bug in micro_mg_cam.F90 (MG1) about e3sm HOT 4 CLOSED

The language about bug fix branches has been changed. Your second version that started from master is the one you should push.

However I don't think there should be a switch to turn it off. Bug fixes should be "on" and the changes to answers accepted unless there is a scientific reason to keep the buggy version around.

from e3sm.

Thanks for the information! I will forward it to our integrators.

from e3sm.

Can someone describe in words what the bug does to the climate or to some output field?

from e3sm.

Below is my understanding of the bug and its fix.

With the bug, the convective transport of cloud droplet and ice crystal NUMBER concentrations will be handled by the second call to the convective transport subroutine (convtran, in zm_conv_tend_2). With the bug fix, it will be handled by the first call to convtran (in zm_conv_tend). The transport of cloud droplet and ice crystal MASS concentrations is handled by the first call to convtran.

zm_conv_tend_2 is called after the deep convection (convect_deep_tend), shallow convection (convect_shallow_tend), and stratiform cloud macro-/micro- physics treatments (stratiform_tend). Because the droplet and ice crystal number concentrations will be changed in stratiform_tend, doing the transport for cloud droplet and ice crystal numbers at a different location will make a difference in the simulated number and mass concentrations of hydrometeors.

In a five-day simulation, the instantaneous relative difference of NUMLIQ and NUMICE can be very large (>50% is not uncommon) and can be associated with a large absolute difference. The long-term impact has not been investigated using the ACME code yet.

from e3sm.