Most of us will be aware that the Earth's atmosphere consists principally of nitrogen, which makes up nearly four-fifths, and oxygen, which constitutes just over a fifth. The remainder of the gases in the atmosphere together add-up to only a fraction of a percent of the total. Thus, rule of thumb would seem to suggest that the atmosphere's physical properties will be mostly those of the gas nitrogen, perhaps with a little of the characteristics of oxygen. To a large extent that is true, but in one very significant respect it is not.
We have seen that the temperature of any object in the Earth's orbit will depend on a balance between two things; the amount of incident radiant heat from the Sun, and the radiant heat loss from all surfaces of the object itself. When these two are equal, the object's temperature will remain stable. If they are unequal, then the object will tend to warm or cool until its own thermal emissions match the inbound energy from the Sun. (insolation)
The curious thing is that a satellite or spacecraft in Earth's solar orbit, without any form of onboard heating, tends to stabilise at a temperature somewhat below freezing. To maintain a comfortable environment for astronauts, supplementary heating is required. This is an unexpected finding, and in fact there were textbooks written in the 1950's which proclaimed that an astronaut's capsule in direct sunshine would overheat if it were not cooled by some form of air conditioning . They were wrong. The Apollo 13 incident demonstrated this fact amply well, with the returning crew having to don all available clothing in an attempt to keep warm in their unheated craft following a power failure.
Yet, we know that the average temperature of the Earth's surface is above freezing, about 14C in fact. Some phenomenon must account for the greater retention of heat by the Earth itself, as compared to a bare satellite. The intuitive thought would be that the difference is in the Earth having a deep external atmosphere, which perhaps forms an insulating layer around the planet. However, this is incorrect. The reason it is incorrect is that nitrogen and oxygen, like most gases, are transparent substances. Transparent, that is, to both visible light and infrared. As such they offer little or no barrier to prevent heat radiated by the Earth's surface from exiting to space.
As an aside, it is worth recalling that if these gases allow radiative heat to pass through freely, then Kirchhoff's Law of Thermodynamics tells us two things: That the gases will not themselves by heated by this outgoing radiative heat, and that if the gases themselves were to become heated by some other means, they would not be able to warm the surface of the earth by way of radiating heat. (Although, they could possibly do so by convection or conduction.) A neater way to express this is to say that typical gases are not black bodies, and unlike solids or liquids, do not emit or absorb black body radiation. More on this point later, as it is important.
Thus, the picture we see here is that the Earth's atmosphere is NOT an 'insulating blanket' in the manner of the glasswool in your loft. It cannot prevent radiated heat from the surface exiting to space. The fact that it might block convected heat loss is largely immaterial, since no convection is possible in space anyway. It may thus come as a surprise that the simple two-gas atmosphere is a poor thermal insulator, and cannot account for the Earth's relative warmness. So, what does?
The answer involves special properties of some minority constituents of the atmosphere, notably water vapour, carbon dioxide, and to a lesser extent hydrocarbon gases such as methane. These substances show the phenomenon of molecular resonance in response to stimulation by specific wavelengths of infrared radiation. These molecular resonances result in certain wavebands of infrared (heat) radiation being absorbed by the molecules of greenhouse gas, and then after a brief delay, re-emitted in a random direction. Sometimes this process is termed absorbtion, but that is a term of convenience used by scientists It would be more correct to describe the effect as random scattering. An explanation of how molecular resonance operates perhaps deserves its own page. When I get time. Meanwhile, this CalTech page on spectroscopy explains it quite well. (no need to do the coursework though!)
Thus, a photon of infrared leaving the surface will have its direction of travel changed numerous times during its passage through the atmosphere. In some cases it may return to the Earth's surface from whence it came. In that event the heat energy it contains is effectively returned to the Earth so the net result is like having insulation. Other photons will eventually make it through to space, in which case that is heat lost.
Theoretically this would suggest that with a substantial greenhouse gas presence, 50% of all photons would be returned to the surface and 50% would escape. Furthermore, once that ratio was reached, increasing the amount of greenhouse gas would not alter it to any significant degree. More collisions with molecules would simply increase the time taken for the photon to either escape or be brought back home, but would not alter that odds.
In practice this is not quite true because CO2 molecules can and do collide with nitrogen and oxygen molecules. In fact, the interval between such collisions with the bulk gas, and the time for which an absorbed photon is held by CO2 before being released again, are about equal at 10 nanoseconds. Thus, some energised CO2 molecules will be forced to give up part or all of their vibrational energy to a bulk gas molecule as a result of a collision, in which case they will be unable to re-emit the photon. In other cases, collisions with bulk gas will impart sufficient vibrational energy to a CO2 molecule for it to emit a brand-new infrared photon of its own accord, without having had to absorb one. Probably, both happen to some extent. This is where things get a little complex, because it is extremely hard to predict which of these two mechanisms will predominate. It may be that some of the 'greenhouse effect' trapped heat is transferred to the bulk gas, warming the atmosphere, or that the bulk gas excites CO2 molecules which in turn radiate 50% of that heat to space and the other half back to the Earth. Opinions on this, and its likely effect on the greenhouse effect, differ.
As well as being a powerful greenhouse gas, water vapour has another special property in that it may condense into ice or rain. Thus, because they are not pure gas but contain solids or liquids, clouds have rather different properties, thermally speaking, from the surrounding air. We mentioned that gases are not in principle 'black bodies' and therefore do not participate in the exchange of thermal radiation with other warmed surfaces. The same limitation does not apply to clouds, which are capable of absorbing bulk heat radiating from the surface, and of radiating heat back toward the surface.
Clouds, due to their high albedo, are also capable of acting as straightforward reflectors of both visible and infrared radiation, mirror-style, which likewise does not involve thermalisation or black-body effects. This has two distinct effects: In daytime, incoming sunlight is reflected back to space reducing solar warming, and at night, outgoing heat from the planet is reflected back to its surface, reducing heat losses. Thus, the effect of clouds is to stabilise surface temperatures, reducing the day/night difference. This effect is empirically obvious; a cloudless sky in winter typically means a very cold night, but a cloudless sky in summer means a scorching hot day. Cloudy skies mean more moderate weather in both seasons.
Now, clouds are principally a product of the solar heating of oceans and wetlands. Therefore it raises the thought-provoking consideration that an enhanced greenhouse effect will have a secondary result of increasing that which stabilises temperature differences between day and night, namely cloud cover. Thus, since extreme weather typically results from dramatic temperature shifts along weather fronts, may well be that global warming reduces the frequency of extreme weather events. No formal proof of this to hand, but it is certainly an intriguing thought.
The work of Svante Arrhenius on carbon dioxide (NASA)