I've just become aware of a significant difference between how inputs are handled for the CAM3 and RRTMG radiation schemes. Both accept a list of volume mixing ratios for greenhouse gases (CO2, CH4, N2O, CFC11, CFC12 ... and some others for RRTMG).

But CAM3 is only set up to handle single scalar inputs for all those gases, imposing the well-mixed assumption, e.g. https://github.com/climlab/climlab-cam3-radiation/blob/7019e3ac28eb226fa906cb120c7515e135e6a78f/climlab_cam3_radiation/Driver.f90#L96

The climlab wrappers need to be updated to reflect this difference as raise an exception where necessary.