Opened 6 years ago

Closed 6 years ago

#1052 closed help (fixed)

Vertical levels in idealised mode output files

Reported by: rmbyoung Owned by: um_support
Component: UM Model Keywords:
Cc: Platform: Other
UM Version: 8.2

Description

I am running in idealised mode and I redefine the vertical levels at the start of the run. The relevant namelist parameters are

grid_number = 1 ! Regular grid spacing
height_domain = 20000.0 ! Top of domain / m
first_theta_height = 10.0 ! (d) Height of lowest theta level

The input dump I use has the top of the domain at 80000km (it's the standard start_dump supplied with the ExtUM). At the end of a test run I save both an instantaneous dump and a PP file containing the U velocity field after one day at the same physical points. The U velocity fields should be identical.

The U velocity values themselves *are* identical at every point. But the vertical levels are different, both when viewed in xconv and in the headers for the two files. In the headers (produced with pumf):

(1) The level dependent constants are the same, i.e. the headers in the PP file contain the Z level values (parameters 5 and 7) that were in the input dump. Shouldn't these be the redefined levels?

(2) The lookup tables in the dump output contain the original (i.e. incorrect) vertical levels (words 46, 52, and 53), but the lookup tables in the PP output contain the correct (i.e. redefined) levels. Shouldn't the levels in the dump file be the redefined levels too?

(3) Finally (and perhaps most strangely), in the REAL HEADER, word 16 in the dump output contains the height of the top level in the redefined vertical grid (i.e. 20000m), while in the PP output word 16 contains the height of the top level in the original vertical grid (i.e. 80000km). If anything I thought this would be the other way round.

Are these discrepancies intentional? I thought that by defining the vertical grid all the relevant parameters would be overwritten throughout the code, but it seems that they are not.

Roland

Change History (4)

comment:1 Changed 6 years ago by grenville

Hi Roland

I have forwarded your query to a UM-Collaboration newsgroup. We are finding it difficult to advise you since we can not see your job set up, data files etc. and can not run an equivalent job for debugging.

Could you set up the job through the umui - that way we would have a better chance of helping.

Grenville

comment:2 Changed 6 years ago by rmbyoung

[From um-collab newsgroups]

Are the parameters in the RUN_IDEAL namelist rather than the VERTICAL namelist? If you did change the VERTICAL namelists did you run the reconfiguration before the UM after redefining the vertical level information in the VERTICAL namelist?

As for the discrepancies. The idealised model has some very special handling of the settings and in a way independent of the non-idealised setup.

In general a new approach to setting up the idealised models might be better since if
the idealised and the non-idealised usage drifts too far apart it then becomes difficult to maintain. Something to think about in the future.

Just in case you are not aware there is a um-collab.um.general newsgroup for more specific UM questions like these.

Tom

Last edited 6 years ago by rmbyoung (previous) (diff)

comment:3 Changed 6 years ago by rmbyoung

This has fixed the problem.

Once I define the vertical levels and model top in the VERTICAL namelist rather than the IDEALISE namelist, reconfigure, and set grid_number=10 in the IDEALISE namelist (to use the levels from the input dump), then all the discrepancies disappear.

Thanks!

Last edited 6 years ago by rmbyoung (previous) (diff)

comment:4 Changed 6 years ago by grenville

  • Resolution set to fixed
  • Status changed from new to closed
Note: See TracTickets for help on using tickets.