You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Many modeling groups "block" 3-D and 2-D variables differently.
Ex: 2-D PSL may have ntime=1460 but 3-D Z may have ntime = 360.
Currently, to track on PSL, files need to be "lined up" using NCO (ncrcat Z files together or split PSL using ncks) or other operators.
In an ideal world, one could pass a completely unordered list of files into TE. TE could then determine a "full time series" by ordering all files with the tracking var (i.e., PSL) and build a lookup table to match times in other files with different vars based on the required processing/outputs (ex: Z, U10). Perhaps this can be done on a master task when DetectNodes is first called but IANASE.
Probably logical addition: a quick check could also be implemented that either exits if a matching time isn't found for each required variable passed in or the "full time series" can be truncated to only include overlapping times between PSL, Z, U10. The former seems safer from a UI perspective.
The text was updated successfully, but these errors were encountered:
Latest version of DetectNodes indexes on time, so as part of a single input you can include files where overlap between time indices is only partial. The first file specified on the line is used for master indexing.
I have to wonder how often it'll be the case that there is climate data that isn't organized using proper subsets.
Many modeling groups "block" 3-D and 2-D variables differently.
Ex: 2-D PSL may have ntime=1460 but 3-D Z may have ntime = 360.
Currently, to track on PSL, files need to be "lined up" using NCO (ncrcat Z files together or split PSL using ncks) or other operators.
In an ideal world, one could pass a completely unordered list of files into TE. TE could then determine a "full time series" by ordering all files with the tracking var (i.e., PSL) and build a lookup table to match times in other files with different vars based on the required processing/outputs (ex: Z, U10). Perhaps this can be done on a master task when DetectNodes is first called but IANASE.
Probably logical addition: a quick check could also be implemented that either exits if a matching time isn't found for each required variable passed in or the "full time series" can be truncated to only include overlapping times between PSL, Z, U10. The former seems safer from a UI perspective.
The text was updated successfully, but these errors were encountered: