The world's climates have changed over the last 100 years. The northern hemisphere is probably warmer now than at any time in the last millennium. It is very likely that most of the observed warming since the mid-20th century has been due to increasing greenhouse gas concentrations in the atmosphere, particularly carbon dioxide from combustion of fossil fuels.
The underlying physical reason for this is well-understood: it is because carbon dioxide and other greenhouse gases absorb heat radiation; thus, raising their concentration in the atmosphere increases the Earth's thermal insulation from space. It is analogous to the effect of cavity wall insulation: a well-insulated house will have a higher temperature indoors than a poorly insulated one, if the same amount of heat is used. In fact, when we quantify the amount by which the mean temperature at the surface of the Earth is raised by increasing the heating due to greenhouse gases, we use a number whose units (W m⁻² °C⁻¹) are the same as the units of the "u-value" which is used to measure the thermal insulation of walls and windows of a house.
The most important application of climate science is to predict changes in climate and sea level in coming decades and centuries. To do this reliably and precisely, we need an improved quantitative understanding of the processes at work.
© Jonathan Gregory 2003
© IPCC 2007 (WG1 Fourth Assessment Report)
|Glaciers worldwide have been contracting since the 19th century. The moraine along the right-hand side of this glacier, which is the Glacier de Moiry in the Valais, Switzerland, shows its size at that time. This is a clear indication of widespread warming, and makes a substantial contribution to global mean sea-level rise.||Combined millennial Northern Hemisphere temperature reconstruction from various indirect kinds of evidence (proxies, such as tree rings). The shading shows the extent to which various datasets agree. The black line shows the instrumental record (measurements with thermometers). (There is more data available for the northern than for the southern hemisphere.)|
The main tool for making such predictions are computer programs, usually referred to as "climate models" (and more specifically as "atmosphere-ocean general circulation models", AOGCMs), which encode mathematical descriptions of our physical understanding of the behaviour of the climate system, based on observations and theory. These are complex programs, hundreds of thousands of lines long, because there is a lot we understand. On the other hand, there is also a lot that we don't understand. Models which make different assumptions about things which aren't well-known make consequently different predictions. Most of my research is concerned with aspects of understanding the causes of these model differences and which predictions are most realistic, by comparison of the models with each other and with climate change which has been observed in the last century and which has naturally occurred in the earlier history of the Earth.
© IPCC 2007 (WG1 Fourth Assessment Report)
Church et al. (2011) © The Oceanography Society 2011
|Observations and predictions of global mean temperature change (left) and global mean sea level rise (right) during the 21st century under several different emissions scenarios.|
Different models give a range of predictions of climate change for any given emissions scenario. Much of the the spread of model predictions can be attributed to differences among models in two categories of phenomena. Climate feedbacks are changes occurring in the atmosphere or at the surface which tend to reinforce or mitigate climate change. For example, the reduction in the area of snow which accompanies warming of the climate gives a positive feedback, because it reduces the reflection of sunlight, leading to more heating of the climate. Heat uptake by the ocean strongly influences the rate of climate change. The ocean has a large capacity to absorb heat, and at present it is retarding global temperature rise. Warming of the ocean leads also to sea level rise due to thermal expansion of the sea water.
|Gregory and Forster (2008) © American Meteorological Society||© Crown Copyright 2003, courtesy of the Hadley Centre.|
|A comparison of the climate feedback parameter (red) and ocean heat uptake efficiency (blue) in various climate models. These two numbers measure the resistance of climate change to forcing; that is, a model with a larger value of either parameter will tend to predict a smaller global mean temperature rise.||Global average ocean temperature change (degrees Celsius) in a climate model prediction experiment following scenario IS92a for future greenhouse gas increases. The warming spreads downwards from the surface. The rate of warming is larger in the 21st than the 20th century.|
|Model predictions for the 21st century indicate that
climate and sea level change will not be geographically uniform. The
variations are large compared with the global average change. For practical
purposes, it is regional change which matters and it is therefore essential to
understand the mechanisms which determine the geographical patterns.
For temperature change, climate models agree qualitatively with regard to some features, for instance that warming over the land is on average greater than over the sea, and that warming is enhanced at high northern latitudes. However, the models do not agree quantitatively. For precipitation and sea level change, there is much less similarity between the patterns predicted by various models.
|© IPCC 2007 (WG1 Fourth Assessment Report)|
|The maps show the changes in surface air temperature (above left, at two periods in the 21st century), precipitation (below left, for the seasons December-January-February and June-July-August, at the end of the 21st century), and sea level (right, at the end of the 21st century), predicted by a set of climate models assuming scenario A1B for greenhouse gas and other emissions. In the precipitation maps, the white areas are where the models do not agree on whether precipitation will increase or decrease, and the stippled areas where 90% of them agree on an increase or a decrease. The sea level map is relative to global mean sea level rise. That is, the negative areas are where sea level is predicted to rise by less than the global mean. The stippled areas are those where the models have some measure of agreement about the size of the change; these are very limited areas.|