Data had to be gathered from around the world, technology had to develop to allow us to analyse that data, and basic advances in physics and other disciplines had to come about before we could understand it. What we now know about the climate is thanks to literally generations of dedicated researchers.
However, it is also easier to observe human caused climate change today because we've put so much more carbon dioxide into the climate system over the past century that the impacts of climate change are now clearly visible, affecting people and ecosystems all over the world.. More cars, more factories, and more power plants - all are changing the climate faster than was previously possible - and causing more obvious changes.
Taking the planet's temperature
To get an accurate picture of how warm the Earth is, you need measurements from all over because the whole planet does not heat up at the same rate. In fact, some parts might even cool down while the world as a whole heats up. Also, many temperature readings are needed over time to develop an accurate long-term picture. In order to develop a global temperature history, researchers have had to travel to the farthest corners of the Earth, and come up with ways to "go back in time".
Some sources of past temperature data:
- Historical records - Includes sources like ship's logs, farmer's diaries and newspaper articles. When carefully evaluated these can provide can provide both quantitative and qualitative data.
- Personal accounts and oral histories - Useful information can be gathered especially from the older generations of indigenous people who have always relied on nature for their survival, and so are particularly observant of changes over the past decades.
- Direct (e.g. thermometer) measurements - Only go back around 300 years, and are very sparse until about 150 years ago. Also, differences in thermometer types and other variables have to be taken into account.
- Data collected by balloon and satellite - Very useful, but only available since 1979.
- Tree ring thickness - Width and density varies depending on growing conditions.
- Ocean and lake sediments - Billions of tons of sediments accumulate each year. The tiny preserved fossils and chemicals in layers of sediment can be used interpret past climate.
- Coral skeletons - The water temperature that the coral grew in can be determined from trace metals, oxygen and the isotopes of oxygen contained in its skeleton.
- Fossil pollen - Each plant has uniquely shaped pollen. Knowing what plants were growing at a particular time in the fossil record lets scientists make inferences about what the climate was like at the time.
- Ice cores - Over the centuries snow falling on high mountains and the polar ice caps packs down and becomes solid ice. Dust and air bubbles trapped in this ice provide valuable climate data. For example, the air trapped in the ice serves as a record of carbon dioxide concentrations across the millennia.
- Observed melting - Rates of glacial retreat, permafrost thaw, shrinking polar ice caps and reduction in Arctic sea ice are indicators of both short and long term climate change.
The important thing is not to look at any one source of data independently, but instead to take them together. This produces a scientifically compelling picture of a warming world that matches with the corresponding increase in greenhouse gasses.
Predicting the climate future
Global climate models are mathematical representations of the real world's climate. Some models are attempts by scientists to boil the complex behavior of the climate down to (comparatively!) simple formulas in an attempt to understand the forces at work. However, when people talk about specific predictions of long term climate behavior they are usually talking about general circulation models. In these models, the equations are tweaked (within reason) until the model is able to predict past and present conditions, as accurately as possible, when tested against actual observations of past and present conditions.
Since it's impossible to know every last variable, and because the model will never match the real world perfectly, scientists compensate by running each model over and over, while making tiny changes to the starting conditions (increasing the wind speed over Detroit by one percent, for example) and other factors. This way they can get an idea of the different possible outcomes. If one result occurs more frequently then another then it's the more likely outcome.
In the end, each model predicts a range of possible outcomes. For example, the IPCC, taking into account all of the different available models, settled on a projected global temperature rise of about 3 to 8° Fahrenheit (1.4 - 5.8° Celsius). No one can say exactly how much the temperature will increase over the next hundred years, but with a couple of caveats it is a safe bet that it will be within this range.
One thing climate models can not predict are all the possible effects of feedback mechanisms, which might help stabilise the climate or cause the climate to change much faster and in unpredictable ways. Of course, it would be irresponsible to ignore the climate models and hope for the best because of these uncertainties.
Another thing these models can not really predict is human behavior, and ingenuity. We could burn more fossil fuels than expected, and end up with a hotter planet then even the worst case scenario. Or we could deploy renewable energy and energy efficiency solutions faster then thought possible - eliminating the likelihood of the higher temperature ranges.
US National Oceanic and Atmospheric Administration - Climate proxy data
American Institute of Physics - History of climate change research
The Climate Change Connection - Climate research
ClimatePrediction.Net - Distributed computing climate modelling project (join the Greenpeace team on the project!)