Cannot run single column model (u-ar989)

Hello there, I’m using the single column model (u-ar989). We tried running the standard suite, however, we found that the results do not make sense. The setting used was
















The results, for example, the boundary layer depth is constant at 36.7 m all the time,the total water content looks reasonable for the first time stamp only while it remains 0 at the rest time stamps and altitudes. I changed the rest setting to True, but the model couldn’t run. May I ask for any suggestions?

All the best,


When I ran a copy of u-ar989, job.err has messages like

?????????????????????????????? WARNING ??????????????????????????????
? Warning code: -777
? Warning from routine: SCM_TRAP_NAN
? Warning message: [TS 1] : NaN detected: dt_lwpc2 from RAD_CTL
? Warning from processor: 0
? Warning number: 23

and eventually

?!!!?!!!?!!!?!!!?!!! ERROR ???!!!?!!!?!!!?!!!???!!!
? Error code: 777
? Error from routine: SCM_TRAP_NAN
? Error message: [TS 1] : 50+ Diagnostics contain NaNs. Stopping run.
? Error from processor: 0
? Error number: 73

When developing ACCESS-CM2 I used a version of the SCM with CABLE. Testing showed a number of uninitialised variables in the SCM code. Exactly what sort of problems these cause depends on the compiler version and code that worked on raijin may fail with the newer compilers on gadi.

These errors are fixed in the MOSRS branch branches/test/martindix/vn10.6_scm_fixes
With this branch included the SCM completes without error.

See suite u-cs845 which also has a couple of other minor updates.

I’m not sure what your intended use is, but it should be possible to set up a version with the same science options using the latest version of the UM code vn13.1. This is better tested and doesn’t require any extra branches to work around bugs.

Hello Martin,
Thanks a lot ! I’ll try running suite u-cs845 when GADI back online. The main idea of using single column model is to investigate the structure of boundary layer during heatwaves in Victoria. For example, we are interested in how important is the downward long-wave radiation to the boundary layer in Melbourne given the idealised horizontal advection (force data).

All the best,

How are you generating the SCM forcing data? That might be something of broader interest.

Not yet, will do soon! Thanks!

Hello Martin,
Thanks for your help with the alternative (u-ar845) I could run. We are going to use the forcing data generated by ERA5 (e.g. temperature advection). We are wondering whether there is a standard SCM job for the latest UM version?