|
|
|
|
The UK climate
UKCIP02
Headline messages
UKCIP02 guidance
UKCIP02 maps
Documentation
UKCIP02 extras
FAQ
|
 |
Sources of uncertainty
Global climate models are the only tools currently available to us for simulating the complex set of processes that determine climate at global and regional levels. It is important to remember that they represent our current understanding of how the climate system works.
Climate change scenarios can be defined as coherent and internally-consistent descriptions of future climate given certain assumptions about the growth of the emissions of greenhouse gases and about other factors that may influence climate in the future. As long as they meet these criteria (and the UKCIP02 scenarios do ) then uncertainty is not relevant. However, in practice the scenarios are used as predictions of future change in order to estimate impacts (e.g. by UKCIP02 users). As predictions, they have considerable uncertainty associated with them.
The uncertainties associated with the modelling of future climate scenarios have been divided by the Hadley Centre into three broad categories: (i) emissions uncertainty; (ii) natural climatic variability; and (iii) modelling uncertainty (see Handling uncertainties in UKCIP02 scenarios of climate change by Geoff Jenkins and Jason Lowe, November 2003).
(i) Emissions uncertainty
Uncertainty about future emissions arise because we can not know with certainty how populations, economies, energy technologies and other social factors that influence greenhouse gas emissions will change in the future. The best we can do is consider a range of plausible ways in which the world might develop. The IPCC Special Report on Emissions Scenarios (SRES) undertook this in 1999 and produced four non-interventionist scenarios (i.e. not incorporating explicit implementation of climate policies such as the UNFCCC or the Kyoto Protocol). These emissions scenarios diverge immediately and rapidly (Figure 1), however SRES clearly state that it is not possible to put relative likelihoods on any of their emissions scenarios, that they are not equally probable, and that there is no single "most likely", "average" or "best guess" scenario.
|
|
|
|
|
| |
|
Figure 1. Emissions of carbon dioxide (Gt of carbon) in four of the SRES emissions scenarios.
|
| |
The climate model used by the Hadley Centre to create the UKCIP02 scenarios used a wide and representative range of SRES emissions scenarios (Figure 1). The A1F1 SRES emissions scenario corresponds to the High Emissions UKCIP02 scenario, A2 corresponds to Medium-High Emissions, B2 to Medium-Low Emissions and B1 to Low Emissions.
Figure 2 shows the predictions of global temperature from each of these. The full range of model results from all the IPCC modelling centres involved in the Third Assessment Report using all SRES future emissions scenarios is shown by the dashed lines. These show globally-averaged mean warming ranging from about 1.5°C to 6°C by 2100.
It is noticeable that despite the immediate divergence of the future emissions, the warming until around 2040 is very similar across all four UKCIP02 scenarios. This is mainly due to the considerable inertia of the climate system; much of the warming over the next few decades is already built into the climate system due to past and current emissions. It is also partly due to the offsetting (cooling) effect of sulphate aerosols in the SRES scenarios. This means that for climate change impacts up to around 2040, the amount of emissions uncertainty is relatively small. Beyond this time, the emissions uncertainty increases rapidly with the UKCIP02 results giving only an indication of the range of possible climate futures.
|
|
| |
|
Figure 2. Global mean surface-air temperature rise estimated by the Hadley Centre CM3 model, resulting from the four emissions profiles shown in Figure 1. |
| |
(ii) Natural variability
A second source of uncertainty in the prediction of future climate is the natural internal variability of the climate system. Climate in the future will vary from year-to-year and decade-to-decade, just as it does at present, due to the chaotic nature of the climate system. For a given future period, natural variability could either reinforce or counteract a human-induced change. Since this variability cannot yet be accurately predicted or modelled, the best way to quantify this aspect of future climate is to run multiple simulations of the same model (called an ensemble), each starting with different but plausible initial conditions in the ocean-atmosphere system.
For the UKCIP02 scenarios, a three member "initial conditions" ensemble was used to analyse the potential effects of natural variability on climate futures for England and Wales (Figure 3). The results of the ensemble are averaged to reduce the “noise” of natural variability and to give a better estimate of changes in climate trends.
