See Eli’s post.
We can summarize the data in the figure above adding that ~40 W/m2 go directly from the surface to space as IR radiation of the 398 W/m2 leaving the surface. In and out in the table … [AT THE POST] … means into and out the surface the atmosphere and space respectively. In is taken as a positive addition to the heat content and negative a decrease …
… The important point is to realize that surface IR radiation absorbed in the atmosphere is rapidly (10 μs) thermalized and converted into random motion of the molecules in the atmosphere, just as is latent heat from condensation of water vapor and from sensible heat. Very little, less than a part per million, is directly radiated back to the surface and we can neglect that.
The 342 W/m2 of back radiation is OBSERVED, so this ain’t a model or a theory, where does it come from? It comes from ALL of the sources pushing heat into the atmosphere, from the convective and radiative heat transfer from the surface.
(Emphasis in the original.)
Perhaps it’s just my physics training, but I never understood the primacy some (even scientists) put on the primacy of convection in terms of climate. I mean, sure, a Bunch of energy can come into the convective-dominated part of the climate system, and it might reside there for a duration, perhaps even long, but, really, that doesn’t matter. Eli’s point is that if a different bunch of the same amount doesn’t leave the climate system, it’ll warm. And it doesn’t matter how long the first bunch is in the climate system, or what path it takes through it, or anything of the kind.
So, to me, this idea that the various oscillations, like NAO or PDO or ENSO somehow have something substantial to do with the overall climate and problem is specious. Yeah, there are big energy flows from one part of the system to the other, just as there are big flows of Carbon to and from oceans to atmosphere, but that’s just slosh, and the point is the net balance. And human emissions of (about) 10 GtC per annum are affecting that a lot.