Select Committee on Environment, Food and Rural Affairs Minutes of Evidence

Memorandum submitted by the Met Office (FL 120)


  1.  Exceptional summer precipitation was well forecast. Individual rainfall events over this summer were exceptional while the accumulated totals over the period were unprecedented. The weather events were well forecast, indeed the weather forecasts leading up to the major July flooding event were the most accurate and detailed provided for any major flooding event in the UK.

  2.  Current warning mechanisms and structures have led to confusion. Responsibility for issuing severe weather warnings lies with the Met Office, while river (fluvial) flood related warnings are a matter for the EA. There has been some confusion amongst user groups particularly when severe weather events and consequent flooding occur at different times and places. This communication issue has been identified in some regional lessons learned exercises.

  3.  Improved weather forecasts will lead to improved flood forecasts. Advances in probabilistic forecasting techniques combined with the operational use of higher resolution models will give higher confidence in the broader pattern of weather and the precision in the detail. Although a very high resolution model was used to significant effect during the July flood event, this is not an operational system. Operational implementation awaits enhanced supercomputer power.

  4.  Integrated models will increase the effectiveness of warnings. The Met Office is developing its range of ensemble based forecasting techniques to provide a probabilistic framework within which to deliver enhanced decision support tools to government agencies. Development of these tools will require close cooperation between agencies[5]. Integration of ensemble weather forecast and river models would provide more reliable river flood forecasts at extended lead times.[6]

  5.  Climate change may mean we see more intense severe weather events. Any specific individual weather event cannot be attributed to climate change. Broadly speaking, climate change in the UK is predicted to produce wetter winters and drier summers with the potential for increasing intensity of severe weather events.

  6.  The past is no longer an adequate guide to the future. There is significant uncertainty over the impact of climate change on flooding events and the probability of them occurring over any given period. As climate research develops, and higher resolution models can be run, there will be increased certainty about regional scale impacts and its effect on flooding. This is important because, under a changing climate, the use of return periods[7] (based on previous climatology) for deciding on national infrastructure may not be the most appropriate mechanism; decisions should draw heavily on the best available regional climate predictions.


  7.  The Met Office is a Trading Fund Agency owned by MOD. It is a world leading organisation, both in the field of weather forecasting and climate prediction, operating on a 24/7 basis with the highest standards of operational resilience. Responsible for providing forecasts on all timescales (from an hour ahead to 100+ yrs), the Met Office is uniquely positioned to support the UK's response to changing incidence of severe weather events due to climate change. It is responsible—through its Public Weather Service—for providing the National Severe Weather Warning Service for the whole of the UK.

  8.  Unlike many other nations in Europe where weather and flood forecasting and warning are integrated into a single meteorological and hydrological agency, the Met Office has no direct responsibility for river or coastal flood forecasting.

  9.  In England and Wales, river and coastal flood forecasting are the responsibility of the Environment Agency with whom the Met Office works closely, providing to them storm tide warning services, weather forecast information (daily) and weather warnings (6 hours before an event). Currently no agency provides warnings of localised flash flooding (also known as pluvial flooding), although the Met Office will give an indication of such risks alongside its weather forecasts.

  10.  Similar responsibilities for river and coastal flooding apply to SEPA in Scotland. There is neither a system nor single delegated authority for issuing flood warnings in Northern Ireland.


  11.  The Met Office routinely utilises three atmospheric forecast models over land: a Global model at 40 km resolution; a North Atlantic and European (NAE) model at 12 km resolution, and; a fine scale UK model at 4 km resolution. The regional models are deployable and can be run for any location in the world in support of military and commercial customers.

  12.  In addition, the Met Office can also run a model at 1.5 km resolution, which provided significant benefit during the July flood. However, given current supercomputer power, this must be restricted in frequency and carefully scheduled to avoid impacting on both the capability required to run the operational models and on computer time allocated to research in weather forecasting.

  13.  The Met Office has also introduced a probabilistic system of forecasting using multiple model scenarios to simulate and quantify uncertainty. This system currently has a 24 km resolution.

