Online climate change projections report Box 2.1
The only way we can calculate how climate will change due to human activities is to use a mathematical model of the earth’s climate system, known simply as a Global Climate Model (GCM). This describes the behaviour of the components of the climate and interactions between them. Firstly, the atmosphere; the way it moves horizontally and vertically, plus physical processes that occur in it, such as the formation of clouds and precipitation, and the passage of terrestrial and solar radiation through it. Secondly the ocean, because there is a continual exchange of heat, momentum and water vapour between the ocean and atmosphere and because within it there are large currents which transport heat, water and salt. Thirdly the land, because it affects the flow of air over it, and is important in the hydrological cycle — not just the land surface but soils beneath it — and changes in the land surface (both natural and human-made) affect the climate. Lastly the cryosphere; ice on land (snow, glaciers and ice sheets) and on sea. All of these components of the climate system interact to produce the feedbacks which play a large role in determining how climate will change.
Typically, a global climate model breaks up the surface of the earth into a number of latitude/longitude grid boxes. It divides the atmosphere into layers, from the surface to the stratosphere, and does the same for the ocean, from the surface to the deepest waters (Figure 2.5). At each of the points on this three dimensional grid in the atmosphere a number of equations, derived from the basic laws of physics, are solved which describe the large-scale evolution of momentum, heat and moisture. Similar equations, but including different variables, are solved for the ocean. The third Met Office coupled ocean-atmosphere GCM, HadCM3, has a resolution over land areas of 2.5° latitude x 3.75° longitude, with 19 vertical levels in the atmosphere and four layers in the soil. The ocean model has 20 vertical levels and a grid size of 1.25° latitude x 1.25° longitude. In all, there are about a million grid points in the model. At each of these grid points, equations are solved every time the model steps forward (typically 30 minutes of model time) throughout an experiment which typically lasts 250 model years.
Parametrisations in climate models
Many of the most important processes in the climate system (for example the drag exerted by hills as air flows over them, and the formation of clouds) take place at a scale much smaller than the grid size of GCMs — these are called subgrid-scale processes. These cannot therefore be described explicitly, so we develop relationships, known as parametrisations, which estimate them from grid scale variables such as winds, temperature, humidity etc. which are explicitly described in the model.
We illustrate this by taking the example of cloud amount. This is defined as the proportion of each model grid square which is covered by cloud at each level in the atmosphere. To calculate cloud amount in HadCM3, we use the model’s calculated mean temperature and water vapour content for that square and level; this is known as parametrising cloud amount in terms of the large scale model variables. Now the equation relating water vapour and temperature to cloud amount contains some parameters, the values of which are based on results from, for example, aircraft measurements or high resolution process models such as cloud resolving models. The values of these parameters are uncertain, and this is a major cause of model uncertainty. So, to quantify this model uncertainty, we vary these parameter values between plausible limits to form variants of a number of configurations of the model, in order to generate the ensembles of simulations which form the primary basis for the PDFs in UKCP09.
But the parametrisation which predicts cloud amount from the modelled large scale variables may be different in models from other centres; not just the parameter values but the actual form of the parameterisation scheme itself; this is illustrated schematically in Figure 2.6. This is an example of a structural difference between models; the effect of structural differences cannot be taken account of using variants of a single model alone. In UKCP09 it is taken into account in the probabilistic projections by using a number of models from other centres, as explained in Chapter 3.
Basic greenhouse theory tells us that when the concentration of a greenhouse gas, such as CO2, increases in the atmosphere, it alters the balance between the amount of incoming energy from the sun and that leaving the earth as infrared energy (the radiative balance). Given enough time, the climate system adjusts to this new condition by increasing the surface temperature of the earth. The direct radiative effect of a doubling the concentration of CO2 in the atmosphere would eventually cause the surface temperature of the earth to increase by about 1ºC. However, once a greenhouse warming starts, a number of consequent changes start to happen which can act to either reduce or increase the direct greenhouse warming; these are known as negative or positive feedbacks respectively.
We illustrate this with some examples. Firstly, as the atmosphere starts to warm due to the direct greenhouse effect, it can “hold” more water vapour — and models indicate that water vapour concentration increases to maintain time-averaged relative humidity (which also depends on temperature) approximately constant as climate change proceeds. As water vapour is a powerful greenhouse gas this effect will further increase warming — a positive feedback. Secondly, as the oceans start to warm some sea-ice will melt. Sea-ice reflects back a lot of solar radiation, but the open ocean it exposes when it melts absorbs more radiation; this will reinforce the original warming effect — another positive feedback. Thirdly, one of the most critical feedbacks, but also one of the most complex, is that due to changes in clouds. In the present climate, clouds have a large effect on climate; high clouds act to increase surface temperatures but low clouds tend to cool climate; the net effect is a cooling one. Greenhouse gas — driven climate change can alter many characteristics of clouds at all levels — their amount and altitude, and the properties of their contituent water droplets and ice crystals, for example. Such changes can alter the radiative properties of clouds — the effect they have on incoming solar radiation and outgoing long wave radiation — and the net effect could be either positive or negative. The last example is that of changes of land surface vegetation (from forests to grassland, for example, or desertification) due to changes in rainfall or temperature which in turn can alter local and global climate. There are many other feedbacks, both positive and negative, in different parts of the climate system.
