Map channels to Var Channels - #181
Conversation
… ChannelIndices might have been in the wrong place.
…151 in opsinputs_varobswriter_mod.F90 when going to the opsinputs_varobswriter_fillchannumandandnumchans function.
…dd it as an explicit entry in opsinputs_varobswriter type instead of being inside the channel_offset type (which was removed)
… changes have been made.
…names. Not really what we want but want to save a copy.
…ize of the varobs array.
…nputs into feature/map_chans
|
The latest KGO output is here http://fcm1/cylc-review/view/olewis?&suite=sith_map_chans_kgo_231023&no_fuzzy_time=0&path=log/job/20210701T1200Z/glu_jopa_compare_output/09/job.out all passing apart from GPSRO which is currently failing the main nightly KGO. |
James Cotton (james-cotton)
left a comment
James Cotton (james-cotton)
left a comment
There was a problem hiding this comment.
Hi Owen, some comments and questions
…opp_ug_1dvar.pdf, change name of sizeVarObs and add comments.
James Cotton (james-cotton)
left a comment
James Cotton (james-cotton)
left a comment
There was a problem hiding this comment.
Hi Owen, coming back to this
| compress_var_channels: false | ||
| size_of_varobs_array: 4 | ||
| varChannels: 2, 4 #[5,7] | ||
| increase_chan_array: true |
There was a problem hiding this comment.
This option doesn't seem to have an impact (output is the same as the previous test). Does this option need a different test?
James Cotton (james-cotton)
left a comment
James Cotton (james-cotton)
left a comment
There was a problem hiding this comment.
Hi Owen, I am generally happy that the current behaviour is replicated as expected, but as noted there is room to improve the way this works in future.
Chris Thomas (ctgh)
left a comment
Chris Thomas (ctgh)
left a comment
There was a problem hiding this comment.
The routines opsinputs_fill_fillreal2d_norecords and opsinputs_varobswriter_fillchannumandnumchans now both contain long nested if blocks. What do you think about creating two helper functions to move some of that functionality outside of those functions? That might make the logic of the code a bit easier to follow and maintain. Just an idea - no problem if you think it won't be useful.
I think this would be useful but also at this point I think we should get this code in and ready to use for the Geostationary radiances so they can be finished for the global. |
Chris Thomas (ctgh)
left a comment
Chris Thomas (ctgh)
left a comment
There was a problem hiding this comment.
Sure, fine for me.
|
Owen Lewis (@orlewis) there is some pressure to get the geos done before xmas. Is this ready to go. Has the latest code been re-run with the sith-kgo. If not could this be set to run asap? |
|
There is KGO from 31/10/23 but there has been one change since then which was adding IMDI. http://fcm1/cylc-review/taskjobs?per_page=15&user=olewis&no_fuzzy_time=0&suite=sith_map_chans_kgo_231031&task_status=runahead&task_status=waiting&task_status=held&task_status=queued&task_status=expired&task_status=ready&task_status=submit-failed&task_status=submit-retrying&task_status=submitted&task_status=retrying&task_status=running&task_status=failed&task_status=succeeded&page=2 there are failures related to GPSRO KGO but these were consistent with the nightly tests at the time. Because it has been a month I''l rebuild and run them again today. |
|
Hi guys, I just need to let you know that David Simonin has requested that the GEOs get treated as priority which means that this PR should be seen as high priority. I've not been following the action here, but if you can wrap it up that would be greatly appreciated. Thanks. |
|
Chawn Harlow (@chawnharlow) We have two approvals so this can be merged if the KGO tests are passing. I can do the merge this afternoon if everything is OK. |
|
http://fcm1/cylc-review/view/olewis?&suite=sith_kgo_mapchans_231204&no_fuzzy_time=0&path=log/job/20210701T1200Z/glu_jopa_process_background_ssmis/01/job.out Getting this error across several obs on the JOPA run. Haven't been able to find where it is coming from this morning. |
|
|
Owen Lewis (orlewis)
added
coordinate merge
|
Good news. Michael Cooke (@mikecooke77) I have added the 'ready to merge' label if you would like to take one more look. |
8 participants
This PR adds the functionality to map channels defined in JOPA to channels in Var for the varobs writer.
As part of this change the offset channels option has been removed as this can nw be covered using the varChannels mapping option.
use_actual_channels has been replaced by a compress_var_channels. This determines whether to compress channel numbers together i.e 3, 5, 100 in just three indices or if do not compress this would have an array of size 100.
This will require changes to radiance yamls in Sith.
Requires this change to be coordinated.
https://github.com/MetOffice/sith/pull/262