Examination of Witnesses (Questions 549
- 559)
WEDNESDAY 12 DECEMBER 2007
DR ANN
CALVER, MR
TERRY MARSH,
PROFESSOR JOHN
MITCHELL, MR
STEVE NOYES,
DR CHRIS
WEST AND
MS JACQUI
YEATES
Q549 Chairman:
We move on to the second part of this afternoon's evidence session.
May I welcome the star-studded cast before us. We are very grateful
to have you with us and also for your written submissions. Are
you able to help us at the Met Office because we would not mind
knowing what the forward forecast is in terms of winter and summer.
It might just give us a way of evaluating your forward forecasting
techniques. In your written evidence, and indeed in some of the
material you have been sending to Members of Parliament, you give
a mouth-watering array of different forms of forecasts that you
are involved in, even up to trying to forecast 100 years ahead,
which seemed a bold challenge, bearing in mind that some people
say you cannot even tell us what is happening tomorrow. Just give
us a flavour for the winter period and summer. Are we in for a
good one or a bad one?
Mr Noyes: In terms of the winter
ahead, it looks as though it is going to be colder than last year
but not colder than normal. We do issue a seasonal forecast for
the winter and then again for the summer. For the coming winter,
it looks like being colder than last year—remember last
year as particularly mild—but not colder than normal and
in terms of rainfall, round about normal rainfall amounts. We
have not yet produced our seasonal forecasts for the summer; that
will be something to look forward to in the spring.
Q550 Chairman:
At least it sounds like normal. I think one of the problems about
the science in meteorology obviously which underpins some of our
discussion is this question of deviation from what you describe
as "normal" and over what kind of time periods "normal"
is established. Looking at the factors which you describe that
affected the summer's weather, I think you were indicating the
opposite of the El Nino effect had occurred this summer and we
had things blowing in from further south than they should have
been and therefore we had an atypical condition in the United
Kingdom. Of course, as a result of that, you will have heard a
lot of discussion as to whether this is the beginning of some
kind of stream of predictable or unpredictable weather, extreme
weather events: is this climate change? Perhaps you could just
set the scene for us on the question about normality and whether
you need something abnormal, like the events to which you referred,
to occur for us actually to get the severity of flooding which
has occurred. One of the points that we want to pursue later in
our inquiry is the difference between the probabilistic effect
of predicting flood events and the severity with which they are
occurring. Perhaps we could start with the Met Office but others
can pitch in if you want.
Professor Mitchell: You have alluded
to the wide range of timescales that we cover, so the daily forecasts
that you see on the television for the next few days is very much
what we call a deterministic forecast. We would expect to see
those events unfold as they are forecast. We do take into account
uncertainties by running a set of forecasts from slightly different
initial conditions, and that gives a range of forecasts that allows
for uncertainty, but for the daily forecasts, we would expect
things to evolve very much as directly produced by the model.
When you go to seasonal forecasting, what we are looking at is
much more on a statistical basis. You cited El Nino as being one
of the world-wide, global variations in climate that we see on
anything from a three to seven year timescale. Certain phases
of that are associated with more common frequencies of certain
kinds of weather. This is particularly strong in the tropics,
so for seasonal forecast in the tropics, we will look at sea surface
temperatures. That is a very powerful indicator, particularly
of how rainfall around the tropics will be distributed. But for
our latitudes, the signal is much weaker, and that is why I think
Steve was quite right to be slightly more cautious in making a
forecast for winter, where we are looking very much at a statistical
approach. What we are saying is that this is what the most likely
outcome is. The third timescale you alluded to is forecasts for
10 to 100 years and there we are looking very much at the long-term
effect of increases in greenhouse gases and other factors. There
is an uncertainty there and we have to decide what the concentration
of those gases is going to be over the next 50 to 100 years, and
that is a socio-economic forecasting problem. Given a scenario
of greenhouse gases, we can then put those in our models and see
statistically how that changes the distribution of rainfall and
rainfall extremes into the future temperature and so forth. As
you go to the very long scales, they are very much statistical
in nature. We are not forecasting that on Christmas Day in 2055
there will be a particular temperature; we just say, "This
is the range of temperatures that we would expect at that time",
and the range of rainfall and so forth.
Q551 Chairman:
Can you help me to get weather forecasting timescales into some
perspective? When are you satisfied that a new norm has been established,
because obviously normal is made up of a series of short-term
events against a previously determined trend. At what point do
you say: "We are now in a new weather scenario"?