  14.  The data used to initialise numerical weather models comes from various sources—including direct observations of the atmosphere, oceans and land surface conditions, as well as previous model runs. Both the Met Office and the EA utilise information from rain gauges and from radar observations in rainfall and flood modelling respectively. However, ownership and responsibility of rain gauges across the UK is split between the agencies and consequently not all of the information is available to either party in real time.

  15.  The Met Office leads and hosts the National Centre for Ocean Forecasting (NCOF) and utilises a suite of models, including a surge model and a wave model, to provide forecasts of tides and storm surges to Government agencies, particularly the EA and Marine Coastguard Agency, as well as to commercial customers.

  16.  The Met Office Hadley Centre, funded through Defra and MOD, provides the best climate prediction science in the world[8] which supports the UK Government's high profile policy role on Climate Change issues.


  17.  The unusual pattern of weather experienced this summer occurred as a consequence of the location and strength of the jet stream[9] combined with unusually high Atlantic sea temperatures. For much of this summer the jet stream was further south and stronger than is typical, resulting in many weather systems crossing southern and central parts of the UK. These depressions have been more intense with some interacting with the very warm and moist air to the south, generating exceptionally heavy rainfall events.


  18.  The cumulative rainfall total in the UK for May, June and July 2007 was unprecedented; Met Office records show that 414.1 mm of rain fell across England and Wales, making it the wettest May to July since the England and Wales Precipitation record began in 1766.

  19.  The exceptionally heavy rain culminated in two severe, disruptive flooding events.

    a.  On 24-25 June, a deep and slow-moving area of low pressure brought a prolonged period of heavy rain causing widespread flooding in parts of Yorkshire and the Humber, Derbyshire, Lincolnshire and Worcestershire. The flooding was a result not only of the heavy rain during the 24-25 June, but also the antecedent conditions of a wet May and June, especially the heavy rain during the 12-14 June, which caused the ground to be saturated and water levels to be high prior to this event.

    b.  The second flooding event was the result of exceptionally heavy rain which moved northwards across the UK from late on 19 July throughout 20 July, the heaviest being reported at Pershore College (Worcestershire), with Met Office data showing 157.4 mm recorded in the 48 hour period. Widespread flash flooding occurred on the morning of 20 July across Southeast England, and later in the day across the Midlands as the system moved north-westwards, causing widespread disruption to the motorway and rail networks. River flooding followed over the subsequent days along the Severn and its tributaries in Gloucestershire, Worcestershire, Herefordshire, Shropshire, and along the Thames and its tributaries in Wiltshire, Oxfordshire, Berkshire and Surrey. More detailed rainfall statistics are at Annex A.


  20.  The Met Office issued an Early Warning through the National Severe Weather Warning Service (NSWWS) on 22 June which gave 3 days notice of potential disruption[10]. An update issued on 24 June focused correctly on the worst hit areas and provided highly accurate estimates of rainfall totals correctly forecasting that 50-100 mm of rainfall would fall within 24 hours of 2200 BST on Sunday with the worst affected areas being Yorkshire and Lincolnshire.

  21.  Forecasting of the 19-20 July rainfall event was perhaps the most detailed and accurate ever achieved by the Met Office for a high profile severe weather event. To be able to accurately forecast 100 mm of rain at county level with 24 to 36 hours lead time is an indication of the process wide improvements that have recently been made. Significant use was made of multi model probability forecasting techniques, while the "on demand" high resolution (1.5 km) forecast model was run for parts of the affected area to give more detailed forecast information. The potential for excessive amounts of rain on Friday 20 July was communicated to the EA and other customer groups, together with the uncertainties. Confidence was considered high enough to issue an NSWWS Early Warning on Wednesday 18 July.

  More detailed timelines of Met Office activities prior to the flooding events on the 24-25 June and 19-20 July are at Annexes B and C.

  22.  During the most intense rainfall on 20 July, a technical fault arose with the radar rainfall measurement system. The result of this was a degradation in the quality of the quantitative rainfall estimates issued to the EA for the areas of heaviest rain. The fault has now been corrected, and re-analysis of the data is underway to confirm the scale of the impact, in consultation with the Environment Agency and other users of the data.


  23.  As a Trading Fund the Met Office places a strong emphasis on meeting customer requirements. Its core mission is to use the best science to develop and tailor services that allow them to make the best informed decisions possible. Due to its role as the National Weather Service and the body responsible for the National Severe Weather Warning Service it has very strong links with UK's emergency response infrastructure, its customers (both government and private sector) and the general public. It has well established and trusted mechanisms for communicating warnings in particular through its website and its partnership with the BBC. During the flooding events the Met Office provided support to numerous organisations including the EA, HPA, HA, Network Rail, local authorities and Gold Commands, as well as attending Civil Contingency meetings at COBR.

  24.  Although communication of warnings did take place successfully, there was evidence that the underlying message was confused. This occurred for a number of reasons:

    a.  individuals and agencies can confuse severe weather warnings and flood warnings;

    b.  some organisations or individuals look to one organisation for warnings—on occasions customers expected flood briefings as well as weather warnings from the Met Office;

    c.  although severe weather and flooding are linked they do not necessarily occur at the same time (for example the Met Office web site showed no weather warnings at a time when parts of the UK were experiencing severe flooding).

  25.  Although lessons learned exercises undertaken by regional authorities highlighted Met Office forecasts as being of high quality, it was generally acknowledged that the fragmented responsibilities for warnings hinder understanding and therefore response.


  26.  Research in the Met Office is focussed on improving all aspects of our forecasts, primarily through improvements to the performance of our numerical weather prediction models. In line with our understanding of public and customer requirements, work is currently focussed particularly on two major areas of research:

  27.  Higher resolution forecasting. The purpose of high resolution modelling is to enable forecasts to become more precise about the weather expected in any locality. The results of our research are extremely promising. The graphic below shows the potential benefit of higher resolution models by applying them retrospectively to the major flooding event in Carlisle in January 2005. The figures show how an increase in forecast model resolution from 12 km (b) to 1 km (c) provides a much increased improvement in accumulated precipitation when compared to observations (a).

  28.  Probability forecasting. Due to the chaotic nature of the atmosphere, the further in advance forecasts are made, the more small scale weather features become unpredictable. How severe this unpredictability becomes is dependent on the particular weather pattern and location. Consequently, although we can provide a "best" forecast it is not possible to indicate how likely this really is. This is particularly important for severe events where the chosen response is likely to vary with the confidence associated with the forecast. In response to this problem we have a research programme aimed at forecasting the probability of certain weather events occurring—through running "ensembles" of individual forecasts. We are now looking towards combining this approach with the high resolution model to look at forecasting the probability of intense local rainstorms.

  29.  Both of these research advances will lead to significant improvements in the information available from Met Office forecasts. Greater use of probabilistic forecasting techniques in principle allows more informed decision making—particularly for emergency response organisations. For example, a 25% chance of an event occurring may require a different response to a 75% chance. In reality, probability increases as an event gets closer, so the customer response process should be able to react to changing uncertainty. We will continue to work closely with customers and other Government agencies to ensure the additional information available is used to the best effect.

  30.  These research advances require substantial increases in computer power before they can be implemented in routine operational forecasting. In 2009 we expect to be able to purchase a new supercomputer with sufficient power to implement a UK scale 1.5 km high resolution forecast model and a small ensemble model with a 12 km resolution. The combination of high resolution modelling and probabilistic ensemble forecasting is very important—neither approach alone will give both confidence in the predicted weather nor the precision in the detailed location.

  31.  Further funding would be needed to fully realise the benefits (better pinpointing of areas at risk) that can be delivered through operational use of a 1 km ensemble.


  32.  With the recent severe flooding in the UK this summer, climate change and what this may mean for extreme rainfall in the future, is very much at the forefront of people's mind. Although any specific individual event cannot and should not be attributed to climate change, we are able to make statements about the risk of such events altering as a result of climate change. It seems likely there will be a greater risk of heavy rainfall across the UK in the future, particularly in winter.

  33.  The rainfall that caused the flooding in the UK this summer was as a result of weather systems occurring as part of the natural variability of the climate. The overall weather pattern is broadly consistent with conditions during previous La Nina[11] events, although with the low pressure situated slightly further south. This weather pattern is not associated with climate change. Nevertheless, the amount of rainfall could have been larger because of climate change. In particular, warm sea surface temperatures in the vicinity of the UK this summer probably contributed to the high levels of atmospheric moisture.

  34.  A change in the proportion of summertime rainfall falling as heavy downpours would be likely to have a significant impact on flooding; short period intense events tend to cause local flash floods. However, for flooding over larger areas, longer period rainfall is required, with antecedent conditions being very important. In general, models suggest that a possible signature of global warming may be relatively greater increases in rainfall for those extremes which are rarest and of shortest duration (ie the most intense).


  35.  Although there is a general consensus on the broad features of expected climate change, there are still uncertainties, particularly when considering how the climate may change locally. Although climate models capture the key processes identified as important for climate change, it is not possible to represent the full complexity of the climate system. Generally we use the ability of a model to reproduce the climate of the recent past as an indicator of its likely skill in predicting the future.

  36.  Natural variability of the atmosphere is a further source of uncertainty. Natural variability leads to our familiar daily weather patterns and on longer timescales to phenomena such as El Nino. Future projections will in part reflect natural variability and it is important to distinguish this from an underlying shift in the climate caused by increased greenhouse gases. In addition, natural variability is a significant factor when considering extreme events on a local scale.

  37.  Although we are able to make confident statements about increases or decreases in extreme precipitation in some regions, the magnitude of these changes remains uncertain. This is particularly pronounced over central Europe and the UK in summer as these areas are in a transition zone between increased and decreased precipitation.


  38.  Under a warming climate the probability of extreme heavy rainfall events—and hence flooding—is likely to change, and this will become evident over the coming decades. In an ideal world investment in flood defences would be derived from accurate models of future climate. However currently there is a great deal of uncertainty about the regional impacts of climate change—particularly at the local scale and with regard to precipitation. For example, although we may be confident of increases in extreme rainfall across the UK in winter, the actual magnitude of these increases is uncertain. Understanding how extreme rainfall may change in detail locally in the future remains a key challenge for climate scientists. In time, with the ability to run higher resolution models, improved science and the investment of resources in large ensembles of models sampling uncertainties, improved regional scale predictions will be available.

  39.  The development of a national flood defence infrastructure has, to date, placed a heavy emphasis on return periods derived from historical records. However it is crucial to understand that the past is no longer an adequate guide to the future; large-scale investment in flood defences and other parts of the critical infrastructure would be much better derived from improved regional scale predictions. In the context of recent flooding events, increased supercomputing power beyond that currently planned would also lead to improved weather forecasts, better regional detail, better input to flood models and improved warning lead time.


  40.  Against this background, the Met Office has prepared a proposal for increased supercomputing power for consideration within the context of the Comprehensive Spending Review (CSR). The proposal would provide around 20 times the current computing capability by 2009 rising to around 40 times by 2012.

Met Office

September 2007

5   A number of studies are taking place to evaluate options for expanding the scope of flood warning services as well as using probabilistic and higher resolution forecasts. Back

6   Current flood modelling techniques rely on observed (rather than forecast) precipitation, but rainfall forecasts could be used to better effect, particularly for rapid response flooding events. Back

7   A return period denotes a recurrence interval. It is a statistical measure of how often an event of a certain size is likely to happen. For many of the recent rainfall events return periods were greater than 200 years. Back

8   An independent review of the Met Office Hadley Centre from Risk Solutions, commissioned by Defra and MoD, was published by Defra on 15 May-it concluded that the Hadley Centre was at the pinnacle of world climate science. The review is available on the Defra website: Back

9   A ribbon of very strong winds in the upper atmosphere which largely determines where the weather systems that bring rain to the UK will develop and move across Western Europe. Back

10   As opposed to likelihood of rain-confidence of heavy rain time of issue was 100%. Back

11   The opposite of El Nino, which occurs naturally every 3-7 years. Back

previous page contents next page

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

© Parliamentary copyright 2008
Prepared 7 May 2008