Can we isolate some biases in a deep convective case to non-convective parameterizations? For example, we could generate idealized deep convective tendencies associated with one of the cases (TWPICE) and use the suite simulator to run suites that don't have deep convection (RAP, HRRR) for the same case.
From @ligiabernardet