Professor Mitchell: Traditionally,
we have taken a period of 30 years. That tends to be the norm
specified in the World Meteorological Organisation, and of course
if you have a climate that is not moving, is not changing, it
is not very critical what period you choose. Now that we have
strong evidence that we are influencing climate on a decadal timescale,
it then becomes important to reconsider, as you have noted, what
period you do use. So the current normals are from 1971 to 2000.
Previously we used 1961 to 1990, so we do tend to move them on.
Where it has really become an issue, again as Steve has alluded
to, is that in giving our seasonal forecasts, we now have to say,
"It will be a cold winter but not as cold as the previous
one", and the meaning of "cold" has changed in
people's minds. People tend to remember the last few years. So
if we have had a succession of mild winters, and that is part
of an underling warming trend and we predict that the next winter
will be colder than perhaps the last few winters but not colder
than average, you have to be very clear that because of the long
underlying trend, what was a mild winter will appear as a cold
winter. However there is no particular methodology for choosing
what the reference period is.
Mr Noyes: If I could add to that,
this is a particularly important point in the context of how we
communicate our forecasts because traditionally, as John says,
we would refer to the 30-year period from 1971 to 2000. If you
take the recent summers and recent winters, all of them have been
above the average for that period, or most of them have. So, whilst
in communicating our forecasts to the public and to our customers,
we might well say that it is going to be warmer than normal, one
might say that that is the most likely scenario in an environment
where climate is gradually warming up. What we find, in terms
of making it much more relevant to the people who use our seasonal
forecasts, as John implies, is that we have to really link it
back to the most recent summers and most recent winters so that
people can put that into context. The reason for that is partly
because people's memories are relatively short and a 30-year mean
from seven years ago does not mean an awful lot, particularly
for people who have not lived that long, and also because the
climate is changing, so the backdrop is changing. It is becoming
quite difficult to explain what a warm summer really means if
you refer back to a 30-year period of some time ago.
Dr West: Could I add something
to what you said about a normal climate? I think we fail to appreciate
what is normal. Even without the fact that climate is changing,
we have short memories, so that in this room I guess some of us
can remember the 1962 winter, which was very cold, and rather
fewer of us the 1947 winter, which was very cold, and yet it is
those extreme hot and cold, wet and dry years that describe and
define "normal". What we perceive in the climate we
are familiar with is only a subset of the true range of what is
normal.
Q552 Chairman:
Part of the reason why I think this is an interesting area is
that the investments which have to be made in flood defences,
or shall we say "response to flooding", are of a very
long-term nature. What you have to deal with in terms of capacity
is, by definition, related to the weather events which cause the
flooding. Just give us a hint as to how much you can contribute
to that debate in territory in which you feel comfortable rather
than just taking a punt, bearing in mind it could affect the expenditure
of many millions of pounds.
Professor Mitchell: Certainly
in terms of looking at longer term infrastructure, one would be
prudent to take future climate change into account. I think there
are a number of points to make on that. The first is that in terms
of the size of the projected changes over the next 100 years relative
to the year-to-year variability, in terms of temperature we have
quite a strong signal to noise; in other words, the trend and
the change over 100 years is large compared with the year-to-year
or decade-to-decade variability in temperature. With rainfall
that is much less true. I think in England/Wales rainfall, if
you look at the year-to-year totals, they can be anything from
an average of about 60 mm, varying from 10 to 100, so there is
huge variability from year to year. Defining what happens in the
future, particularly in terms of extremes, is quite difficult
because of what in statistics we would call a low signal to noise
ratio. That is the first point to make. The second point is that
there are some seasons over the UK where we could have a fairly
good idea of what the changes are going to be. In winter we expect
increases and precipitation; that is true of all the models that
we looked at in the Hadley Centre, but also going to the recent
IPCC report, that of the Intergovernmental Panel on Climate Change,
pretty well all the models there show increases in rainfall over
the UK in winter. The difficulty comes with the summer, and it
is particularly relevant to this committee, and there we find
about half the models get an increase over the UK, particularly
the southern UK, and half the models get a decrease. The reason
for that is that when you look globally, you find at high latitudes
there is a consistent increase in rainfall and snowfall. Around
the Mediterranean region in particular, there is persistent decrease,
and in summer that divide just happens to fall in the middle of
the United Kingdom. That is why we are unable to give particularly
clear indications of what is going to happen in summer. I think
the other thing that we should bear in mind is that an event such
as we had in the summer was certainly quite a large anomaly in
the large-scale circulation and the ability of models to simulate
those anomalies is limited. For example, in UKCIP 08—this
is a series of scenarios which we are producing in conjunction
with Defra for UKCIP—those will certainly be very much the
best available predictions that we can make at the current time,
but there is still room for improvement. Essentially, as to where
we are on climate change, we have established that climate change
is happening; what we are unable to do at the moment is to make
reliable, accurate predictions on a regional scale, and that is
something which unfortunately will need more investment.
