier to track? But I didn't want to make that change if a lot of directories were made the other way.

[hvdosser]

I'm working through all the delayed mode data for all missions, and would appreciate input on the following:

• How to avoid interpolating across big gaps in pressure? This functionality seems to be buried inside the DBD reader code, and the issue will also happen with the new interpolation in the SeaExplorer code. Clearly it's an issue for ECOpuck data. I don't have time to figure out a fix, so I'm running the delayed-mode processing as-is, but I know that's not ideal.

• How can I prevent links from showing up on the website for files that don't exist (e.g., the link to the delayed mode data, if those files haven't been created yet). This is well beyond my ability to troubleshoot for the website.

I'm currently waiting on James to upload the delayed mode data for the two 2022 Line P missions, and will process it ASAP when it shows up (hopefully today or tomorrow - he's located it and I'm just waiting on the upload).