Select Committee on Environment, Food and Rural Affairs Written Evidence

Memorandum submitted by the Met Office


  Note that research at the Hadley Centre, on which this submission of evidence is based, is largely funded by the Global Atmosphere Division of Defra, with additional resources from the Ministry of Defence and the European Commission.


  The Terms of Reference of this inquiry are concerned with the future adequacy of water supplies and with flood management. Central to both of these issues is the effect that climate change will have, and may already be having, on precipitation. This submission from the Met Office summarises the changes that have recently happened to precipitation climate, and those which are predicted to take place in the future. It demonstrates that good progress is being made in predicting detailed patterns of change over the UK. Nevertheless, substantial uncertainties remain which are a hindrance to estimating impacts and hence to efficient adaptation. The challenge for the next few years is not only to reduce these uncertainties but also to quantify them. The probabilistic predictions which will become available over the next few years will allow planners to incorporate the effects of climate change into formal risk assessments and hence plan adaptation more cost-effectively.


  1.1  Substantial changes in climate have been observed over the past century, and particularly over the past 3 or 4 decades. Since the 1970s, global temperatures have risen rapidly by about 0.5ºC. Research at the Met Office Hadley Centre has demonstrated that this rise is extremely difficult to explain other than by invoking human activities, particularly the enhancement of the greenhouse effect by fossil fuel burning. More recently, we have been able to attribute recent warming at a continental scale (for example, over Europe) to human activities. And, although this attribution cannot yet be made at the scale of the UK, it would not be unreasonable to link human activities to the warming of almost 1ºC in the Central England Temperature that has been observed over the past few decades.

  1.2  Changes in precipitation have also been observed. England and Wales precipitation has increased in winter months and decreased in summer months, and the proportion of winter precipitation which falls in heavy events has changed, too, over the past 40 years; with increases in winter months and decreases in summer. Such changes are in line with model predictions of the effect of human activities although, once again, we cannot formally attribute them to that cause. The observed changes in rainfall over the UK are larger than we would expect from the current level of warming, possibly due to a contribution from changes in circulation patterns (winds) which may or may not be associated with human-made changes in greenhouse gases.


  2.1  We expect to see climate changing in the future as emissions of greenhouse gases, particularly carbon dioxide, increase from fossil fuel burning. In order to estimate the change we use a global climate model which represents the climate system in the atmosphere, land, ocean and cryosphere; in our case the Hadley Centre HadCM3 model which is recognised to be world-leading. We drive this with scenarios of future emissions of greenhouse gases which have been developed by the Special Report on Emission Scenarios of the Intergovernmental Panel on Climate Change; these depend upon assumed changes in population, energy use, technology etc and range (in the case of carbon dioxide) from relatively little change in emissions over the next 100 years to a quadrupling of emissions. Under the lowest of these scenarios we predict a temperature rise over land by 2100 of almost 3ºC, under the highest scenario this becomes about 7ºC.

  2.2  The climate model also predicts patterns of change, in climate quantities such as rainfall, at a coarse resolution of about 300 km. In order to obtain more detail in the predictions over the UK, we "downscale" the global results to a resolution of 50 or 25 km using a regional climate model (PRECIS). Results from this were used to generate new climate change scenarios for the UK Climate Impacts Programme (referred to as UKCIP02), launched by Mrs Beckett in April 2002. The results predict that, by the 2080s, under the highest emissions scenario, rainfall over Southern England will be reduced by more than 50% in summer and over most of England it will increase by 25% or more in winter (Fig 1). For other emissions scenarios, and in earlier future periods, changes will be correspondingly less, as they will also be over more northerly and westerly parts of the UK.

  2.3  As well as changes in amounts of rainfall, we predict that the frequency of intense rainfall events will also change, such that in wintertime over most parts of the UK these will at least double. This has obvious implications for the frequency with which river and urban flooding could occur, and for the intensity of the 1-in-100 year flood. Research in collaboration with CEH Wallingford will translate the climate scenarios we have developed into river flows and river levels, so that change in flood risk can be estimated. In summertime the opposite will be the case.


  3.1  The UKCIP02 scenarios are probably the best available for any country, and are being extensively used to study the impacts of climate change on various sectors and regions of the UK. However, as is clearly pointed out in the report, but not always heeded, the scenarios contain substantial uncertainties which should be taken into account when impacts are assessed.

  3.2  The first uncertainty is due to the lack of certainty in future emissions; this is handled in UKCIP02 by showing scenarios for the full range of possible future emissions. In the event, because of the great inertia of the climate system, climate change up to the middle of the century doesn't depend much on emissions to that point, because most of the early-century change is already built into the climate system by current emissions and those over the past few decades. At the end of the century, climate change is very dependent on which emissions path the world follows.

