Hi NCAS,
I am currently setting up a GC5-central piControl suite on MONSOON3 (u-dv344), as a preparation for later paleoclimate simulations.
During the coupled run, I observed that NaN in UKCA tracers suddenly emerge within a single timestep.
UKCA_TRACERS_COPY_TO_UM: Copying tracers out.
tracer: 1 1 H2O2 =
0.1572603885168801E-10 0.1339563825575584E-10 NaN
tracer: 2 2 DMS =
0.6820798255761047-155 0.6736828766935020-155 NaN
tracer: 3 3 SO2 =
0.3336495792351384-154 0.3320419915821740-154 NaN
tracer: 4 4 H2SO4 =
0.2205347896000025E-38 0.2205347895999976E-38 NaN
tracer: 5 5 DMSO =
0.8582867651132799-155 0.8469225087207077-155 NaN
tracer: 6 6 Monoterp =
0.1416995633673715-154 0.1398943792319969-154 NaN
tracer: 7 7 Sec_Org =
0.2763652209483305-154 0.2728252194723871-154 NaN
tracer: 8 8 Nuc_SOL_N =
0.0000000000000000E+00 0.0000000000000000E+00 NaN
tracer: 9 9 Nuc_SOL_SU =
0.0000000000000000E+00 0.0000000000000000E+00 NaN
tracer: 10 10 Ait_SOL_N =
0.0000000000000000E+00 0.0000000000000000E+00 NaN
tracer: 11 11 Ait_SOL_SU =
0.0000000000000000E+00 0.0000000000000000E+00 NaN
tracer: 12 12 Ait_SOL_BC =
0.0000000000000000E+00 0.0000000000000000E+00 NaN
tracer: 13 13 Ait_SOL_OM =
0.0000000000000000E+00 0.0000000000000000E+00 NaN
tracer: 14 14 Acc_SOL_N =
0.0000000000000000E+00 0.0000000000000000E+00 NaN
tracer: 15 15 Acc_SOL_SU =
0.0000000000000000E+00 0.0000000000000000E+00 NaN
tracer: 16 16 Acc_SOL_BC =
0.0000000000000000E+00 0.0000000000000000E+00 NaN
Before this crash, the model crashed with very high horizontal wind speeds in the upper levels, which I believe may have been related to a developing numerical instability near the poles. I have applied vertical_damping, increasing eg_vert_damp_coeff up to 0.2, which seems to have relieved the upper-level wind speed explosions.
However, despite this adjustment, the tracers still go NaN, causing crash.
Any advice on diagnosing or preventing this NaN behavior would be greatly appreciated.
Many thanks in advance for your help.
Best regards,
Zikun Ren