Error in CLUBB from NaN values in mean wind on component levels (wm_zm and other vars)

[jacob-stu-allen]

Hello!

I'm performing some experiments which limit the surface momentum flux felt by the tropical Pacific ocean (${CASE}/build/cmake-bld/core_ocean/shared/mpas_ocn_surface_bulk_forcing.f90). In my experiments, zonal wind stress values which are less than a specific threshold are held to that threshold. This threshold is a Tau anomaly (difference from daily climatology in the piControl) of -0.050 n/m^2 (on daily time-scales an easterly value this strong in the equatorial pacific is about 2.5 standard deviations from the mean ~ pretty rare).

After 59 years, one of my simulations has failed with the following Error message in the e3sm.log: "Error in advance_xp2_xpyp".

It then prints out the input variables used in the EAM module (CLUBB), advance_xp2_xpyp_module.F90, and the following variables all contain NaN values, resulting in the break: wm_zm, wprtp, thlm and wpthlp (see attached log).

It additionally prints out tau_zm values all equaling 3600.00.

I'm hoping someone more familiar with CLUBB can help me understand why limiting strong/rare easterly wind events in the tropical Pacific mpas_ocn_surface_bulk_forcing.f90 (I'm altering values of the variable, windStressZonal in MPASO) is resulting in NaN values for the momentum level variables in CLUBB and why this is happening only sporadically. I can imagine why this may happen physically, but I'm not at all familiar with the turbulence calculations/considerations in CLUBB and am hoping to find a way to continue this simulation
in a physically sensible way.

Thanks in advance,
Jacob

logs.zip

[jacob-stu-allen]

This has been fixed simply by decreasing the CLUBB timestep- must have been a CFL problem.