Crash in downscaling suite

Unified Model
1

Hi Ros,

This is Ben path

/home/n02/n02/bharvey/cylc-run/u-dq099_cm655_dd441_19580216-P_uk-02

And my own

/home/n02/n02/egentile/cylc-run/u-dq099_cm655_dd441_19580216-P_naew-02

We got a similar crash despite us running on two different downscaling domain, so maybe an issue with the driving boundary conditions not sure, any help would be greatly appreciate from us both.

I just add that Ben carried out the following tests that didn’t help, just quoting from his email

  • Reduce timestep to 30s (set in opt/rose-suite-CANARI_uk-02.conf:TIMESTEPS_PER_DAY=2880)
  • Turn on extra diagnostic output (set in app/lm_um/rose-app.conf:PRINT_STATUS=PrStatus_Diag)
  • Change number of processors (set in opt/rose-suite-CANARI_uk-02.conf:NPROCX=12 [also NPROCY])
  • Examine my vegfrac ancil file for obvious NaNs etc (/work/n02/n02/shared/CANARI/ancils-regional/climate/uk/0p02/time_varying_vegfrac/qrparm.veg.frac)

Bests,
Emanuele

2

Emanuele

Please allow us read access to your nvme space.

The fact that both uk and naew domains fail for this storm does point to bad start data, but which bit of data is hard to establish. Can you skip this storm for now?

Grenville

3

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.