RRTGMP complains that tsfc has values outside range

[brianpm]

What happened?

As reported by @cecilehannay and Jim Edwards, in coupled runs RRTMGP sometimes complains that:

radiation_tend: ERROR: kdist_lw%gas_optics: gas_optics(): array tsfc has values
outside range

A workaround was applied by @brian-eaton and is in cam6_4_044, but this just gets us through radiation, the surface temperature is still going to be outside the realistic range.

Then range that is checked by RRTMGP comes from its data files, and is 160 - 355K.

What are the steps to reproduce the bug?

As far as I know, run a coupled case with development version of CESM and RRTMGP (before the workaround).

What CAM tag were you using?

before cam6_4_044

What machine were you running CAM on?

CISL machine (e.g. cheyenne)

What compiler were you using?

Intel

Path to a case directory, if applicable

No response

Will you be addressing this bug yourself?

No

Extra info

We should discuss this in at least the CAM development meeting to come up with a strategy for how to better diagnose this issue.