These indicate that although the variability associated with individual years is large (both from year-to-year for a single emissions scenario and from scenario-to-scenario for a single year), the variability associated with longer term trends is smaller (i.e. the overall trend of all three members is similar). It follows that the uncertainty in the projection of extreme events will be greater than that in seasonal or annual means.
The UKCIP02 climate change scenarios are also "packaged" into three thirty-year time-slices (2011 to 2040, 2041 to 2070 and 2071 to 2100, usually known as the 2020s, 2050s and 2080s), which represent the average future climate simulated across that time period. This further reduces the effects of natural variability relative to a climate change signal.
|
|
| |
|
Figure 3. England and Wales winter precipitation: Observed for 1965–2000 (black line) and three possible future evolutions for the period 2000–2050 for the same emissions scenario using different initial conditions (coloured lines)
|
| |
(iii) Modelling uncertainty
The final broad category of uncertainty in the modelling of future climates results from the models themselves. Different global climate models represent the ocean-atmosphere system and its many interactions and feedbacks in different ways. Each model has a different structure and each model contains different representations of important climate processes such as clouds, ocean eddies and soil moisture. Each model will therefore simulate a different global climate change and a different regional response even when forced by an identical emissions scenario (e.g. Figure 4).
|
 |
| |
|
Figure 4. Patterns of change in summer-average precipitation as simulated by nine global climate models for the 2080s (2071–2100) for the same (A2) emissions scenario.
|
| |
|
It is important to recognise that agreement between models does not show that they are correct, merely that they are telling a similar story. Conversely, predictions from the different models do not have an equal probability of representing the true outcome. However, evaluating the relative credibility of each model in an objective way is difficult to do. To deal with modelling uncertainty, it is important that users of the UKCIP02 scenarios consider results produced by other modelling centres. This is especially the case when engaged in applications where representing a wide range of uncertainty is relevant, or which are likely to lead to significant decision- or policy-making recommendations.
Considerable efforts are being made to reduce the uncertainty in predictions of future climates. The issues of natural variability and modelling uncertainty are being addressed in UKCP09 (the next set of climate change scenarios being produced for the UK by the Hadley Centre through the use of ensembles). An increasingly common method for addressing modelling uncertainty is to undertake a large number of climate simulations (i.e. large ensemble), using a model that is identical for each simulation apart from the values given to certain parameters. These are varied within plausible ranges to reveal the diversity of realistic future climates. More information about how this is being done by the Hadley Centre and others is available in the UKCP09 pages.
In addition to the uncertainties described above (which relate to limitations in our scientific understanding of the climate system and of how future greenhouse gas emissions will change), it is important to recognise that there are also uncertainties regarding the climate impacts themselves. Even if changes in climate could be accurately predicted, uncertainty would still surround the effects these changes would have on society. This is because our knowledge of impacts is based largely on experience of past events (which are not necessarily a guide to future impacts), and because this knowledge itself is imperfect.
Details about uncertainties associated with the UKCIP02 scenarios and guidance on how to deal with them can be found in the UKCIP02 documentation pages. Further discussion about the uncertainties in the UKCIP02 climate change scenarios and in global climate models in general can be found in Chapter 7 and Appendix 1 of the UKCIP02 Scientific Report and in Hadley Centre Technical Note Handling uncertainties in UKCIP02 scenarios of climate change (Geoff Jenkins and Jason Lowe, November 2003).
Whilst science aims to reduce the uncertainties associated with climate change, it will be a gradual process. Meanwhile, a risk-based approach to decision-making ensures that uncertainties are acknowledged and treated rigorously in the decision-making process. Uncertainty is not unique to climate, and uncertainties associated with other future social, economic and environmental changes may also be important for the appraisal of decision options. Climate risks need to be evaluated in comparison to non-climatic dependent risk factors to help decision-makers identify where adaptation to climate may be required. The UKCIP "risk, uncertainty and decision-making" study provides an eight-stage decision-making framework to advise this process.
The UKCIP Adaptation Wizard is designed to help users move through the process from simple understanding of climate change to integration of climate change into decision-making.
Back to top
|
|
|
|
|
|