Scottish Road Network Climate Change Study: UKCP09 update Autumn 2011
2 Overview of the UKCP09 scenarios
The UK Climate Projections (UKCP09) provide projections of climate change for the UK, giving greater spatial and temporal detail, and more information on uncertainty, than previous UK climate scenarios. The following sub-sections highlight information of particular pertinence to this project. For a fuller understanding refer to the UKCP09 technical reports (Murpy JM et al, 2009; Jones PD et al, 2009).
2.1 Purpose and design of UKCP09
Over land, UKCP09 gives projections of changes for a number of climate variables, averaged over seven overlapping 30-yr time periods, at 25 km resolution and for administrative regions and river basins. Similar projections are given for a smaller number of variables averaged over marine regions around the UK.
UKCP09 is the first set of UKCIP[1] projections to attach probabilities to different levels of future climate change. The probabilities given in UKCP09 represent the relative degree to which each climate outcome is supported by the evidence currently available, taking into account our understanding of climate science and observations, and using expert judgement.
The Met Office Hadley Centre has designed a methodology to provide probabilistic projections for UKCP09, based on ensembles of climate model projections consisting of multiple variants of the Met Office climate model, as well as climate models from other centres. These ensembles sample major known uncertainties in relevant climate system processes.
2.2 Emission scenarios
Unlike the UKCIP02 scenarios which had 4 emission scenarios (High, Medium-high, Medium-low, Low) the UKCP09 scenarios only use 3 (High, Medium, Low). Each of the UKCP09 emission scenarios suggests a different pathway (storyline) of economic and social change over the course of the 21 century. Changes in population, economic growth, technologies, energy use and land use are all considered in the determination of the emission scenario. They do not assume any planned mitigation measures and importantly, they cannot currently be assigned probabilities. The following boxes describe the types of futures assumed.
High emission scenario storyline:
Describes a future world of very rapid economic growth with a population that increases from 5.3 billion in 1990 to peak in 2050 at 8.7 billion and then declines to 7.1 billion in 2100. Rapid introduction of new and efficient technologies is assumed, as is convergence among regions, including large reductions in regional differences in Gross Domestic Product (GDP). High use of fossil fuels is assumed.
Medium emission scenario storyline:
As for the High emission scenario, but with reduced reliance on fossil fuels and the use of non-fossil energy sources.
Low emission scenario storyline:
Describes a convergent, more equitable world and has the same population scenario as the above storylines. However, rapid changes in economic structures towards a service and information economy are assumed, with reductions in material intensity and the introduction of clean and resource efficient technologies. Global solutions are found to economic, social and environmental sustainability.
Given that probabilities are not assigned to the 3 emission scenarios it is left to the user to consider the merits of using the different scenarios. All 3 emission scenarios are presented in this report.
2.3 Projection uncertainties
Uncertainty in climate change projections is a major problem for those planning to adapt to a changing climate. Adapting to a smaller change than that which actually occurs (or one of the wrong sign) could result in costly impacts and endanger lives, yet adapting to too large a change (or again, one of the wrong sign) could waste money. In addition there is the risk of maladaptation - adapting to climate change in a way that prevents or inhibits future adaptation. The 2009 projections are the first from UKCIP to be designed to treat uncertainties explicitly, by generating projections of change that are given, where justified, as estimated probabilities of different outcomes (see Appendix A for a fuller description) rather than giving a single realisation of possible changes from one model or a small sample of possible changes from several models. This means that probabilities are attached to different climate change outcomes, giving more information to planners and decision makers.
Uncertainty in projections of future climate change arises from three principal causes:
- natural climate variability;
- incomplete understanding of Earth System processes and their imperfect representation in climate models;
- uncertainty in future emissions.
Uncertainty in projections is presented in the UKCP09 in the form of probability density functions which are developed by repeatedly running the models with each run having plausible but not identical parameter values or initial states (ie. a model ensemble). The outcomes are not identical and offer a spread of estimates. If enough runs are undertaken that adequately sample the range of plausible states then a general picture will develop showing a range of projections with some more common than others (ie. a probability density function). Appendix A provides an extract from the UKCP09 report (Murphy JM et al, 2005) that describes probability density functions in more detail.
The progression to probabilistic projections based on large ensembles has meant that not all of the properties and characteristics of the UKCIP02 scenarios could be carried across to UKCP09 - the direct provision of daily time series from climate model output, for example. Thus the new projections are not a "drop in" replacement or straightforward update of UKCIP02.
Since probabilities have not been assigned to the emission scenarios the UKCP09 probabilistic projections cannot include this aspect. Hence separate Probability Density Functions are presented for each of the three emission scenarios. Differences between the three emission scenario projections are small over the next two or three decades mainly because of climatic system inertia, but will be substantial in the second half of the century.