Mr Marsh: It is clear that the
models are strong in relation to temperature and there is a fair
degree of confidence about where we will be moving. The uncertainty
in rainfall is particularly important in relation to flood risk,
and that uncertainty in a sense is at its broadest in relation
to extreme events. I would make the point that modelling is very
much a complementary activity to monitoring and what we need to
capitalise on is the information that we have in the field, not
just contemporary but historical, to see if there have been any
tendencies or trends over time which are sensitive to climatic
variations.
Q553 Chairman:
Can we move to what did happen this summer. Perhaps you could
give us the official version of the events that led to these by
our standards—and I suppose by normal standards as well—quite
extreme weather conditions.
Mr Noyes: The meteorology at the
time during the summer period from late spring through to July
was atypical in the sense that there were the sorts of weather
conditions in terms of the way that the atmosphere was established
over north-west Europe that one might find, for example, at this
time of the year. In that regard, it was not what one normally
would expect. The particularly heavy rainfall events of 25 June
and 20 July fell on top of a particularly wet period. If you look
at the period from May through to July, across many parts of England
and Wales there was three or possibly four times as much rainfall
as would normally fall during those months, and so the ground
was very wet. Then we have had these particularly intense rainfall
events on those two particular days. In terms of trying to explain
a little bit about the atmosphere and what was going on, at this
time of the year (December) in this part of Europe most of our
weather comes typically from frontal systems that come in off
the north Atlantic, bringing with them cloud and rainfall and
sometimes snow and sometimes heavy winds. During the summer months,
what tends to happen more is that those frontal systems tend to
sit further north and west, often affecting parts of Scotland
and Northern Ireland but less so the southern parts of the UK
where our weather is affected more by continental Europe. In that
regard, what we were seeing this summer was that the frontal systems
were much further south. Why do we not get those sorts of rainfall
amounts in the winter if that is the norm for winter? It is because
typically in the winter the air is colder, the sea is colder.
To try to create, if you like, a context in which one can think
about the rainfall, in the tropics the atmosphere is very warm
and the seas are very warm; there is lots of energy and the air
can hold much more moisture when it is hot. In the summer, if
you get these frontal systems coming further south over the UK,
as we saw consistently week after week during those months, then
you have warmer seas and warmer air and more moisture that can
be put into the atmosphere and the atmosphere can hold more, and
therefore you get more intense rainfall. We have done a bit of
investigation in terms of these sorts of events. If you go back
to the last century, when do we get more exceptional rainfall?
What we discovered, and what we really knew but we made sure we
double-checked on that, is that the summer months are when you
get most exceptional rain. On the one hand, the systems we saw
this summer were atypical in the sense that we would normally
see them in the winter months, but getting very heavy rain is
normally, yes, a summer phenomenon, and that is linked to the
fact that the atmosphere can hold more moisture. There is more
energy in the atmosphere, so you get larger rain droplets and
more rainfall falling.
Q554 Chairman:
One of the things that is very interesting is where rain falls.
One of our colleagues who gave evidence earlier was talking about,
for example, Sheffield, the sodden nature of the uplands and the
fact that when the second deluge came the uplands were already
full and therefore the water ran off very quickly. Are you required
or do you as a matter of course inform anybody, any organisations,
about the amount of rain that has fallen in key areas like that,
so that they might be aware of the fact that areas that traditionally
had acted as soak-aways were filling up rather rapidly?
Mr Noyes: We do a number of different
things in that context. On the one hand, we have a weather radar
network, which we operate jointly with the Environment Agency,
and the equivalent in Scotland, with SEPA, and that radar network
is able to make an equivalent estimate of the rainfall that is
falling at any moment in time. That data is passed in real time
to the Environment Agency that uses that to look at making estimates
of the amount of rain that has actually fallen in river catchment
areas. They then use that in their own models to understand what
is happening to the rivers. Some of our colleagues will be able
to explain a little more about that in detail afterwards. The
Met Office's primary role is to provide forecasts of how much
rainfall is going to fall and then in the context of any warnings
that we issue, we certainly take account of what has happened
before. So we modify our advice in the context of the prevailing
conditions. Our forecasters are very much aware of saturated ground
when rivers are at a higher state of alert and they are in regular
dialogue with their colleagues in the Environment Agency. In the
run-up to both of the major events of this summer, we have, for
example, initiated daily conference calls with our colleagues
from the Environment Agency each morning when we would ask them
about the states of the rivers and the state of the ground and
understand what they think is going to happen from a flood point
of view, and we would be talking to them about the amounts of
rainfall that is likely to fall. Clearly, as I think you are implying,
50 mm of rain at this time of the year (December) probably would
not be an exceptional thing. However, 50 mm of rain falling a
day or two after what we saw on 25 June and 20 July would be significantly
important. Indeed, with regard for example to the COBR environment
which we were involved in, we were asked with the Environment
Agency to run some "what if" scenarios because, in the
context of probability, we were forecasting that there was a small
risk of a high rainfall event for the weekend following the 20
July event. It was not the most likely forecast. The most likely
forecast was for around 20 mm of rain to fall, which the Environment
Agency believed would not lead to any further flooding. However,
we were indicating that there was a small probability of a heavier
rainfall event. Together we worked out whether if we got 50 mm
of rainfall, which was a 15% probability of happening, that would
lead to more over-topping of the banks in the Severn area in particular.
We were able to advise the Gold Commands in Gloucester and also
the COBR community that this probability of further high rainfall
was unlikely to lead to further flooding. That is how we do it.
Q555 Mr Williams:
During that period of time, in Wales we did not get much flooding.
Was that because the intensity of the rain was less or was it
because actually Wales has got better resilience in terms of flood
prevention than the areas that did experience flooding?
Mr Noyes: Could I deal, first
of all, with the rainfall that actually fell and then perhaps
colleagues from CEH could deal with what then happens to it on
the ground? You are right that if you look at the difference from
normal, Wales had significantly less difference from normal in
terms of the amount of rain that fell on both occasions. Further
east from Wales, particularly if you take the July event, the
Worcestershire and Oxfordshire area had a much greater than normal
amount of rainfall than you experienced in Wales. Perhaps I can
pass to my colleagues and see what happens on the ground.
Mr Marsh: I would agree with that.
The rainfall in Wales was less and, relative to the long-term
average, Wales is a wetter country obviously than the lower Severn
basin and the Avon itself. It was principally a function of the
rainfall. I can say a bit more about what happened to the rivers
in 2007 if it is appropriate. Localised intense downpours in the
summer are not rare in this country and localised severe flooding
is not that rare. We remember Boscastle, which was extremely damaging.
In 1952, the Lynmouth flood in the summer sadly took 34 lives,
so that was a disaster, and that was a summer event. What distinguished
this year's flooding was its regional character—indeed,
its multi-regional character. That just does not happen in the
summer. What is the reason for that? We have heard the synoptic
background in terms of the weather, the weather patterns being
unusual. It is often reported that the May to July period was
the wettest on record. That is true but it under-sells the wetness
of that period. The May to July period registered 400 mm for England
and Wales. That may not mean too much but the highest in the 20th
century was about 100 mm less than that. What you got this year
was seven weeks additional rainfall, if you like on top of the
highest rainfall in that period, and only in that period, in the
20th century. This rainfall was outstanding. With rainfall of
that magnitude, you are going to get extensive flooding. If you
look at the rivers of the country as a whole and try to estimate
the outcomes for England and Wales in June and July of this year,
then as a reasonable estimate it was twice what we have ever had
in June and July before. That is only over a period from 1961.
We have data over that successive period. It really was outstanding.
In those areas in a broad belt from Yorkshire down through to
the Bristol Channel embracing much of the rivers and the upper
Thames catchment, many rivers registered their highest flow on
record. Again, that is remarkable for the summer period. For much
of the rest of the country, the experience of river flows was
very notable for the summer but not outstanding in relation to
winter flows. Just to pin a couple down, in the Thames for example
where the flooding was severe in the headwaters and not so severe
lower down, the flow was the highest since 1903 for this summer
whereas where the flows were most severe, in the Doncaster area,
south Yorkshire, Sheffield and in the Warwickshire/Avon and the
lower Servern, the flood magnitude there matched or exceeded that
of the 1947 event. The 1947 flood was the largest; it was the
most extensive flood in the 20th century in England and Wales.
It was entirely different in character to the summer flood we
have now, but in a few areas, 2007 exceeded that. We are talking
about a very singular event.
Q556 Chairman:
Let me try to put together, if I may, the scenarios that you have
painted. Mr Marsh, you have given me what I call a forthcoming
attraction picture. Here in the spring and the summer there was
this increasing amount of rain. You guys in the Met Office are
being asked to make certain predictions. As a layman looking at
this and with the benefit of hindsight, I am thinking to myself:
if they knew that the volumes of water that had been falling in
various locations was so far above the normal level, should not
bells have started to ring somewhere that if it carried on like
that, there might be some problems? One of the parts of the analysis
that we are looking at is: how did authorities react to the flooding?
