A small localized
change in surface temperature can cause a convection burst (thunderstorm) and a large increase
in convection height, improving both reflection of incoming solar radiation, and conveying sensible heat to a higher altitude where it can then escape to space
via radiative processes with far less interference.
The short and lazy answer to Matthew Marler's question about the column energetics that bring the system back to equilibrium is that GCMs do of course represent evaporation, sensible heating, etc
in ways that are undoubtedly imperfect (e.g.,
via «bulk formulas» that transfer energy down - gradient of
temperature or humidity differences between the
surface and air aloft), but they are free to evolve
in climate
change scenarios
in ways that are physically self - consistent.