Ethan Siegel in Forbes:
If you didn't know anything about climate science, about the Earth's temperature, about carbon dioxide or greenhouse gases, but you wanted to, how would you go about doing it? You'd begin by constructing a plan for how you'd accurately scientifically investigate the problem. You'd think about the data you'd need to collect and how you'd gather it. You'd think about the measurements you'd want to make and how to make them. You'd think about the sources of error and how to account for them: how to properly calibrate your data from all over the world and from many different time periods. And then you'd bring it together, under one enormous framework, to try and draw a scientifically robust conclusion.
Your first step would be to go out and try to measure the heat content of the planet. You'd measure the temperature of the air where you are, and you'd attempt to do it all over the world. The continents would be the easiest, and then you'd go after the oceans. The sea-level temperature would be low-hanging fruit, and then you'd have to go beyond the obvious to measure the heat trapped in the upper atmosphere and in the deeper waters of the seas. You'd try and measure it everywhere today, but also to reconstruct what it was in the past, going as far back as you can.
Because you can't go back in time and take measurements that you weren't there to take, you'd then look for proxies, or things you can measure directly that can give you information about past temperature. You'd discover that the science of dendroclimatology — or inferring past climates and temperatures from the properties of trees that were growing at the time — can give you lots of robust information.
More here.