2.4 Projections at a daily resolution
UKCP09 provides a tool known as a weather generator, capable of providing plausible realisations of how future daily time series of several variables could look. It does not provide a weather forecast for a particular day in the future; rather it gives statistically credible representations of what may occur given a particular future climate.
The UKCP09 Weather Generator provides synthetic daily time series of temperature (mean, maximum and minimum), precipitation, relative humidity, vapour pressure, potential evapotranspiration (PET) and sunshine at a resolution of 5 km, for each of the three emission scenarios. The weather generator does not add any additional climate change information over that which is present in the 25 km probabilistic projections. However it does add local topographical information (e.g. hills, valleys) at the 5 km scale, as it is based on observed data which is representative of this scale. For more detailed information on the UKCP09 weather generator refer to Jones PD et al (2009).
2.5 Confidence in climate projections
There is a cascade of confidence in climate projections. There is very high confidence in the occurrence of global warming due to human emissions of greenhouse gases. There is moderate confidence in aspects of continental scale climate change projections. 25 km scale climate change information is indicative to the extent that it reflects the large scale changes modified by local conditions. Higher confidence is attached to longer term average values such as annual values compared to variables relating to short time-step events such as daily or sub-daily. The confidence in the climate change information also depends strongly on the variable under discussion. For example, there is more confidence in projections of mean temperature than for mean precipitation. The probabilities provided in UKCP09 quantify the degree of confidence in projections of each variable, accounting for uncertainties in both large scale and regional processes as represented in the current generation of climate models. However, the probabilities cannot represent uncertainties arising from deficiencies common to all models. The fact that the UKCP09 projections are presented at a high resolution for the UK should not obscure this and users should understand that future improvements in global climate modelling may alter the projections as common deficiencies are steadily resolved.
2.6 Gaps within the UKCP09 projections
UKCP09 projections do not provide information on projected changes in wind speed or snowfall.
It is also worth noting that the UKCP09 scenarios were unable to adequately resolve storm tracking or anticyclones and are unable to represent localised convective activity in the atmosphere that is responsible for heavy summer downpours or thunderstorms.
2.7 Interpretation of probabilistic projections
A key advantage of the UKCP09 climate change scenarios is that they better describe the uncertainties attached to the predicted changes. Earlier sets of scenarios gave a single value for a certain emission scenario, whereas the UKCP09 effectively provides a probability distribution of likely change in the future. This is a much more 'honest' and informative means of relaying what is likely to happen since there are inevitably uncertainties in the process of generating the projections and the selection of just one modelled representation may not be representative.
Figure 2.1 presents a case example of the probabilistic nature of a typical UKCP09 projection. The example is for the 10-year daily rainfall for the 2080s time horizon compared to that of the baseline 1961 to 1990 period. The graph is effectively a histogram of predicted 10-year rainfall depths derived from numerous equally plausible simulations both for the future and separately for the baseline period. Because there are uncertainties within the modelling process and therefore, each predicted 10-year rainfall depth can differ from the others. If a large enough sample of estimates are sampled, then the frequency (or likelihood) of a certain value can be obtained from such a graph. For example, the median estimate (the 50 percentile estimate) is indicated in the figure. For this middle estimate there is an equal chance that the actual value may be larger or smaller than this value. For the 10 percentile estimate there is only a slim chance (a 10% chance) that the actual value will be less than this value and a strong chance (90% chance) that it will be larger than this value. The 90 percentile value indicates that there is only a slim chance (10% chance) that the value will be larger, and so on.
Figure 2.1 - Example of the probabilistic nature of the modelled projections for the 10-year daily rainfall for both the baseline period (1961-1990) and for the 2080s time horizon. (The graph is effectively a frequency diagram of rainfall depths predicted from numerous plausible model runs. The percentage values indicate the position within the distribution of the 10, 50 and 90 percentile values)
The UKCP09 project has standardised upon the use of the 10, 50, and 90 percentiles and attaches the wording that projections are unlikely to be less than the 10 percentile value and unlikely to be greater than the 90 percentile value.
Projected changes that are made with confidence have narrowly defined probability functions. These tend to be for averages of climatic variables that are determined over longer time steps, such as average annual temperature or average annual precipitation. Wide probability functions tend to result for more difficult to model variables such as extreme daily variables (eg. 10-year daily rainfall depth). Uncertainties and hence the width of the probability functions will also broaden the further away from the baseline period. Examination of the uncertainty ranges provided in the tables of this report will demonstrate differing scales of uncertainty for different variables.