Nicholaswogan
commented
1 year ago - I checked two_stream_ir, benchmarking it against the ATMOS CLIMA version. They always agreed to a factor of 1.0e-7. two_stream_ir is not the issue!
- I compared my CO2-CO2 CIA opacities to the opacities in ExoRT. They looked extremely similar. I think the CO2-CO2 opacities are OK
- I re-did the calculation using the CLIMA CO2 k-distributions link. I got 116 W/m^2, so still very wrong.
If I had to guess, I think there are two problems:
- CO2 k-distributions are not quite right
- There is a problem interpolating the k-distributions
I think this because, when I randomly increase the input pressure by 1e6, then I can more closely reproduce petitRADTRANS, when using identical opacities. Doing the same exercise, but using the ATMOS CLIMA k-distributions, I get 87.6 W/m2, and the planet TOA flux vs wavelength looks somewhat comparable to other codes (e.g. ExoRT).
It seems very likely that there is an issue with the CO2 k-distributions because petitRADTRANS gets the wrong answer, yet the code itself is very well-validated.
Figure 2 of this paper shows a radiative transfer calculation benchmarks with 3 models: ExoRT, SOCRATES, and SMART. The atmosphere is a 2 bars pure CO2 atmosphere on Mars with 250 K surface. All model get approximately OLR = 90 W/m^2. The CLIMA model also gets about 90 W/m^2 (see here)
I fail to reproduce this benchmark.
Using my normal opacities, I get 120 W/m^2, and petitRADTRANS get about 110 W/m^2.
Possible issues
two_stream_ircould have a bug