update initial guess to be more formal and general#13
Conversation
evan-greenbrg
left a comment
There was a problem hiding this comment.
Looks great. Just two small comments. Using RT.calc_RT_quantities is much cleaner.
| # TODO - make this a function, and use it here and in radiatve transfer | ||
| # transmit thermal emission through the atmosphere | ||
| transup = rhi["transm_up_dir"] + rhi["transm_up_dif"] | ||
| L_up = Ls * transup |
There was a problem hiding this comment.
We should make this TODO explicit that lines 206-207 are only compatible with rt_mode = 'transm'. If rhi["transm_up_dir"] and rhi["transm_up_dif"] are in rdn units then Ls * transup becomes a rdn * rdn multiplication.
I think we may have flagged this earlier during a working session, and also may be what the TODO is referencing.
There was a problem hiding this comment.
Oooh, good catch, I'll patch.
There was a problem hiding this comment.
Actually as I look more, this should be safe with the convention we discussed this morning (which does still need to get codiied), whereby 6c components should only hold transmittances, and dir-dif (etc.) terms should hold radiances.
There was a problem hiding this comment.
Okay, I've now tried to codify this as a rule.
|
@pgbrodrick This looks clean to me. Only extremely minor comments (see above). So, from what I can get from this PR, it sounds like the following decisions have been made at the meeting this morning:
Is that correct? |
|
I think we're leaning towards the convention that if you're doing a 6c model, it's only the dir-dif / dif-dir terms that should get populated. IE, that holding radiance-based transmittances in that case aren't used, and that we shouldn't rely on them. Outside of the transup case, this shouldn't come up (I don't think?) from this PR....but needs to be resolved & codified more generally. For the second bullet, I'm making the assumption (which I think is true) that calc_RT_quantities returns the directional version of those quantities. I agree we should also quantify our convention about whether sza & cosi are baked into those terms (right now I believe they are not), but that is the property of calc_RT_quantities. |
|
Got it! These are useful bullet points for future updates but shouldn't hold up this PR. |
Proof it works: