limit temperature and pressure in AtmosphericState within lookup table range

[szy21]

And throw a warning when it is outside the range?

[glwagner]

I just ran into this issue as well. I believe for temperature the fix is to apply the limit fttemp = clamp(fttemp, 0, 1) after this line:

RRTMGP.jl/src/optics/gas_optics.jl

Line 86 in 90eadd1

@inbounds ftemp = (t_lay - t_ref[jtemp]) / Δ_t_ref

[szy21]

I just ran into this issue as well. I believe for temperature the fix is to apply the limit fttemp = max(0, fttemp) after this line:

RRTMGP.jl/src/optics/gas_optics.jl

Line 86 in 90eadd1

@inbounds ftemp = (t_lay - t_ref[jtemp]) / Δ_t_ref

Thanks! I thought about that. I haven't done this because I'm not sure if I want to do the clipping in ClimaAtmos or RRTMGP (there is a clipping in ClimaAtmos now).

[glwagner]

My preliminary thought is that it is more appropriately done here, because it is a property of the lookup table, which I guess could be viewed as owned by RRTMGP. For example, RRTMGP could someday support other lookup tables with different limits. Unless RRTMGP has an interface for providing alternative lookup tables?

[glwagner]

Likewise fpress = clamp(fpress, 0, 1) after

RRTMGP.jl/src/optics/gas_optics.jl

Line 106 in 90eadd1

@inbounds fpress = (ln_p_ref[jpress - 1] - log_p_lay) / Δ_ln_p_ref

[szy21]

My preliminary thought is that it is more appropriately done here, because it is a property of the lookup table, which I guess could be viewed as owned by RRTMGP. For example, RRTMGP could someday support other lookup tables with different limits. Unless RRTMGP has an interface for providing alternative lookup tables?

You can make your own lookup table for RRTMGP, but I don't think I will make one anytime soon. Let me think about it a bit more and ask for others' opinions as well.