From Forbes.com (Ethan Siegel). Click through and read the whole article. Here’s an excerpt:
Modeling the Earth’s climate is one of the most daunting, complicated tasks out there. If only we were more like the Moon, things would be easy. The Moon has no atmosphere, no oceans, no icecaps, no seasons, and no complicated flora and fauna to get in the way of simple radiative physics. No wonder it’s so challenging to model! In fact, if you google “climate models wrong”, eight of the first ten results showcase failure. But headlines are never as reliable as going to the scientific source itself, and the ultimate source, in this case, is the first accurate climate model ever: by Syukuro Manabe and Richard T. Wetherald. 50 years after their groundbreaking 1967 paper, the science can be robustly evaluated, and they got almost everything exactly right.
If there were no atmosphere on Earth, calculating the climate would be easy. The Sun emits radiation, the Earth absorbs some of the incident radiation and reflects the rest, then the Earth re-radiates away that energy. Temperatures would be easily calculable based on albedo (i.e., reflectivity), the angle of the surface to the Sun, the length/duration of the day, and the efficiency of how it re-radiates that energy. If we were to strip the atmosphere away entirely, our planet’s typical temperature would be 255 Kelvin (-18 °C / 0 °F), which is most definitely colder than what we observe. In fact, it’s about 33 °C (59 °F) colder than what we see, and what we need to account for that difference is an accurate climate model.
The number one contributor, by far, to this difference? The atmosphere. This “blanket-like” effect of the gases in our atmosphere was first discovered nearly two centuries ago by Joseph Fourier and worked out in detail by Svante Arrhenius in 1896. Each of the gases present has some amount of absorptive effects in the infrared portion of the spectrum, which is the portion where Earth re-radiates most of its energy. Nitrogen and oxygen are terrible absorbers, but good ones include water vapor, methane, nitrous oxide, ozone and carbon dioxide. When we add (or take away) more of those gases from our planet’s atmosphere, it’s like thickening (or thinning) the blanket that the planet wears. This, too, was worked out by Arrhenius over 100 years ago.
But a true climate model is more complex, because there’s more at play than just the atmosphere. The oceans ensure that the amount of water vapor (and cloud cover, which impacts temperature significantly) change dependent on conditions, and if you tinker with one component of the atmosphere — like carbon dioxide, for instance — it impacts the concentrations of other components. Scientists refer to this general process as feedback, and it’s one of the largest uncertainties in climate modeling.
The big advance of Manabe and Wetherald’s work was to model not just the feedbacks but the interrelationships between the different components that contribute to the Earth’s temperature. As the atmospheric contents change, so do both the absolute and relative humidity, which impacts cloud cover, water vapor content and cycling/convection of the atmosphere. What they found is that if you start with a stable initial state — roughly what Earth experienced for thousands of years prior to the start of the industrial revolution — you can tinker with one component (like CO2) and model how everything else evolves.