In the end it is the energy budget that determines what happens. The chaotic part is mostly just internal redistributions of energy.
Not so.
The ‘Lapse Rate’ feedback ( evidently another fail ) is modeled to occur because of ‘internal redistribution of energy’. Circulation matters to how much energy goes to space.
If there were zero convection, the surface would warm to be much hotter ( much more so than for doubling CO2 ) to achieve radiative balance. But the real atmosphere overcomes the greenhouse obstructed radiance from the surface by moving heat via convection instead.
Now, I’ve run a radiative model on global atmospheres and the effect of doubling CO2 is positive and fairly consistent regardless of the variations that circulations impose – FOR THAT PARTICULAR ATMOSPHERIC SOUNDING. But that’s not a ‘fair fight’ because the atmosphere will respond, so considering only the radiance without knowing how circulation responds is still a guess.
Further, models have to make stuff up in parameterization. Little non-linear misunderstood errors in these fabrications can compound in unknown ways.
Further, it doesn’t appear that the models can even get the temperature right and that one would think would be the easiest part.
Which spaghetti strand is correct? None of them: