Science & Technology - Minutes of EvidenceHC 1538

Back to Report

Oral Evidence

Taken before the Science and Technology Committee

on Wednesday 2 November 2011

Members present:

Andrew Miller (Chair)

Gavin Barwell

Stephen Metcalfe

David Morris

Stephen Mosley

Pamela Nash

Graham Stringer

Roger Williams

________________

Examination of Witnesses

Witnesses: Nick Baldwin, Independent Chairman, Public Weather Service Customer Group (PWSCG), Professor Sir Brian Hoskins CBE, Chair, Met Office Scientific Advisory Council (MOSAC), and Professor John Pyle, Chair, Met Office Hadley Centre (MOHC) Science Review Group (SRG), gave evidence.

Q41 Chair: Welcome, everyone. I welcome in particular the Royal Society peers who are sitting in the audience. I hope that they find it a fruitful sitting. Will the first three witnesses kindly introduce themselves?

Nick Baldwin: I am Nick Baldwin, the independent chair of the Public Weather Service Customer Group.

Professor Hoskins: I am Brian Hoskins. I am at Imperial College and the University of Reading. I am also a non-executive director of the Met Office and chair of its scientific advisory committee.

Professor Pyle: I am John Pyle from the University of Cambridge and the National Centre for Atmospheric Science. I chair the Met Office Hadley Centre Science Review Group.

Q42 Chair: Thank you very much indeed. May I start with you, Mr Baldwin? In your role, you are both intelligent customer, and watchdog and guardian of the Public Weather Service. Is that a difficult balance to achieve?

Nick Baldwin: I don’t think that it is a difficult balance to achieve. Perhaps it would help if I elucidate each role. We are the intelligent customer on behalf of the Government and the general public for free at the point of use weather forecast and information services. We have a group of people who are mostly from the resilience community, and are therefore experts in the area of the services that we are buying and using. As a customer, we are responsible for buying service from the Met Office through a customer-supplier agreement, which essentially defines the outputs we get. It defines performance measures so that we know that we are getting what we paid for. It defines a price, and it also sets the Met Office a 3% annual efficiency target to reduce the costs of the service that it provides. I am supported by a very small secretariat, which keeps in day-to-day contact with the Met Office to ensure that it is aware of developments as the informed customer.

As a watchdog, we conduct consultations with both the general public and our users to ensure that we are getting what we ask for, and that we are aware of the needs for future developments in certain areas. The final part of our role as the guardian leads, in part, to the inquiry of the Committee in that we hold the funding of the Public Weather Service. We’re the ones who sign off the invoices and ensure value for money, and we’re doing that to ensure that the Met Office maintains its core underpinning of its operational capacity and its research and development programme. So I think we are well established to conduct those three roles.

Q43 Chair: You set the key performance indicators, and I understand that last year the PWS achieved all of them. Does that mean that the indicators were not challenging enough?

Nick Baldwin: That’s an interesting debate, which we have at every one of our meetings as we go through the year. We are running a rolling scorecard in the course of the year, using the traditional traffic light colours of red, amber and green. We tend to find that we start with a number of reds and ambers and a proportion of greens and, as the year goes on, the focus is brought to bear on all those that are not hitting their milestones or are not at the level of quality that is expected. It is for the Met Office to hit those targets and it’s up to the Met Office to reallocate its resources to ensure that it does hit the targets. We put the pressure on it to make sure it gets there, rather than this being a case of soft targets.

Q44 Chair: Notwithstanding any external pressures, what would be on your wish list to add to the pressures that you’d like to put on the system?

Nick Baldwin: I don’t have a wish list of additional pressures that I would like on the system. I have a wish list of what I want more from them-

Q45 Chair: What improvements do you want?

Nick Baldwin: I want more accurate forecasts and longer lead times on severe weather events, because those are the two fundamental things that we’re about: providing accurate and timely forecasts and ensuring that severe weather events are forecast in advance so that the resilience community can prepare for them and take action.

Q46 Chair: Why can’t you have that?

Nick Baldwin: It takes time, basically. The Met Office has a scientific research programme, which is all the time improving its methodologies of forecasting. In the time that I’ve been doing this job-since 2007-we’ve been able to see improvements in forecast accuracy. But my understanding-my illustrious colleagues next to me will have a better understanding-is that we are dealing with emerging research and cutting-edge technologies in the forecasts that we are producing, so I do not feel that the Met Office is falling behind; it is leading the way.

Q47 Chair: Some of us who were at the Met Office yesterday heard about the expansion of the site-specific forecasts up to 5,000. Has that led to improved forecasting at local level and, all other things being left out of the equation-just looking at the objectives of the service rather than dealing with costs and so on at the moment-do you think that should be further expanded?

Nick Baldwin: We have encouraged the Met Office to move from what was, only a few years ago, 400 sites to 5,000 sites. It will soon increase that again with further forecasts for beaches around the country-an area where there are particular safety concerns. That is on the back of its increased computer capability. So we’ve moved quite a long way forward in being able to provide much more local information for people. We’re down to the resolution of the computer programmes now to enhance the position further.

Q48 Roger Williams: Sir Brian, in your written submission, you said that the Met Office was very well placed to pursue the seamless approach to modelling. Could you tell us first what the benefits of that approach are?

Professor Hoskins: I am very happy to do so. I should perhaps declare my bias to start with, because I think I created the name "seamless" in the international sphere. The idea is that the atmosphere knows no particular bounds at any particular time scale. There are phenomena that occur on daily time scales and weekly. I’m thinking of the low pressure systems and the blocking highs. These things occur on all sorts of time scales. Equally, the models that we use to look at those phenomena have very much common ingredients, so there’s a lot to be gained by using, as much as possible, the same system to look on all sorts of different time scales. You can evaluate a model that may be used mainly for seasonal, but make sure it works well with a daily weather system, because if it doesn’t, can you trust it for the seasonal? Then, by looking at a model on a seasonal time scale, you can perhaps learn something about the land surface that turns out to be useful for a few days. The idea-from even below a day right through to a century-is that there is no boundary between the phenomena and the models used. A lot of the techniques and information are much better if you are using a common system and can learn from one scale to another.

The Met Office is uniquely well placed to deal with this. I don’t think there are any other centres around the world that have developed in terms of the weather and the climate to the extent that it has. I was not pushing this on an international scale because my home institution-the Met Office in the UK-was going to be the one to really flourish, but by accident or design, the Met Office is in a great place to do this. It has the same basic model in terms of the atmosphere that it uses for less than one day out to a century.

Q49 Roger Williams: Thank you for that. Professor Pyle, you have described the success of the MONSooN project in facilitating joint development. Can you tell us what the barriers are to expanding this project?

Professor Pyle: Let me just clarify. First, MONSooN is a facility outside the firewall of the Met Office computer. It is used to develop work co-jointly by the academic community and the Met Office. Prior to MONSooN, when we did develop work together, it was very slow, very laboured and not particularly efficient, so MONSooN has made a huge difference to the academic community’s ability to contribute to what the Met Office delivers to Government. Equally, of course, there have been benefits to the academic community.

There are strategic issues, of course, as to the direction of computing in academic circles, and those are important, but financial issues constrain what happens. Initially, MONSooN is a relatively small additional component to the Met Office computer and is paid for jointly by the Met Office and the Natural Environment Research Council. I think NERC is contributing a little less than £1 million a year. If you want to make it bigger, you have to put more money in, and the question is where the money comes from-that is the issue. The Met Office will upgrade a computer next year, and MONSooN will become bigger along with that, but I suspect that the demand, particularly from the academics, which, as I say, helps the Met Office, will not be met by that service. So this is largely a development service; it is not being used in large part to do very long integrations-it is about proofing code.

Q50 Roger Williams: Can you tell us how the Met Office could make sure that its models are more user-friendly?

Professor Pyle: Again, I think it is a question of somebody being able to find the resource to make models more user-friendly. The reference I made was to mesoscale models, which Professor Hoskins has just mentioned, in this range. Some of the academic community have found using mesoscale models quite difficult. In consequence, they have been picking up codes from north American sites, which probably are not as good, but are easier to use, and the academic, like everybody else, has targets to meet, and wants to publish papers. A number of people in the UK community have gone down that route.

My own feeling is that we ought to be investing substantially in the unified model-the Met Office’s model-as an academic community, but that is a little more difficult to do. Because of that, some people vote with their feet. Essentially, you would need to devote more effort, more people and more resource to making those models portable and therefore usable by the academic community.

Professor Hoskins: May I come in on that? It is worth looking at the history on this. The Met Office provides an operational effort for the UK. There are places in the world whose aim has been to produce a community model, so those places that have the community models-in general, those are developed to be easy to use. The Met Office model started off being for the operational purpose; what we are seeing is an attempt to move across to make them more like those community models. But there is an overhead with it: if you have something that is only being run for today’s weather forecast all the time, it comes streamlined for that and it just runs like that. If it is going to be used by all sorts of users, it has to be maybe slightly less efficient at that, and it is a lot of work to make it more usable.

Q51 Roger Williams: What more could the Met Office do to ensure that it has access to more reliable data to put into the model?

Professor Hoskins: There is a continuing programme of looking for observations that are both different and more accurate on all time scales. The MOSAC committee that I chair, for instance, when the 1.5 km model-a very detailed model of the UK-was coming forward, put pressure on to say, "Do you have the right observational capability to make the most of that model?" As always, one would like more and more observations, but there is a question of getting the right mix so that, together with the model, you can actually interpret what is really there in the atmosphere and it is enough for initial data.

All the time the continuing satellite programmes are incredibly important, as is searching for new data sources and the ability to use those data sources. To use radar, to put into a model, for instance, the model has to be good enough that you almost have the right answer, and then the radar information can slightly correct it. So there is an iteration between the model and the observations, and the range that is available; but it is incredibly important that the observational system that is currently available is continuous, and that we look for ways of getting new information all the time, but there is no golden bullet of, "This is the more accurate data that could be available."

Q52 Chair: Just for clarity, those observational points can belong to anyone, and the key is that we have good co-operation.

Professor Hoskins: Absolutely. I think meteorology has been unique in the world. Even in wartime, countries have communicated their meteorological observations to one another. The world system that produces that observational capability is unique and it is so important to the whole operation. Clearly, if you are looking at the details of the UK, it becomes, "Do we provide enough observations for our locality, as well?" So if we are looking at the Olympics, do we have enough observations to really get the detailed forecasting for the Olympics? That has been something that we on the MOSAC committee have pushed the Met Office to look at, to which the Met Office has responded very positively.

Q53 Roger Williams: Would anyone like to add anything to that?

Professor Pyle: The global data sets that Brian has just been talking about are crucial and, of course, they are provided without the Met Office. They are provided internationally and continued provision, as has just been said, is absolutely essential. The Met Office then has a great record of interpreting those data sets and developing usage of those data sets.

It is also worth saying-this is something that the Science Review Group has been pushing the Hadley Centre on-that, as the models become more complex and as they contain more types of processes, then testing those models at a process level-not necessarily with global data sets but by high-quality local data sets so that you can understand the process-becomes more important. That is one of the things that we have been pushing very hard: to make sure that the Met Office tries to access appropriate data sets to test individual elements of its new models.

Q54 Roger Williams: Do you agree that representing uncertainties in modelling is a critical area for future research?

Professor Hoskins: I think that it is incredibly important for future research, and it is important in communicating the results of the models to the users and the public in general. One of the developments in the subject over the past decade or so is to try to get a real handle on the uncertainty that is present in forecasts. The running of the single model is not sufficient to give you that. Just giving one answer is not sufficient. One needs somehow to see the range of possible answers, given the chaotic nature of the atmosphere. Even if the model were perfect, the butterfly would flap its wings and it would be different.

The techniques are there now to explore the range of possible outcomes. MOSAC has been putting pressure on the Met Office to try, first, to produce that uncertainty. There have been great strides in that: the MOGREPS system at the Met Office has made great strides. The fine resolution model that is to be put in place for the UK will again have a measure of the uncertainty. So, we have techniques for exploring that. If you are going to look at those for the uncertainty in extreme weather, for instance, then you have to calibrate the system. Suppose you run 50 models-50 realisations-and 2 of them say there is going to be extreme rainfall, you have to be able to interpret that. That is only by using your same system over past data. You can then say, "Well, in the past, when it said this would happen 2 times out of 50, it has happened maybe 5 times or not at all." By looking at the past you get a way of calibrating what that model system is telling you.

MOSAC has very much encouraged the actual production and the communication-maybe you are coming on to that later. Communicating that uncertainty to people in a realistic manner is very important.

Q55 Roger Williams: Would the other two gentlemen like to comment on how the Met Office is addressing this particular issue?

Nick Baldwin: We have spent a lot of time discussing with the Met Office how to represent to the general public uncertainty. One development we recently funded was the "Invent" section of the Met Office website, where it puts up experimental methodologies to try to represent what is going on. We have been using that to show variability in temperature associated with different areas, to get people to start understanding that. As we go into longer-term forecasting it becomes more important to explain. The Met Office is currently working on ways to represent the uncertainty with its longer-term seasonal forecasts. With our support, it will soon publish material to help people understand.

One problem we find is that the populace at large is not well educated in probabilities and how you measure uncertainties. People completely misunderstand the risks they are taking or not taking in regular day-to-day events. For instance, it is more dangerous to drive to the airport than fly, yet most people would think it is the other way round. We have the same problem explaining weather forecasts. People do not understand when you say that probability is attached to it. That is part of the education that we are encouraging the Met Office to do.

Roger Williams: I think we will be coming on to that in a few questions.

Q56 Stephen Metcalfe: Good morning, gentlemen. On our visit yesterday we heard about the importance and the dependence on supercomputing capacity, and that the science is available now but not necessarily the computing power to realise that science. Do you consider that a step change in supercomputing capacity is required, as opposed to just an upgrade?

Professor Hoskins: In the past, meteorology and its application in both weather and climate had available to it the cutting-edge computers. That is no longer the case. It is a job to argue that weather and climate are now less important than they were in the past. There is no doubt that big advances could be made, pushing the science to the limit, if the supercomputer power took that step change. For instance, with global models we know there are real advances to be made if you can take the grid down to 10 km or so. We know that is needed for weather. Again, we come to whether we can be confident of seasonal or climate time scales, if we are not resolving the scale that we know is essential for weather. If you are going to run models, either on a very fine resolution for one or two days on a kilometre scale or even smaller or on the 10 or 20 kilometre scale for centuries, and not just do one-because of uncertainty you need to do many-then you need the step change in computer power. We would know how to exploit it. It is not just to do the runs, it is also to analyse the results of the runs. Sometimes it seems as if it is just going to produce results, but you need the whole system, which enables you to look at those results as well.

Professor Pyle: Again, I agree very strongly with what Brian is saying. If you look at the Met Office, it has an absolutely fantastic record, and is something of which I think the UK ought to be very proud. It seems to me that it has also delivered very important products for Government. You have to ask yourself whether you want to maintain the position whereby the Met Office covers that range of activities, in terms of the product that it delivers for the forecast and how it delivers information about climate change and the importance of that. Do you want the Met Office to continue to play that leading-edge role? I do, and I think that if you want that to happen, you have to invest in the appropriate computing, because, as Brian says, the challenge has become trying to understand things at a finer and finer resolution.

There is no point forecasting for 5000 sites if you have only one grid box covering the UK. You have to have models that run at very fine resolution to be able to forecast at that kind of number of sites. I am speaking as a scientist, and the science drive is to those higher resolutions. The interesting science, interesting questions and interesting answers come out of running the models at very high resolution, and for that, the Met Office has slipped down the league table substantially in terms of the computer power to which it has had access. I would like to see it right up at the top again.

Q57 Stephen Metcalfe: What level of investment would be required to get it right back up to the top again and over what period would that investment need to be?

Professor Hoskins: It is not just a single slug of money and that is it. It has to be a strategy, whereby you say, "We’re going to keep to this level." These days, you do not usually buy computers with a load of banknotes and that is it; it is a more continuous process. I have not looked at the details of the costs these days, but one must think in the order of £10 million to £20 million per year as an ongoing thing, if you are to stay at that level. As John said, it is the UK staying there and I believe it is the UK in terms of the Met Office and academia. We have a leading role in the world and we have been acknowledged as having that, and that is partly because we work together very well, but the whole community needs this sort of thing to stay at that cutting edge. Japan and Korea have this, and, yes, we can do our little bits of theory or whatever, but we will not be at the top table unless we have that sort of support.

Q58 Stephen Metcalfe: Do you agree with that, Professor Pyle?

Professor Pyle: Absolutely.

Q59 Stephen Metcalfe: And do you agree with the amount of investment required? Is that a number you would recognise?

Professor Pyle: Yes, I guess it is. John Beddington’s report talked about a factor of four, and what Brian says is consistent with that.

Q60 Stephen Metcalfe: Who should fund that? There’s the crunch.

Professor Hoskins: The UK.

Q61 Stephen Metcalfe: Okay. As simple as that.

Are there any economic benefits that you can identify that come back into society by investing in this, by improving the resolution and the accuracy of predictions? Is there a benefit to the wider economy that would justify the investment?

Professor Hoskins: Well, I think you can take this on various scales. I have mentioned extreme weather. If we are to make progress on the floods that have occurred in the past few years, the snowstorms and the high wind events-you name it-there is clear economic benefit on that time scale from the extra information that you can give to the user. They can then say, "Well, you’re telling me now that there is a 60% chance of this event somewhere in this location." You are refining gradually as we get closer to the event, and you should be able to take measures to combat it.

On the longer time scales, climate change is clearly such an important issue, whatever the economic situation, and there are decisions to be made on infrastructure in terms of adaptation, and mitigation in terms of actually knowing the latest. The amounts of money involved in putting in that infrastructure are just too large. Last year, for instance, the Department for Transport was asking, on the shorter time scale, "Are we going to get more winters like this?" That would have huge implications for the roads, the airports and the railways. If one could get more information to help in such decisions, the economic benefits have got to be huge.

Professor Pyle: I am just looking at the John Beddington report. He says that "external research on the value of the Met Office’s public weather service has shown that the service contributes at least £614m to the economy, based on a sample of the services".

Q62 Stephen Metcalfe: Assuming that we can’t make that economic argument convincingly enough to persuade the UK to top up the cash, are there any lower-cost alternatives that could provide the computing capacity needed to get up to the levels that you are talking about?

Professor Hoskins: I said UK, Met Office and academia, so one could imagine there being a shared facility in that way, which would slightly reduce the costs for one organisation. The facility would not be for academia in general, because the weather/climate problem is rather different from any of the other supercomputing problems. So I don’t think it would be a joint-purpose academia machine; it would be a shared dedicated weather/climate machine. That would be a way of reducing costs for one organisation.

There has always been discussion about a European solution, which seems to remain as far away as ever. One can imagine that it might happen at some time in the future, but there doesn’t seem to be any immediate prospect of a European solution. But on a decadal time scale there may be a reduced number of centres around the world that have such computational power available to weather and climate, which would then be shared with the region. I don’t think we can look on that as the solution now. Japan and Korea have their power now, and there is no sign that Europe will find such a solution within the next few years.

Professor Pyle: The Met Office already does come up with this. Brian mentioned Korea, and there is a substantial collaborative programme with Korea. There are benefits, in terms of a bit of computing power, that that brings. The Koreans do some things in collaboration with the Met Office. A long-term solution that relies on that rather piecemeal approach-a bit of Europe, a bit of Korea and a bit of Australia-is not sustainable, in my view.

Q63 Stephen Metcalfe: Are there technical barriers to doing that that you would identify? Or is it that you would rather have the capacity here?

Professor Pyle: I think there are logistical barriers. It takes effort to put a code on a different machine. For example, the unified model was put on a very high-power machine in Japan a few years ago by scientists from the University of Reading. It was a big effort to move the code from there to somewhere else. If you say, "Well, let’s try to do that in several places," you need several groups of people to be making that effort, which is not very cost-effective.

Professor Hoskins: There is running a model and there is actually looking at the results, which is often the more difficult thing if it is run remotely because you have this huge mass of data-this is very much the weather and climate problem-and you have to look at them and analyse what is going on. Do you bring those data back from the remote computer? That has difficulties, so where we are going to analyse those data is a real problem, too.

Q64 Graham Stringer: I shall ask a couple of questions in this area. Professor Hoskins, you are dealing with non-linear chaotic systems, aren’t you? Is there a limit, a theoretical limit at least, to what you can predict?

Professor Hoskins: There are different questions that one must try to answer on different time scales. If you are thinking about a projection through to the late 21st century, you would not try to predict the weather on 1 January 2080. So the idea that the system is chaotic and our ideas of what may be predictable and what is not, change on the different time scales. There is no particular limit that we know of in terms of slightly improving on different time scales. There are phenomena that occur on all sorts of time scales and the more we can understand those phenomena, the more we have the data to initialise those states of the climate system, the better one can hope for the prediction.

Let me give an example: on the multi-decadal time scale, there is an overturning circulation in the Atlantic that may well have predictable parts to it. We think there are and if we can have data, understand it better, get data for the ocean in the Atlantic and maybe the global ocean, one can hope to have an element of predictability on the decadal time scale, even though the weather will fluctuate around that.

So there is no time at which everything disappears into a chaotic mass. There are predictable items that we can pick out and try to use those ideas of uncertainty and get more predictive power for them. Whether that predictive power in all time scales will be useful, we don’t know until we have explored it.

Q65 Graham Stringer: May I put a case against what you are saying for the investment? I should be interested in hearing your or your colleagues’ comments. You want extra computing power. Computing power gets cheaper over time according to Moore’s law, so if you wait nine years, it is a lot cheaper, so you have to give the benefit of doing it now as opposed to over nine years.

The scientific community has come to a consensus, as far as climate modelling is concerned, that there is a 90% certainty that climate change is happening. Given those two factors and the fact that you are competing for money against, say, stem cell research, which could provide a cure for Parkinson’s disease, or the search for the Higgs boson, which could give us knowledge that was known before so we do not know where that is leading, where is your case? It is very weak when you talk about being at the top table or at the leading edge. Where is the hard benefit in that?

Professor Hoskins: On the mitigation side, I would agree that we know enough that we should be reducing our greenhouse gas emissions. In terms of the international discussions, it will be very useful if we could say, "These are the implications for China and these are the implications for India." That could be quite persuasive. At the moment those ideas are rather vague.

But I talked about adaptation where the decisions are being made now for the next Thames barrier for all sorts of infrastructural things with huge amounts of money associated with them. If we say, "Okay, we’ll go out of the race for 10 years and we’ll rely on what information we can get from others around the world", I don’t think they will be too keen to provide their information for our infrastructural decisions. On shorter time scales down to seasons, and if you are thinking about the benefit to humanity, then I would suggest that if we could refine the seasonal forecast and the implications for regions in Africa, it would have huge implications-perhaps more than some of the other things you have discussed.

On the daily time scales, again, with extremes, there are major human decisions that have to be made and benefits that could be attained. So it is not that we could just put this off and other things are more beneficial; I think this is as beneficial as any.

Professor Pyle: The science question has changed. It is no longer a question of whether greenhouse gases will warm the climate; it is about what will happen, for example, to areas of the United Kingdom in 2030. What will climate change look like at that kind of scale? That is why we need an investment. We cannot do that with our current modelling infrastructure. We can’t do it very well. We would like to do it much better.

Q66 Pamela Nash: Mr Baldwin, while customer satisfaction with the meteorological library and archive service is reported as being extremely high, the Committee has heard evidence that the bulk of the available data is only accessible to the public on payment of very high data charges. In the US, however, that data is more widely and freely available. Is that a complaint that you have heard often? Do you think it is possible for the Met Office to make that information more widely and freely available?

Nick Baldwin: It is not a question that I have heard often, but I have heard it from a number of sources. The important thing to recognise first is that we already provide a lot of data via the Met Office website that is free and downloadable. We also provide a lot of historical records through the archives. If people want to get large amounts of data from the archives and want it put into an electronic form, that costs money. Therefore, we have to have a process to recover those costs. We do not have budgets to fund all requests from people who are after large amounts of historical data. The intention is that people are charged the recovery cost for that.

We have been in long discussions with the Met Office about what further data information we can make freely available. We would of course actively want to encourage that, because, from our perspective, the more information that people have and understand, the more useful it is to them. The question is how to do that in a cost-effective way. In the way we work at the moment, there is no bottomless pit from which we can fund everything that people want.

Q67 Pamela Nash: But it is being looked at-whether it is possible to expand the available information.

Nick Baldwin: It certainly is, yes.

Q68 Pamela Nash: If you could achieve improved access to the information, how do you think that would affect the private sector in this country?

Nick Baldwin: We have had a number of discussions with private sector providers about this, and it is quite clear that there are a number of areas where private sector providers believe that if more data was available to them, they could produce different and potentially better weather services than we currently provide. If they can, it is great idea, because the intention is that we are providing a national resource and if we can do that, that would be a good idea. It goes back to the point that you have to find a way of providing the data cost-effectively. The current way that we are structured and our funding does not mean that we are a bottomless pit able to provide everything that everybody wants.

Q69 Pamela Nash: If more funding was available to you, that would be something-

Nick Baldwin: Yes. I am sure that the Committee is aware of the ongoing discussions about the consultation on the creation of a public data corporation, and the debate about what is a financially viable business model to release more data is a key part of that consultation.

Q70 David Morris: Sir Brian, do you see a potential conflict of interest between your roles as chair of MOSAC and as a member of the Met Office board? If so, can you clarify how that would work in practice?

Professor Hoskins: I was chair of MOSAC before I was actually a member of the Met Office board. My membership of that board has provided a very good conduit from the science that occurs in the Met Office and the independent view of that science through to the board. Personally, I feel no conflict whatsoever. I am a pretty independent sort of guy, and I assure you that I do not let it hold me back in any way.

As a member of the Met Office board, I suppose you might think that my chairing of MOSAC might be influenced by that, but it certainly is not. Our discussions are frank and that is encouraged. They are constructively critical, and they are attended by more than just those who are presenting to us. Many members of the Met Office come into those discussions. It all works, and I am pleased with the way it works. I produce a report at the end, which is agreed with the members of the committee, and I can then communicate that directly to the board with all the warts that it implies. I think that puts me in a very strong position to do that.

Q71 David Morris: How are members appointed to the various weather service groups-MOSAC, the Met Office Hadley Centre, the Science Review Group and the Public Weather Service Customer Group?

Professor Hoskins: I will answer for MOSAC first. I ask them to take part in it. I clearly consult the Met Office on who should be a member of it, but in the end it is my decision.

To tell you the membership, at the moment we have, I think, six academics, one of whom is a professor in a German Max Planck institute, and five are from the UK. The other seven members are the equivalents of chief scientists in met services elsewhere around the world. I sometimes have to explain to the rest of the board, who feel that this is actually parading our programme in front of our competitors, which is not normally done in industry, that this is the way that real progress is made in science, and in science-based organisations. I make sure that the input we are getting from these other chief scientists is real input, so that they are not just a sponge of what they are getting, and actually, the input that we get from academia and from these chief scientists at met services is very valuable.

Q72 David Morris: Do you think the different Met Office advisory groups maintain their independence from one another? For example, do they adhere to the code of practice for scientific advisory committees?

Professor Hoskins: I don’t think anyone would say that we do not act in a very independent manner. I think we do, but perhaps I should defer to John.

Professor Pyle: It is worth explaining that the Science Review Group that I chair is tasked by the Department for Environment, Food and Rural Affairs and the Department of Energy and Climate Change, which are the primary funders of the climate programme at the Met Office, to ask whether the Met Office is delivering science that is appropriate to the needs of DEFRA and DECC. We are actually a DEFRA/DECC committee, looking at the Met Office programme; we are not a Met Office committee, which MOSAC is.

We are subtly different-well, not very subtle, but we are different. We operate in very much the same kind of way that Brian does. We have a rotating membership and people typically serve for three years. They are all scientists, half of whom are from overseas. Scientists like to argue and disagree with one another and that is where the health of the committee comes in. People are there because they enjoy being better than somebody else, and that is the way the committee works. It has been effective.

I think there is an issue, from the Met Office perspective, about whether it is being over-reviewed. It is reviewed by MOSAC, and there is an overlap with the annual review that happens through the Science Review Group that I chair. On top of that, the Met Office has had a number of ad hoc reviews in recent years, such as the Sir John Lawton review, the Beddington review, and so on. It is entirely appropriate that it should be accountable for what it does, but I wonder whether it is being asked to do the same thing too many times.

Professor Hoskins: May I add something to that? The equivalent for the Hadley Centre on the weather side is the Core Customer Group, so the manager of the Public Weather Service is present at our discussions, as is the head of operations. My chair’s report goes through to the customer group, and I have presented it personally, or the Chief Scientist has presented it to them.

Q73 Chair: Before we move on, Sir Brian, you referred to the international players with whom you co-operate. I am intrigued by that, although it is not the core of our inquiry. In other sectors where UN treaties drive co-operation, the tendency is for things to move at the pace of the slowest. Why does that not happen in the met service?

Professor Hoskins: The drive to produce the forecast, which has been a collaborative operation often way before these UN bodies started to oversee it, has made everyone realise that this is the only way to progress the matter. The spirit of collaboration is probably unique in the weather/climate area. Everyone is competitive in the sense that they would like their model to be up there, but they are collaborative.

It very much reflects the mixture of competition and collaboration in academia-there is always an edge, a tension between the two. This is an operational area in which creative tension has managed to continue to the benefit of all. It has been a fantastic way of progressing. Research is shared and the research of today becomes the operations of tomorrow. The medium of journals and conferences is always a sharing operation of what we are doing. In weather and climate, it is a pretty short time from research taking place to its really making a difference at the coal face.

Q74 Gavin Barwell: I should like to ask questions about collaboration, which several of you touched on in answers to previous questions. To start with a general question, are there areas in which the Met Office could do more to collaborate either with the academic community or others in the UK, or on an international level?

Professor Hoskins: If you had asked me that 10 or 20 years ago, I would have come in with a load of comments, saying that the Met Office could be doing more. The collaborations and partnerships are so much better now, and they are progressing in the right way. Perhaps in the spirit of my previous answer, the individual scientists have always collaborated very well around the world. Met Office scientists have played a very important role in the international scene. Collaboration in the UK was quite rocky 20 or 30 years ago, but we have reached a point at which academia-Met service collaboration is probably just about the best in the world.

Internationally, again, in the past I would have said that the Met Office tended to be slightly, "Hands off! We’ve developed this model; we have the IP; we are not going to let you into this." There is a very different attitude now of realising that the partnership of others using your model is extremely important. A few years ago, MOSAC said that the strategy with the European Centre for Medium-Range Weather Forecasts should be rather better, with the scientists collaborating, but there are moves to get that in place.

So we are looking at a pretty good scene now. We are realising the values of collaboration and partnership. The UK is No. 1 helped by the collaboration in it, and internationally, that is essential to the Met Office place in the world scene. It is very productive to get the input of others into your framework as a way of having a bigger effort. This is realised now and there are very good programmes for doing it.

Professor Pyle: When I talked about MONSooN, I referred to the collaboration between the Met Office and the academic community. MONSooN is part of something called the Joint Weather and Climate Research Programme between the Met Office and the Natural Environment Research Council. As Brian said, that is a very new thing and we all welcome it. We are all aware that this is not moving quite as quickly as perhaps we would like. That is for perfectly good reasons on both sides as the Met Office and the academic community learn how to effect collaborations in the most effective way. Collaboration is great, and I am looking forward to seeing it get better and better.

Q75 Gavin Barwell: To pick up on a couple of specifics, Sir Brian, in the written evidence that you submitted to the Committee, you said that you thought the requirement on the Met Office to gain financial reward from its products might be damaging to collaboration. Will you elaborate?

Professor Hoskins: Certainly. Let me talk from the point of view of an academic, for instance, who might be collaborating with the Met Office to improve the ability to produce seasonal forecasts for India. Understanding the Indian monsoon would have tremendous impact on the ground, so that research is for the good of the subject and the good of humanity. But academics might not be quite as willing to share in the collaboration if they felt that the Met Office, at the next moment, would use that research to make money for itself. That is a tension that has to be handled. Recognising the difficulty is part of the solution. If, as the commercial pressures become stronger, this becomes more difficult, it becomes true with the data problem as well. If you make all your data freely available and do that before you exploit it yourself, you will not make as much money out of it as you would if you did it the other way round. There will always be a tension between the commercial operation and working for the good of all. We have to handle that tension.

Q76 Gavin Barwell: To probe a little on the example you gave, if in that situation the Met Office was looking to commercially exploit the model or forecast it had developed in relation to India, would the financial benefit from that be shared with the academic community that had been involved in developing that work?

Professor Hoskins: I am not sure I know of a particular example where that has happened. In general, we are talking about the academic working with the Met Office to do research that will be published. In general, the academic will get their reward in that way from that extra input to their career. That would be the natural way. There has to be a clear line between that and the commercial operation, which will exploit some of the benefits that have been obtained through that research and others.

Q77 Gavin Barwell: Professor Pyle, you said that the JWCRP was a good thing, but that it was not moving as quickly as you would like. Can you explore in a bit more detail for the Committee what you see as the barriers to making the progress that you want to see?

Professor Pyle: I think that there is just a bit of institutional inertia. As I have said, two sides have to come together to collaborate, although this is more than collaboration, essentially. In a sense, what happens is that if I was collaborating with another scientist, I would do something and the scientist would do something. We would pull those two things together and hopefully have something that is better than the two individual things.

What we need as part of this JWCRP is a more strategic consideration of the collaboration. As an example, I collaborate currently with the Met Office. We have been putting atmospheric chemistry schemes into the unified model-the climate model. The way that that has been done is that my group has done that work and we have then added it to the Met Office model. The best management way of doing that would instead have been to have somebody with sole responsibility for that, who could tell people in the Met Office or in my group at the university exactly what to do. At the moment, I tell my people what to do and people at the Met Office tell their people what to do and hopefully we build the thing that way.

I can see that there are big advantages to the United Kingdom, both for the academic community and the Met Office community, if we have a more strategic approach. That, in part, is how JWCRP is going. It is not that there is anything wrong; it is that people are perhaps gradually getting used to having to lose a little bit of their sovereignty. It is not just a question of collaborating; it is a question of being involved in the management structure rather more than they have been in the past. We scientists don’t like management.

Q78 Gavin Barwell: I have one final question to probe further on how JWCRP works. Is there a programme director? To take your example, is there a single person that is directing that programme or is it a loose negotiation?

Professor Pyle: I am trying to remember exactly what the structure looks like. There is a committee responsible for science and there is a committee responsible for facilities-I cannot remember what they are called-and each committee has two chairs, one from the Met Office and one from the academic community. There is the discussion currently about the development of a UK earth system model. There, the idea of this management structure that I have talked about will take place. There will be somebody who is responsible for that. This is starting to happen.

Q79 Stephen Mosley: In the evidence we received, the National Oceanography Centre stated that there is a common public perception that the Met Office does not provide reliable seasonal forecasts, largely due to sensationalist media reporting and shortcomings in how probability and risk are understood by non-experts. Do you think that is a fair assessment?

Professor Hoskins: It probably is, yes-we’re putting in probabilities. The seasonal forecast is a very interesting scene. Any retired van driver can look at the berries in his garden and produce a forecast, and a newspaper can take that to get a headline that will sell newspapers. It is then a difficult space for the Met Office to be in and giving likelihoods of different outcomes, but that is the way it has to be done. It has to be done in the manner of probabilities based on the best possible science.

MOSAC-and I in particular-have always encouraged the Met Office to produce more information about likelihood than has tended to happen in the past. Again, all the media pressure tends to be on the weather forecast; it has a shorter time, and the manner of communication becomes more important. There is so much information to communicate now, which I believe the general public as well as more sophisticated users would find useful.

Let us take the example of the hurricane that was likely to inundate New York. US television was showing 12 possible tracks provided by 12 different models for the hurricane. I do not believe that that sort of information is difficult for the public to assimilate, and we should not underestimate the public’s ability to take on odds and make their own decisions based on those.

For many years we have encouraged the provision of more likelihood information, although I don’t think the BBC was too keen on that. These days, with red buttons to press or websites, it is possible to provide sophisticated information for those who want to dig deeper, and the Met Office, with its great strength based on science, is well placed to provide that information. The tabloid headline will never be a good way of communicating the seasonal forecast. That leaves the Met Office vulnerable, but that is the world we have to live in.

Nick Baldwin: We were heavily involved in a discussion about withdrawing the previous seasonal forecasting approach. The consultation we undertook showed that people did not find it very useful in the way it was presented, and that they would rather have received a shorter-term forecast so that the three-month forecast was replaced with a 30-day rolling forecast. A lot of work has gone on since then with the Met Office, and over the next week or so it will introduce a new seasonal forecasting methodology for civil contingency communities, which includes a better explanation of the uncertainty facing us. I am sure that everybody is aware of the work that is going on at the moment in preparing for the winter, and the desire to warn people to be ready for potential extremes in the weather. It is important that people are organised and have a good understanding of that forecast. We have been funding that information and it will be released through the Cabinet Office.

Professor Pyle: I think the communication issues are quite difficult. If you say that there is a 60% probability that winter will be colder than average, it means there is a 40% probability that it won’t be. How do you decide after one winter whether your forecast was right or not in a probabilistic sense?

Professor Hoskins: I totally agree with that. You can never say that you are right or wrong at the end of one winter, given a probability forecast. What you can do as a responsible organisation, however, is say that in the past, when we have said that there is a 60% chance of a cold winter, that has happened on 55% of those occasions. Unfortunately, that rigour of evaluating how good a forecast has been is not always present in those who produce the tabloid headlines.

Q80 Stephen Mosley: The way you have described it, it is very understandable, but part of the problem we have is that most people get their weather information, whether long term or short term, from the TV, and of course with the pressure on TV schedules, the amount of time that weather forecasts are on for has shrunk. Is there an easy way of getting that information across in, say, a 30-second weather forecast?

Professor Hoskins: I don’t believe there is in a 30-second weather forecast, but I believe there is a way of opening it up to saying, "And if you want more information in the form that you are increasingly used to, you can find it here." That sort of likelihood information should be readily available to people. We now have the 5,000 sites with the information. There is no reason why we cannot give the probability information on all time scales and make that available to those who wish to delve. People are pretty good these days at pressing the red button or saying, "Yes, I want to find out more about this," and going to their computer. Then, a responsible organisation such as the Met Office can say, "We do evaluate these things, and this is the skill we have-the reliability of our forecasts." People will get this as an iterative thing and have confidence in it. We have to believe that the public are able to assimilate this sort of information. They do in their daily lives. I don’t see why they can’t for the weather.

Q81 Stephen Mosley: From time to time, I’ve seen American weather forecasts, which quite frequently say, "There’s a 20% chance of rain here and a 40% chance of being sunny." Occasionally you see that on the UK channels, but not often. Do you think that sort of presenting might be a way forward?

Professor Hoskins: One of the points, of course, is that that television channel is probably beaming down only to the particular city you’re in, so it’s able to handle that detailed information, and those channels seem to think they have more time to handle it and that it’s of more public interest, too, so you find that their weather forecasts are longer. As long as one can access to what that 20% probability of rain means and that is communicated, it is extremely useful to have it.

Chair: Thank you very much, gentlemen. This has been a very informative session.

Examination of Witnesses

Witnesses: Phil Evans, Government Services Director, Met Office, John Hirst, Chief Executive, Met Office, and Professor Julia Slingo OBE, Chief Scientist, Met Office, gave evidence.

Q82 Chair: We now move on to our second panel of witnesses. I would be grateful if you introduced yourselves for the record.

John Hirst: My name is John Hirst, and I’m Chief Executive of the Met Office.

Professor Slingo: I am Julia Slingo and I’m the Chief Scientist at the Met Office.

Phil Evans: I’m Phil Evans, the Director of Government Services at the Met Office.

Q83 Chair: Thank you for coming before us today. I also place on the record our thanks for your very informative tour of the facility in Exeter yesterday. We have heard already this morning about issues relating to money, and I’ll start at that messy end of the business. How should long-term funding for Met Office services be secured?

John Hirst: That is a very short question for a quite complex area. First, it is important to understand that we don’t receive voted money. We sell our services to Government Departments and public sector customers, just as we sell our services to the private sector. We have contracts from the Public Weather Service Customer Group. That is the only contract that we have that runs for more than a year. The climate programme has an unsigned contract as yet, and the defence programme has a contract that is for no fixed term. That creates some uncertainties and difficulties in resource allocation over the medium to long term, and as we’re in an area of activity that requires a long-term perspective for the scientific and operational development, that causes some tensions.

The benefit of having a contractual relationship is that it focuses the customer and us, as the supplier, on exactly what the customer requires and what the benefits are so that we are directing our activities to make sure they get what they want. We are bringing new ideas to their service. I can’t think of anybody in the world who wouldn’t want a longer-term perspective on the funding, so that we had a better planning horizon to operate in.

Q84 Chair: Are there ways in which non-governmental revenue could be increased? Some of the very sophisticated tools you have are of enormous value to the insurance industry and a wide range of sectors of the UK economy.

John Hirst: We have major customers in a whole range of different areas-utilities, for example; water utilities and power utilities. Insurance companies work with us, particularly the reinsurance sector where they are pricing their risk over the long term against historic climate records and frequencies of extreme weather events, and the destruction that they cause. We do build, and have built over the past few years, the revenues we get from those customers by deploying our expertise in their service.

It is a pretty tough economic time at the moment, so it is slightly harder to sell those things than it might otherwise be. But we continue to make steady progress. Moreover, because of the collaborations we do around the world and our sharing, we leverage other people’s spend in the area that we have to reduce the burden of cost. We are also now drawing the attention of national met services around the world, which see that we can deliver benefits in terms of products and want to take advantage of the experience. We have got to help them and take our products under licence.

My current estimate from that funding is that we reduce the burden on the public purse by about £20 million to £25 million a year. Our ambition is to build on that so that we give the UK’s Public Weather Service increasing value for money as we go.

Q85 Chair: In terms specifically of the Hadley Centre, has the DECC-DEFRA relationship provided a sensible, stable funding mechanism?

John Hirst: We have very good relationships with DECC and DEFRA. They have been very good in supporting our Hadley Centre programmes over many years. We do, of course, see their struggle with funding, challenges and the money that they have available to spend, so they are constantly challenging to see whether they can do things in a different way. I reiterate that it would certainly be better if we had a slightly longer term perspective over a couple or three years in the funding that was clearly committed and supported. It is not quite there at the moment.

Q86 Chair: In terms of things that could change, your plea would be for longer term time horizons?

John Hirst: That would be very welcome.

Q87 David Morris: Do you accept that the Met Office needs to make its models more user-friendly? If so, what will you do to address this issue?

John Hirst: We do, for a number of reasons. The more we can get people to collaborate with us, the more our science and our services will advance, because we bring in the scientific capabilities of other people who use our model to develop and do their experimentation. That could be in the academic community around the UK or internationally, or even between us and our national met service partners that use our models in South Korea, Australia, New Zealand, South Africa, Norway and others, who bring to us their expertise and some support in funding.

We have permanent stress that I think you have already heard about. We are having to run our model so that we deliver many thousands of forecasts for various users every day, and do the scientific development and experimentation we need to do to roll the science forward. It is probably slightly better in the climate area than it is in the weather area where there have already been some commitments from the academic community to build that relationship and put in the resources to make sure that the things are translatable and easier to use. We are now working with the weather academic community to do the same thing and make it easier to use, recognising the constraint that we have to deliver the forecast every day.

Q88 David Morris: What more should the Met Office be doing to ensure that it has access to, and makes more use of, accurate source data to input into its models?

John Hirst: One of the benefits you have heard about is the fact that in the meteorological community, which is a subject of fascination to me coming from the private sector in the past, there is massive collaboration. Of the data we use to produce our forecasts, we ourselves produce something less than 4%. The rest comes from the world community of met services and academia.

We stretch continuously to find additional sources. One I use, for example, is that mobile phone companies try to remove the distortion from their signals. Quite a lot of that distortion is caused by atmospheric conditions, so we work with Ofcom to see whether we can get those distortions, which we will interpret to understand where moisture and atmospheric disturbance are. We do that all the time. We also work collaboratively to deploy buoys in areas of the ocean where we do not have good data, and there is a rolling programme of development. You will never satisfy any met service that it has enough, but we do work hard on building that network.

Q89 David Morris: Professor John Pyle described the success of the MONSooN project in facilitating joint model development. Do you intend to expand on the MONSooN project? How else can you encourage the joint model to develop?

John Hirst: Yes, we do. In our last procurement and in the delivery of new capacity in the second phase of that, we are building the size of the allocated space for the MONSooN project, and we are already in discussions with NERC about the next phase to develop that.

Q90 David Morris: Do you agree that representing uncertainties in modelling is a critical area of future research? How is that being addressed within the Met Office?

John Hirst: We have a number of serious challenges in the organisation. There are big scientific challenges. Representing and understanding the uncertainties in the science is an issue in itself. As you have already discussed, representing the uncertainties in a communications sense, so that people understand what it really means, is also a big area. I think it is an area we have addressed over the past few years much more intensively than in the past, because we have learned from focus groups, academic studies and work with experts at Cambridge University, how difficult it is for people to understand uncertainty and integrate it.

I had conversations with international colleagues. A couple of weeks ago I was talking to the guy who heads the Canadian met service, where they have used concepts of uncertainties and probabilities a lot. In their survey work, although people are used to it, they do not always understand what is exactly communicated by those uncertainties and probabilities. That is an area that will keep us busy for some time.

Q91 Graham Stringer: You were in the room listening to a discussion about the case for increased supercomputing capacity making a big step change. How important is that?

John Hirst: It is vital. We have an aim in the organisation to provide to the UK the best weather and climate service in the world. The aim is not because it is a good thing in itself, but because the earlier you can give warnings of extreme weather events, the more you can give reliable indications of emerging weather and climate patterns, and the more Governments, communities, businesses can use that intelligence to plan their activities, save lives and cost, and build their businesses. Being the best is about pushing those barriers down, ensuring you can get more reliable information out earlier.

If you look at the insurance industry, which has done a lot of work on this, two-thirds of the world’s insured losses are as a result of natural hazard events. Wind kills more people, water causes more cost. Curves on graphs show that the earlier you can warn makes a significant improvement to the losses that they engage in.

Professor Slingo: I was also sitting through that discussion, when the question was asked about possibly waiting nine years or whatever until Moore’s law takes you to the power that you need. It is clear-as the panellists said-that the science is ready and waiting. We know we have science; we understand; we have models ready and waiting to roll out, and we are testing them.

Investment now will give a very rapid return on that in terms of the economic value to the UK and its interests nationally and internationally. That is perhaps distinct from other investments you could look at in research. This is not entirely for research; it is science ready and waiting. It just reinforces the comment that Sir Brian Hoskins made that weather forecasting and climate predictions sit very close to the bleeding edge of research. As an organisation and through our partnerships, we are now very skilled at taking the latest science through into our operational delivery, so you will get a very rapid return on the investment.

Q92 Graham Stringer: You talk about the insurance companies’ curves that show the savings that can be made. As was discussed before, that is very difficult to communicate through the Daily Mail or the Daily Mirror. Is it possible to give a real example of where you have focused on an extreme weather event and-saving lives is particularly difficult to demonstrate-you have stopped damage to property and, in all likelihood, saved lives?

John Hirst: There are a couple. When the Cumbrian floods occurred a couple of years ago, we were working jointly with the Environment Agency and our-newly established at that stage-flood forecasting centre. We were able to give 24 hours’ better notice than we had ever been able to do for a level of rainfall that was beyond any historic record of rainfall in this country. So the level of the confidence that we had from the quality of the science that we had gave the responding community there an opportunity to get things in place and manage the event much more effectively.

There is another example that Phil talks about, which is Tewkesbury, where the fire chief, having had three or four days’ indication of likely floods, was able to bring boats and inflatables to that area from Cumbria and the Lake District. He was therefore able to get people out of their houses and to safety in a way that they would not have done had they not had that advance notice to shift the equipment to the right place.

Q93 Graham Stringer: Is it possible to get the extra computing power by collaboration, using something akin to the Airbus model, rather than just putting it all into one supercomputer in Exeter or anywhere else?

Professor Slingo: That has been looked at over many years by the international community in weather and climate science, and the problem is that the nature of the problem that we are trying to solve computationally does not lend itself to distributed computing. It has been tried and we cannot get either the efficiency, in particular, or the timeliness.

In weather forecasting, you have a very small time window before the forecast is out of date and beyond its shelf life, so you have to work within that very tight time window. That means that you have to be able to both run the model very efficiently and also gather and process huge amounts of data. That requires a very specific type of machine architecture.

The nature of the codes that we use means that we require large memories per processor, which is not true for many other sciences. We also require very fast interconnects between processors, because we exchange a large amount of data. Finally, we need very large bandwidth to get the data out of the machine, because we are producing huge amounts of data as the model runs, and on to a huge data archive.

Those sorts of systems are not typical of the sorts of architectures that are around and available on group computing or other set-ups like that. That is the conclusion that we have come to, but it is one that all the major weather and climate modelling centres around the world have also come to. We look at it every year and assess these things.

Q94 Graham Stringer: That is very interesting. I have one final question. I can see how you can verify the accuracy of your forecasts over a day; you just see whether it has rained or not. Even seasonally, you can check the probability over a few years. In terms of your climate predictions, which you have only been doing for 20 years or so, Professor Hardaker told us last week that you had predicted the flattening of the temperature curve over the past 10 years or so. We have heard evidence at different inquiries before the Committee that doubts that. Can you give us documentation that shows that you got the shape of that curve for temperature change over the last 10 years right?

Professor Slingo: What you have to understand is that we have done a lot of work to understand how that flattening of the curve has arisen. It is part of the natural variability of the complete system. Professor Hoskins talked about the slow-

Q95 Graham Stringer: I understand that. What I am interested in is whether you have got it right in advance.

Professor Slingo: When we initialise, this would have to be only through our decadal prediction system because that flattening was a result of the current state of the full coupled climate system at the end of the 20th century, in the 1990s. If you initialise the model from that state and run it forward then, yes, we got some of that flattening.

If you are running a free-running climate change prediction model, it would not exactly replicate that particular 10-year period, but what it would have, within the richness of its projections, is decadal periods where the rate of change in temperature would slow down or it might be a bit faster. That is all within the natural variability of the climate system. So the attribution of that flattening is that it is part of the natural system.

There may also be a contribution from increased aerosol loading. Again, if you were predicting that ahead of time we would not know that. That is an anthropogenic forcing, but looking back we can reproduce that and say, "Yes, the aerosol loading would have that effect on the temperature curve." As you know, we are now picking up again. The final area here is that we are also going through quite a deep solar minimum. So if you combine aerosols, deep solar minimum and the slowly varying ocean circulation, it is very easy to justify that short-term, decadal-time-scale, decline in the rate of rise of temperature.

John Hirst: It may be helpful to express something that I learned when I joined the Met Office about how the science has developed. The easiest example is a weather example. When we introduce new science for models, we go back to an event in history that we might not have got precisely right, reinitialise the model and run it through that event to see how much better we get it. That proves the quality of the science. That builds our confidence as we go forward that if that or similar events occur then we will get it right, or righter than we did in the past.

The same applies to the way we develop climate modelling. We will go back and initialise the model 20, 40, 50, 80 years ago and run it through and make sure that we are replicating the observed trends in the Earth’s atmosphere. So it is not that we do not have any checking. Clearly, there is a difference between making a forecast for tomorrow when you experience tomorrow very quickly, but we are going back and modelling how the climate has evolved in history to make sure that our models replicate what actually happened. That, again, gives us the confidence that we have the chemistry and physics right which will help us predict the future.

Q96 Stephen Metcalfe: Can I take you back to the supercomputing capacity? In the first session we heard an estimate of around £10 million to £20 million per year to provide the capacity. Is that a number that you recognise? Do you think it is more or less?

John Hirst: We always need to say what problems we are solving. There are short-term problems that we can get better at. There are longer term problems that we can better at. We need to balance the availability of science, the availability of infrastructure, including observations to verify and check how we are going, and the supercomputing capacity. It is not just about saying, "Here is a cheque for a supercomputer." It was quite rightly observed that we have science now waiting for application and a lot of scientific experimentation waiting for computing capacity, despite the fact that we run this pretty efficiently.

Having thought about this a bit, I think that if you take specific examples in the short term, getting weather events more precisely, you could allocate capacity and other bits of about £7 million a year to that in the next three to four years. Then if you take the monthly and seasonal forecasting that we do-I am trying to not call it seasonal; I am trying to call it the three-month rolling forecast-you would allocate some £7 million a year to the problems that we have ready. It comes to about £14 million over three to four years. During that period, you would constantly review what the next challenges are and what the next science availability is, so that you would refresh that as you go forward.

Q97 Stephen Metcalfe: That is quite a lot of money. There were suggestions about where that £14 million might come from in the earlier session. I am sure you have your own ideas and I do not want to get caught on that. Have you done a cost-benefit analysis on that additional spending and what the benefit is to the wider economy?

John Hirst: Yes we have. When we did our last procurement we had three A, B, C and D cases. The case we went to was not pushing the envelope as far as would have been ideal, because it was based on affordability-not that we were not grateful for it. At that stage, we had a return ratio of 10:1. That meant that every £1 invested got £10 return. That is not in the next week, but it builds progressively over months and years. We could have justified that same ratio with a computer three times bigger. That ratio still exists. Again, it joins with the earlier question of why we would not wait 9 years, but we are losing £2 billion a year in losses to the community, because of disruption due to non-preparedness for extreme events. You can save that money as you go and get the return.

Q98 Stephen Metcalfe: You said that it is 10:1, which happens to be a very round ratio. Can you justify that in any way? Can you give practical examples of where the additional computing capacity could lead to a better prediction, which could save x pounds?

John Hirst: Most of that 10:1, at the time, was built around preparedness for floods. It does not include the kind of benefits that we have explored with the Department for Transport for better preparedness for extremes in winter. There are many quotes on how much the disruption in winter last year cost. You can build that in. I am confident that we would not fall short of that 10:1. It is a rounded figure, because it would be spurious to say that it is 9.687:1. The estimates that you would have to make over several years are just not that accurate.

Stephen Metcalfe: But there is practical application behind that number, so it is justifiable.

Q99 Pamela Nash: I want to turn to the science strategy of the Met Office now. To what extent did you consult with external organisations and the meteorological community when you were forming the science strategy?

Professor Slingo: As Chief Scientist, it is my job to define the science strategy for the Met Office to ensure that it is fit for purpose in 10 years’ time and to guide the science to deliver our operational requirements. I came, as you know, from the NERC community, where I led its climate modelling programme. There was consultation through the Met Office Scientific Advisory Committee, so it was involved in the development of the strategy. Colleagues were, of course, involved. As Sir Brian said, we have six academics sitting on the Science Advisory Committee who were engaged in those discussions. At the end of the day, our science strategy has to be one that I am confident is right for us as a business and as a public sector organisation that has to deliver our public task and one that I believe is achievable within our own resources-that does not mean that I do not welcome the consultation with the academic community; as we develop the implementation plan, that is very much in our minds. At the end of the day, it has to be my judgment that this is the science that we need to do to provide the highest-quality, most useful forecasts going forward, whether it is a few hours ahead, or out to a century ahead, in response to what I now know, or understand, as the cross-government needs.

It is clear-I see this very clearly having come from the academic community-that the Met Office research programme has to be very directed towards our public task, whereas in academia you follow the Haldane principle, which says that you do the best research for research’s sake. The development of the strategy would be very different for me as Chief Scientist in the Met Office to the strategy that I was developing as head of the climate science programme in NERC at Reading University. I sit on NERC committees and I know what the big science challenges are. It is not surprising that the academic community would recognise the same challenges.

Q100 Pamela Nash: Did you feel that the level of consultation that took place was adequate?

Professor Slingo: It was absolutely appropriate for the job.

John Hirst: May I offer a thought as well? There are masses of exchanges of information that go on all the time in the Met Office. It is a structure and almost an osmotic process. There is MOSAC and other advisory bodies. There are the bilateral arrangements that we have. There are hundreds of collaborations with scientists in academia in other institutions around the world that go on all the time. It is impossible for those not to help formulate the strategy. When the strategy has gone out, it is commented on all the time and will evolve progressively as people make comments about potential gaps or things that are emerging in science. We operationalise those contributions. There are both structured and natural flows of information, so consultation is almost unavoidable.

Q101 Pamela Nash: I appreciate that, but my question has arisen out of the memorandum that we received that said that the research councils would have appreciated a greater opportunity to have been consulted on when the science strategy was being developed. Why were the research councils not given a greater opportunity to contribute during the consultation?

Professor Slingo: I was rather surprised by that comment because the director of the National Centre for Atmospheric Science, who is the most obvious NERC representative, is on our Scientific Advisory Committee. The strategy was discussed with them. One of the things that is sometimes misunderstood is that a strategy is not an implementation plan. The NERC community tends to see strategies as implementation plans. NERC is mentioned throughout the strategy. Now that we are going into the implementation plan, as Professor Pyle talked about, the role of the Joint Weather and Climate Research Programme with NERC will be critical to the implementation of that strategy. It is at the implementation level that you really can work together. As Chief Scientist, it is my job to define the scientific direction of the Met Office so that it is fit for purpose in operational weather forecasting and climate prediction for policy decision making by Government.

John Hirst: We have already agreed that we will go back and ask what particular areas they might have been consulted on and what concerns they have because we want to draw in that information in a deep and thorough way.

Q102 Pamela Nash: On the implementation plan, I understand that the science strategy was published a year ago today, yet the implementation plan was only published last week. You said that those are two different things and it might have been misinterpreted. Why has there been that year of delay in producing an implementation plan?

Professor Slingo: Clearly, it takes time to develop an implementation plan. It does not mean that we have been doing nothing. We have actually implemented an awful lot of what is in that science strategy in terms of the science partnerships and the integration and restructuring of the science programme. Many of the recommendations have already been followed through, but we have to see the science implementation plan as part of the corporate plan. We have gone through a major refresh of our corporate plan in the last year. The science implementation plan has to be then put in the context of our business implementation plan and our implementation plans for Government services. There is a timetable to that and a year is what it has ended up being. We have implemented an awful lot of things during that time.

Q103 Pamela Nash: Would you not agree that a year is a long time? Has a level of uncertainty been fostered between Met Office scientists and external collaborators who are having to wait such a lengthy period of time?

John Hirst: No, it has not. I do not make my judgment about that on the basis of details within a science strategy, but, on having published that, the other met service in the world that is ranked in the top two or three, and which we respect very much, is the Japanese met service. It is on the MOSAC advisory committee and said, "This takes the strategic science strategies of all met services to a different level, and we will be following it." That is a kind of peer acknowledgment.

One of the things that we strive to do in being the best is attract first-class collaboration. We want the best scientists to want to collaborate with us, and there has been no let-up in that collaboration. People see what we are doing on a day-to-day basis, and want to come and work with us and share their expertise. The number of scientific papers that are published jointly with other institutions and leading scientists around the world are in the hundreds. I do not see any let-up or delay in that.

Q104 Pamela Nash: It has remained the same over the past year.

John Hirst: It has improved; it is increasing.

Q105 Stephen Mosley: In the previous session we heard that some people are saying that only a tiny fraction of the available historic weather information is accessible to the public without incurring large data charges. Do you recognise that as a widespread complaint?

John Hirst: I hear it. Some people complain about it quite a lot and persistently, and others don’t complain about it at all. Every time we get a specific request or issue, we try and address it and ask, "What precisely do you want?" For example, I received a letter that was addressed to the Ministry of Defence about a year ago, grumbling about the lack of data. I went to the person and asked, "What is it you’d like?" I haven’t received any response. Sometimes complaints are a general perception about history, and sometimes they are genuine things.

I am not diminishing the fact that there are data that people would like access to. Some of that stuff is still held in paper format in the archives, and it would cost money to put it all into digital form so that it can be readily accessed. We are constantly managing a tension between how much money should be spent to develop that and whether it would satisfy specific requirements more directly and more cost-effectively, or whether we need to make all those things accessible because there is a general demand. At the moment there is a general demand in some areas and, in consultation with the Royal Meteorological Society and others, we are looking at how much data we need to put into that kind of category to make the information more readably available. We are testing demand in other areas to see whether it persists and would provide good value for money, or whether it should be done in a different way. It is pretty open, and we make what data we can available.

Q106 Stephen Mosley: You were talking about the different types of data-I guess you have raw data, previous forecasts, actual weather forecasts and the effective results of what actually happened. What are people actually interested in when they contact you?

John Hirst: It varies a lot from people who want to know what the weather was like on the day they were born-it starts with very small things-to people who are conducting research experiments in areas of preparedness for wind events or particular areas of science. There are people who are writing books about the development of the coal industry in south Wales and want to look at weather events. It ranges massively over a whole spectrum. The most consistent demand is from private sector weather providers, and we make data available as we can. It is worth noting that through our international relationships I help the UK private sector get access to data in other countries where there is not as much openness and flexibility as in this country.

Q107 Stephen Mosley: You are talking about private weather forecasters and the private sector. Do you think that if the information could be made more easily available and more cost-effective, it would help grow a more vibrant private sector?

John Hirst: To a certain extent, yes. Some private sector providers, I have to say, are concerned about the amount of data that we are providing free of charge. The 5,000 sites that are being made available in the Public Weather Service will cause some distress to some private sector providers, because that is the kind of thing they have been doing and charging customers for. There is a wide range of different kinds of providers. Clearly, some are much more sophisticated and some less so. There is not one answer that applies universally to all the private sector providers, but clearly, the data available are a key issue. We have a commitment to move up the volume and relevance of the data that we make available.

When you listed some of the things that you talked about, very few people want yesterday’s forecast, to tell you the truth.

Q108 Stephen Mosley: I think Graham Stringer wanted 15 years ago.

John Hirst: The information is time-expired. We try and make available, as part of our mission, as much of the data as we possibly can behind the forecasts and behind the observations that we have, so that they can be used, checked, developed, enhanced and absorbed into other people’s activity.

Q109 Stephen Mosley: Last week, on Thursday 27 October, the Government consultation on the public data corporation finished. Are you in favour of the Met Office forming part of the PDC, and what impact do you think that it will have?

John Hirst: You are asking me to comment on an area of public policy, which is not-I will respond to the Government’s decisions on this.

The consultation, as I see it, is part of shaping what the PDC might be and how it might operate. We have been working with the teams in the Cabinet Office and in BIS to help them understand what we do already and what contribution we make. We can do some things collectively, but whether they are worth investment is for other people’s judgment. We have a plan to develop partnerships with other environmental science organisations, which we have called the environment science to service partnership, and which works with colleagues in the NERC institutes to help bring them some of their science-fantastic, world-leading science-out into greater use, by developing applications and uses for that science across the boundaries between us. If we can make use of the PDC as a vehicle to do that kind of thing, it would be very beneficial.

Q110 Stephen Mosley: Surely you responded to the PDC consultation, did you not?

John Hirst: It is a public consultation, and I am in the process of writing a letter recommending some things that should be taken into account. However, as it is a public consultation and given that I am not a member of the public in this role, I have not responded to it. There is a different route.

Q111 Gavin Barwell: I want to explore collaboration, which I touched on with members of the previous panel, and I think you were all in the audience at that time.

First, a general question for all three of you: do you think that the widespread use of common modelling systems encourages collaboration, and if so, are there specific systems that you think the Met Office should be sharing more widely?

John Hirst: I will ask Julia to comment on some of these things in a second. One thing to make clear is that we recognise that it is impossible for us to do all the science that we need in order to deliver the best service possible ourselves. We cannot do it, so we reach out in a very structured way now to draw in the best science collaborations we possibly can, and to operationalise that for the best benefit of the UK and our customers. An underlying philosophy is that we collaborate. We have created a post; one of our very senior scientists is now head of partnerships. He works on developing the partnerships on the unified model and on developing collaborations internationally between us and scientific institutions. We are acting on that need to collaborate.

Julia, it is probably worth quoting some of the numbers from the collaborations that we do, and how much we rely on it.

Professor Slingo: As John said, as part of the science strategy, we highlighted the role of science partnerships in a much more formal way than hitherto, where it would have just been going on at a grass-roots level. We have a head of science partnerships who was in charge of our climate programmes and who has now moved into this. He is a leading scientist.

We have started a process of quantifying the benefits that we accrue from such partnerships. Our statistics for last year show that we had 165 projects with an estimated value of £15 million, within year, so that is a third of our science budget within the Office.

Q112 Gavin Barwell: So that was 165 projects, with an estimated value of £15 million?

John Hirst: That is leveraging to value assigned to other people’s efforts.

Professor Slingo: More than a third of that came through the Joint Weather and Climate Research Programme that we established with NERC, which John Pyle talked about. That is now developing very strongly. I, with the director of operations at NERC, chair the strategic programme board. Between us we have implemented what I think is a very good structure to allow the programme to take off.

About another third comes from our international partnerships, which is the leverage we get from the countries that use the unified model within their operational weather and climate predictions, where, again, we now have much more formal arrangements in place. It is more about partnership and ownership than licensing. To begin with we licensed the model, but now they have implemented the model in their forecasting systems and are beginning to do research and development on that model, we are coming to a much tighter partnership arrangement. We recognise that we are all in this together.

The benefits this year are £15 million. Within the past year we have also launched the academic partnership scheme with three major universities, which draws together about 1,000 scientists working in HEIs, rather than just in the research councils. That is a different leverage, a different way of stimulating science within the academic community, as well as translating that science into improved services. I imagine that, as the JWCRP gathers momentum, that £15 million will grow and grow over the next few years. We need to quantify that year by year, and we will.

Q113 Gavin Barwell: In their memorandum to the Committee, the Government strongly support the proposal for stronger partnerships in the science strategy, but they said that the proposed partnerships should include Government representation. Do you think that is appropriate? If so, how do you intend to facilitate that?

Professor Slingo: We need to be careful that, particularly with our academic partners, we don’t conflict with the Haldane principle, which we need to recognise.