Feedbacks naturally arise in the climate model because the processes which lead to them (in the second example above this is the formation of sea-ice and its reflectivity) are explicitly represented or parametrised. Many feedbacks take place at a small scale and capturing their overall effect in the model therefore depends upon the parametrisations of small scale processes. Hence the strength of the feedbacks, and thus future changes in climate, will depend on the form of the parametrisation used (part of the model structure), and the values of its constituent parameters. This is one of the main causes of the differences between projections from different models. The methodology developed for the UKCP09 projections is designed to sample these uncertainties, to the extent that this is presently possible, in a systematic way.
The carbon cycle and the sulphur cycle represent two important processes in climate change, yet, as with standard processes in the atmosphere and oceans, they carry their own large uncertainties. Here we give an overview of the processes, the uncertainties, and how UKCP09 includes them in the final probabilistic projections; more detail resides in Chapter 3.
Currently about half of the emissions of CO2 from human activities (fossil fuel combustion and land use change) are taken up by sinks on land (vegetation and soils) and in the ocean (seawater and ecosystems within it), leaving the remainder of the CO2 in the atmosphere where it increases concentrations. But as climate starts to change, carbon sinks can also change, so may be able to absorb more, or less, CO2 from the atmosphere. For example, as soils warm they increase their respiration of CO2 back to the atmosphere and their ability to remove CO2 will weaken, leading to atmospheric concentrations being higher than they would otherwise be — a positive feedback. On the other hand, a warmer climate will encourage the growth of boreal forests which would take up more CO2 from the atmosphere — a negative feedback. There are a host of such feedbacks, both positive and negative, although the net effect is a positive one. Uncertainties in estimating atmospheric concentrations resulting from emissions were not dealt with in the IPCC Third Assessment Report (TAR) in 2001, and hence could not be taken into account in UKCIP02. In UKCP09 these feedbacks are included, and the uncertainty they add to climate change projections is estimated using two sources of information. Firstly, using variants of the Met Office coupled climate — carbon cycle model with different values for the land carbon cycle parameters within it. Secondly, using results from a project (known as C4MIP) which compared results from a number of international models which include the carbon cycle. Further detail is given in Chapter 3. Note that, although UKCP09 projections include the feedback from both land- and ocean-carbon cycle projections, they only include the effect of uncertainties in the feedback from land, which has been estimated (in C4MIP, see Friedlingstein et al, 2006) to be several times greater than that from the ocean component. Because the processes involved in climate — carbon cycle feedback are less well understood, and projections are less constrained by observations, our ability to assess the uncertainty in these is more limited than for other aspects of the climate system.
Sulphur gases emitted from fossil fuel burning, and naturally from the oceans, takes part in chemical reactions in the atmosphere to form small particles — sulphate aerosol. These are eventually removed from the atmosphere by rain and clouds, having a typical lifetime of a few days, but whilst in the atmosphere they can have a substantial cooling effect on climate in a direct and an indirect way. The direct cooling effect arises when a suspension of aerosols in the clear atmosphere reflects back some of the incoming solar radiation before it has a chance to warm the ground. The indirect effect arises from the ability of sulphate particles to act as additional nuclei on which water vapour condenses to form clouds. Such clouds would therefore have more water droplets, each of which (for a given amount of available water) would be smaller — the total surface area would therefore be greater and the cloud would reflect back more solar radiation — a further cooling effect. Both the direct and indirect effects described above are included in the HadCM3 model.
A second indirect effect occurs within sulphate-laden clouds. Because their droplets are smaller than those in clean air, the processes which lead the droplets to grow heavy enough to form rain are slower, and hence the clouds persist (and reflect back solar radiation) longer — a further indirect cooling effect. This is a much more complex process, and is only now becoming understood well enough to be included in models (such as the Met Office earth system model, HadGEM1) but is not included in UKCP09. Because atmospheric sulphate burdens are expected to decline in the future, the omission of this effect may lead to an underestimate of changes in the first few decades of the UKCP09 projections.
Constituents included, and not included, in the probabilistic projections
The atmospheric constituents included in HadCM3, its corresponding simple-ocean configuration and the regional climate model, are shown in Table 2.1. With the exception of the cloud persistence effect of sulphate aerosols, the projected combined effect by 2100 of changes in those constituents not included is unlikely to add a significant amount to overall uncertainty. Similarly, although the Met Office model includes the effect of chemical reactions in the atmosphere which determine concentrations of methane and tropospheric (low altitude) ozone, no attempt was made to estimate the consequent uncertainty in concentrations; this would also be expected to have a minor effect. Uncertainty in the climate effect of northern hemisphere stratospheric ozone changes is also likely to be small relative to those quantified.
In contrast, other components of the methane cycle, such as climate-induced emissions from wetlands, melting permafrost and methane hydrates, do have the potential to modify future climate change significantly. However, these feedbacks are so poorly understood as to make estimates of their effect very uncertain, and hence they are not currently integrated into any climate model.
|Constituent||Whether included |
|Carbon dioxide ||Yes|
|Nitrous oxide ||Yes|
|CFCs, PFCs, HFCs, HCFCs, SF6 ||Major ones |
|Tropospheric ozone ||Yes|
|Stratospheric ozone || Yes |
|Sulphate aerosols — direct effect ||Yes|
|Sulphate aerosols — cloud albedo effect ||Yes|
|Sulphate aerosols — cloud persistence effect ||No |
|Black carbon aerosol ||No|
|Organic carbon aerosol ||No|
|Mineral dust ||No|
|Sea salt aerosol ||No|
|Land cover (albedo effect) ||No|