  3.3  The second uncertainty is due to the climate models themselves, both the global one required to estimate broad-scale changes and the regional one used to downscale these to a resolution useful for impacts assessment. The models incorporate our best representation of the climate system, and in particular the interactions between different components in it (feedbacks) but this understanding is limited (largely by lack of observations in the real world atmosphere, oceans, etc) and the models themselves cannot even represent this limited understanding (in part because of the limitations of supercomputing power and hence the resolution of the global model).

  The change in annual global average rainfall predicted by the 11 models shown in the last IPCC report ranges from 1% to 9%. On the scale of the whole UK, the situation is worse: in wintertime the increase ranges from 1% to 60%. And for East Anglia it ranges from a small decrease of a few percent to an increase of over 50%. Although the Hadley Centre model has been extensively validated and is a world-leader, and its predictions fall roughly in the middle of the range from all models, we cannot formally claim that its predictions are any more or less likely to be correct than any of the others.

  3.4  Faced with this uncertainty, the flood defence planner is put in an unenviable position; if he plans on the basis of the least sensitive model his defence costs will be small or even zero but he runs the risk of major flooding if the most sensitive model turns out to be close to reality; if he plans for the worst he could waste billions of pounds on protection which turns out not to be needed. At present we cannot even rank the various model predictions in terms of credibility.


  4.1  What planners really wants is not several deterministic predictions of unknown relative quality, but a probabilistic prediction of change, which tells them the probability of different levels of change which can be factored into formal risk assessments. So, for example, this might say that the probability of winter rainfall in SE England increasing by 10% or more by 2050 is 90%, the probability of it by 30% or more is 15%, etc (values given are illustrative). For the first time, this is now possible through the use of model ensembles. The technique involves taking a climate model and changing within it one of the representations of processes in the atmosphere, ocean, etc to make a new variation of the climate model. This is then run to predict changes (for example, in SE England winter rainfall by 2050). A further process in the model is changed and a different prediction ensues. By doing this a large number of times, rating the credibility of each of the models by reference to its performance at simulating the current climate, and combining all the predictions, we can form a probability distribution of change. In the Hadley Centre this has been started using 50 variations of the climate model, and Fig 2 shows the initial probability predictions that have been made. These are not yet useable, as they only include an initial selection of representation uncertainties, but they will develop over the next 2 or 3 years to the point where they do provide useable information to planners. Hence one valid option for planners would be to delay decisions until they have better information to work with.

Fig 2. Preliminary probability predictions of % changes in South-East England precipitation in summer (left) and winter (right) by the 2080s under a Medium-High emissions scenario.

  4.2  Up to now, these probabilistic scenarios have been generated using global climate models, which have a poor resolution (300km) and do a poor job of estimating changes in extremes. Both these limitations are overcome by using regional climate models (RCMs) as mentioned earlier, and the goal is to have high-resolution probabilistic predictions by the time of the next set of climate change scenarios for UKCIP in about 2007.


  5.1  The last main uncertainty in prediction future climate is natural variability. We know that, because interactions between the atmosphere and oceans makes the climate system a chaotic one, climate varies widely from one year to the next unconnected with any human influence. The UK, because it is at the end of Atlantic storm tracks and at the interface between continent and ocean, has a very variable climate, and for precipitation this variability is likely to be greater than any underlying human-made change over the next few decades. Until recently it was taken for granted that the variability was unpredictable, but recent research is indicating that some useable predictability exists, and useable forecasts over the next decade ("decadal" forecasts) will be possible in the future. This is partly due to developments in modelling, and partly due to the increasing availability of good observational data from the oceans (which, because of their inertia, determine the state of the climate over the next few years). Hence, any planning for flood defences or water availability should also take account of decadal forecasts once they become available.


  6.1  Global warming due to human activities will cause the sea level to rise, mainly due to the expansion of ocean waters and the melting of land ice. There is a wide range of uncertainty in the global average rise over the next 100 years, ranging from 10cm to 90cm. The rise will not be uniform across the globe, and local effects add to this uncertainty. Flooding mainly occurs when storm surges, produced by a combination of high winds and low pressures, hit vulnerable coastlines. Surge heights for which coastal defence will have to cater will change both by rising sea levels and changes in storminess; the latter is notoriously difficult to predict. Nonetheless, modelling done for the UKCIP02 report showed that even a models rise in sea level, of 30cm by the 2080s, coupled with changes in land levels, would change the height of the typical "50-year" storm surge by up to a metre or more in some parts of the coast, with other parts seeing 30 or 40cm. Again, uncertainties in sea level predictions are substantial, arising from both the climate model and the storm surge model. Probability predictions will be made, on a somewhat longer timescale than those of precipitation and river flows.

The Met Office

April 2004

previous page contents next page

House of Commons home page Parliament home page House of Lords home page search page enquiries index

© Parliamentary copyright 2004
Prepared 16 September 2004