One of the factors, for example, was: should the Environment Agency
in the specific circumstance have moved flood defence barriers
to a certain location? No, they could not because the flooding
had already started; the roads were blocked and they could not
get the kit in the right place. I think what I am asking is: do
we learn any lessons from marrying up, Mr Marsh, if you like,
your scenario of "oh look, what's happening; this is not
really the normal thing" with, Professor Mitchell, your guys
saying, "We seem to be seeing something rather abnormal in
the way that weather is happening" so as to be able to say
to local authorities, the Environment Agency and water companies,
"Something may be out there which is not right. Be prepared".
I do not think I am getting the message that these two bits of
information were put together ahead of events. It was only afterwards
that everybody started to think about what was coming next.
Mr Noyes: Can I deal with the
Met Office's forecast, if you like, and the warnings that we issue?
In regard to saying that these types of things happened only after
the event, I would certainly challenge that in that the Met Office
for both events gave early warning in advance that the two particular
events that led to the flooding were likely to happen. We were
particularly pleased with how accurately they were forecast. There
has been a significant step-change in capability over the last
five years. If we go back five years, it would have been much
more difficult for us to have given so much confidence to that
early warning as we were able to do this year. We have seen an
improvement in our capability. There is still more improvement
that we want to be able to deliver to enable even earlier warning.
Chairman: Forgive me for interrupting.
Coming back to Mr Marsh's very interesting picture, the message
I get, hearing him describe if you like the run-in to this, for
the first time in a very objective way is that if the kind of
accuracy of forecast that you were dealing with was visited upon
the scenario that Mr Marsh has enunciated, you might have said,
"If we are going to get this amount of rain and we are forecasting
it just a little bit ahead here, it is going to come on top of
what Mr Marsh says is already a pretty serious abnormal situation
and it will amplify the effect on the territory where this rain
is going to come down." The question I am groping towards
is: should those who had access to this information have spotted
what was coming earlier than they did? Patrick Hall wants to follow
up on that.
Patrick Hall: I understood Mr Marsh to
be describing the events after they had taken place. I was tempted
to think that one might say, in the face of the very exceptional
nature of those events that Mr Marsh so succinctly described,
that there is a limited amount one might be able to do anyway.
That does not mean you do not try. That was the message I received.
He was talking about what had happened after the event. That,
he said, did not contribute to forecasting. It is the Met Office's
job to forecast.
Q557 Chairman:
I appreciate that. Let me ask this question. At what point could
you have, if you like, produced the data which said: we have already
had an awful lot of rain and it is above the norm?
Mr Marsh: I should stress that
we have no operational role here, and so it is correct that this
is post hoc after the event and after the event you can
see that it was extraordinary. If during the event it was possible
to recognise a synoptic pattern which tells you something about
how the condition is likely to evolve, clearly that is something
that has be capitalised on, but it is not my area of expertise.
Q558 Chairman:
You are not saying, as an expert who looks at the way that rainfall
and rivers interact, that during the course of the spring and
the early summer when this lot is starting to build up you could
not have said, "Something unusual is occurring"?
Mr Marsh: I would not run with
the term "build up". I was describing the period as
a period. With that totality of rainfall, you would get extensive
flooding.
Q559 Chairman:
I am leaping ahead in my analysis of where you can contribute.
I accede to Mr Hall's cautionary tale that I should not build
too much on what Mr Marsh has said.
Mr Noyes: I would not accede too
much in the sense that I think there is more that could be done.
I think the Environment Agency will probably have to answer for
themselves in terms of how much they were aware of the situation
that was happening on the ground. Because they monitor the rivers
in real time, I am sure they were very much aware of the state
of the rivers during those few months. What we were able to do
during the summer, which we would not perhaps have been able to
do so effectively a number of years ago, was work much more closely
with the Environment Agency to look further ahead to what might
be going to happen a week in advance. The difficulty we have is
that at the moment, if you think of the environment as being a
system in itself—and you are probably familiar with the
water cycle where the water falls from the atmosphere, lands on
the ground and goes back up again—is that typically at the
moment we model that system in two different parts. We forecast
the atmosphere and then forecast separately what is going to happen
to the water once it has landed on the ground. At the moment,
with the way that we are able now to forecast a long way in advance
and from that produce early warnings, the way the Environment
Agency system works is that they then model the rainfall once
it has fallen. So their models are dealing very much with more
of a real time nature rather than a long-term early warning. Where
I think there is room for gaining on capability is to integrate
those two capabilities much more into a whole system where you
are extending the longer range ability to project what is going
to happen to the rivers when the rainfall falls five days hence.
|