The science partnership programmes are presented at MOSAC, and we make it clear to the Hadley Centre Science Review Group that they are the collaborations that benefit the climate programme. I think that is appropriate at this stage. As the JWCRP gathers momentum, we could include somebody from the appropriate Government Department, probably BIS in this case, on the strategic programme board, alongside myself and Phil Newton from NERC.

John Hirst: Having also read that memorandum, I don’t understand the thoughts behind it. We need to understand precisely what people want to do.

Q114 Gavin Barwell: It came as a surprise to you?

John Hirst: Yes. We need to understand a bit more about what people would like to achieve, and we will work hard to accommodate that.

Q115 Gavin Barwell: I want to give you the opportunity to answer the question I asked Sir Brian. In his written evidence, he said that the requirement on you to gain financial reward from your products could damage collaboration. You heard me put that point to him in the earlier evidence session. What would you like to say about that?

John Hirst: It is a tension that I think is quite healthy. At the extreme there is a risk. In the past there has been some confusion about what is the best way to develop this, but we have a clear view that the more we collaborate, and the more we develop our science, the more we will be able to bring in supportive revenues to our operations. It is not in the data; it is in the interpretation and contextualisation of that information, interpreting it for different users and understanding their requirements, where we really attract the right kind of revenues.

The benefit of having real customers who require real delivery and pay money for it is evident throughout the organisation. That keeps us focused on delivering and making sure that we are efficient and that we work in a businesslike manner. Although there is a risk that we must be aware of, it does not drive any inappropriate behaviours in the organisation-indeed, to a certain extent, it drives better behaviours than if we did not have customers.

Q116 Roger Williams: There is a perception that the Met Office’s seasonal forecasts are unreliable. Is that because the forecasts are unreliable, because of how you communicate them to the public, or because the public do not understand probability and risk in such matters?

John Hirst: Long-range forecasting is an area of developing science and significant scientific challenge. But it also has massive potential benefit, because it addresses areas that businesses and responding communities are thinking about in preparing to make their activities more efficient.

In that whole area, the Met Office is one of a small number of leading institutions in the world. We are breaking down scientific barriers as we go. We prepare those forecasts principally for professional use, because they are quite complex at the moment.

As Brian Hoskins said, these are not the kind of forecasts that say, "You can take your deck chair to Weston-super-Mare beach on 15th of whenever." They are about general patterns, likelihoods and probabilities of general patterns emerging. They are useful in the UK for professional users, utilities companies and the civil contingencies secretariat. They are useful internationally for Governments - planting and developing crop plans in Kenya, for example, where we have examples of people benefiting from our input as they help feed their population. So there are real benefits.

Long-range forecasting is a complex area of science to communicate. In refreshing our service to Government, we are saying, "Here are the limits and the extents of what you can decide." We are also working hard to find out how we can communicate that information to make it more simple to understand.

I have said a number of times that I have been responsible in my past, when I was a group treasurer in a major international corporation, for buying and selling £14 billion or £15 billion worth of foreign currency a year. Even at this stage, given that it is based on physics and chemistry, some of the seasonal-long-range output-forecasting is more reliable than some of the economic forecasts I was using to make purchases.

Q117 Chair: Tell that to the Treasury.

John Hirst: They are not Treasury-given forecasts; they are commercially available. We have a lot of work to do in the area.

Q118 Roger Williams: Obviously, the work is to explain it to the public, but it is a bit like a chicken and egg situation isn’t it? Is it your job to educate the public to have a better understanding of probability?

John Hirst: We cannot say, "Here is our information-it’s up to you now." That would be derogation of duty. But we must understand what people do with our information and make it available in as digestible a form as we can to help them make the right decisions and understand the implications of what we are saying.

Q119 Roger Williams: I think it was Mr Baldwin who said that his group had suggested that that type of forecast should be made available not to the public but to particular groups.

John Hirst: We have clear rules that state that, if we are going to make information available to the Government, it must be made available publicly-it has to be put on websites. We have an obligation to ensure that the limits and uses of information are clear. We have tables that state what kinds of things to use a one-day, three-to-five-day, monthly, seasonal or decadal forecast for, so that people understand the context that we are operating in. But it will take a little while for people to get used to that kind of uncertainty and probability.

Q120 Roger Williams: You mention making greater use of your website. How will you give that information?

John Hirst: It will be available on our website. First, we have improved the context and amount of data in that forecast. It will be made available with explanations of the kind of things that are on there, what kind of use you could put them to and what the uncertainties do, and we will give a commentary around the data. That is what we are doing. That will now evolve over months as we develop the service.

Q121 Roger Williams: What should the Met Office be doing to communicate to the public the underpinning science of the forecasting you do? Can I just say that putting up maps and having symbols for the sun, rain and such things is rather dumbing down the science, not giving people better understanding?

John Hirst: May I challenge that? That is not the way of describing the science, to be fair. It is a way of communicating in a very short period what the general weather patterns will be. That is not the way we communicate the science. We have other plans to communicate the science. It is very helpful for people to understand a little more of this, and we co-operate with broadcasters-we have had two film crews in the office this week studying the science and how it can be communicated to build an understanding of what underlies it. Julia, would you comment on the science?

Professor Slingo: I came in as Chief Scientist and I have really pushed the need to be more open and transparent about the science we do and to get much more information on to our web pages. For example, we now have a research link on the front page that takes you into all our science areas and tells you what we do and who does what, so if you want to find out who a scientist working in a particular area is, you can.

We have just launched "Research news", where there are hot topics of the moment-one of which went up this week on declining Arctic sea ice-presented in language suitable for an interested member of the public. We have started looking at all sorts of other ways of communicating science, such as YouTube videos on topical issues, which get a lot of hits actually-an increasing number. We use Twitter and all sorts of media. We have just started a whole series of educational posters with accompanying videos, to explain the very basics of how the climate system works, for example, and therefore why one thing-carbon dioxide-changes everything.

The whole business of education is really important and it is really very difficult, but I think we have a clear aim within the science programme to put far more information out about where you are publishing and who is doing what. I hope that the "Research news" page will take off as somewhere to which interested members of the public will go and say, "I wonder what the Met Office has been doing this month. What’s the latest stuff on the role of the sun?" The sun is another story on there at the moment.

Q122 Graham Stringer: Roger Harrabin at the BBC is setting you a test, comparing your forecast, both short and medium term, with private providers. How do you feel about that? Will you co-operate?

John Hirst: I would not say he was setting us a task. I think we are setting a task together to explore the area. We have had lots of conversations with Roger and the colleagues he has gathered around him helping to shape that study. To be fair, it is an area of verification and comparison that has been very challenging and has eluded a conclusive answer for the meteorological community for many decades. It is not a trivial exercise.

Q123 Graham Stringer: Can you expand on that a bit? Why isn’t it trivial?

John Hirst: If you say, "We are forecasting showers", did the shower occur in your garden or someone else’s? Therefore, in terms of showers in that area, was the forecast right or wrong? Forecasts for specific uses can vary. The information we would provide a fast-jet pilot on crosswinds on a runway he is going to use for take-off would not be appropriate for the person running the farm next door.

Ensuring you have the right context, purpose, time scales and degree of granularity is really important. Then you need to go over a sufficient period of time-to go back to what Brian Hoskins was talking about-to make sure that you’re not checking just one event, but a whole series of events to see the aggregate performance.

Q124 Graham Stringer: Do I take it that you will co-operate?

John Hirst: We are co-operating.

Q125 Graham Stringer: You are in the process. You’ve said it is difficult. Do you have any assessment at the moment of how your medium-range forecasts compare with those of the private sector? To go back to the tabloids, the Daily Mail regularly says that Piers Corbyn-to name just one person-does it better than you do. Do you have any assessment?

John Hirst: Yes. We are collecting information on our and other people’s forecasts in this domain, and we now have a growing database that shows that actually our forecasts are generally more reliable than most others. We look at that for two reasons. One is to make sure that we understand whether the claims made are right or wrong.

There’s also an issue of sensible humility. If someone’s getting it right more than us-if someone’s doing better than us-we want to find out why and make sure that we take advantage of their understanding. There are a number of people-I don’t know whether to talk about names-who claim these things but don’t publish anything about their work and don’t share information about the science, if there is any, behind what they do, so it is sometimes difficult to make comparisons that have underlying scientific sense. That creates a difficulty.

Professor Slingo: Particularly in the area of seasonal forecasting, which I think Roger mentioned, it’s a probabilistic forecast, so you’re not right or wrong and you need a whole history of forecasts to decide the level of skill and what we call the reliability of the forecast. As an international community, we have still to come together and work out how to do this in a way that makes sense.

Again, Roger’s initiative is pushing us in that direction, so I welcome it, but we have to be careful, particularly when we’re going to probabilistic forecasts, that we don’t go into this in a naive way and try to approach verification in the way that we would for a deterministic forecast, which is what we’ve been doing for years, in terms of our performance measures.

For the things that Nick Baldwin talked about earlier, where we are set targets, we have had, with the World Meteorological Organisation, a long history of defining robust, verifiable targets or indices of forecast skill. We have now to go through that process for dealing with probabilistic forecasts and seasonal forecasts as an international community. We’re beginning that process, but it will take some time.

Q126 Graham Stringer: I don’t know whether my final question is answerable, but given the difficulties that you’ve just been through, is there anyone in the private sector or other national agencies who you feel is ahead of you in terms of their ability to do medium-range forecasting?

John Hirst: We are part of a consortium of leading met services working together under the auspices of the World Meteorological Organisation to develop these skills. Therefore we’re already engaged with scientifically based people doing this kind of work. Although we always keep an open mind, I don’t think there is an institution or a provider that gives a consistently and traceably reliable forecast better than the kind of work we’re doing.

Chair: Thank you very much for a very informative session.

Prepared 17th February 2012