Examination of Witnesses (Questions 101
WEDNESDAY 7 NOVEMBER 2007
Good afternoon, ladies and gentlemen, to a further evidence session
in the Committee's inquiry into flooding. I would like to welcome
Professor Edmund Penning-Rowsell, the Head of the Flood Hazard
Research Centre at Middlesex University and Professor Howard Wheater,
who is the Professor of Hydrology at Imperial College. Gentlemen,
you are very welcome. Can I thank you in advance for the written
submissions which you have both made. Professor Penning-Rowsell,
I would like to start by inviting you to try and put our problem
of flooding in the summer into some kind of context and perhaps
compare the experience we had in this country with other flooding
incidents, perhaps in a European context or perhaps, with thoughts
of what is happening, sadly, in Mexico, in an international context.
How serious was it?
Professor Penning-Rowsell: How
serious was 2007?
Professor Penning-Rowsell: It
depends what you take as your metric of seriousness. Some people
lost their lives, not very many fortunately; quite a lot of property
was flooded, but not tens of thousands; there was a lot of dislocation
of the local economy and people's lives in Gloucestershire and,
to a lesser extent, further up the River Severn, but if you look
internationally then the flooding problem we have in England and
Wales is not as serious, for example, as that experienced in the
United States of America, where regularly scores of people are
killed each year as a result of flooding, which tends to have
a different mechanism, largely driven by hurricanes and coastal
storms. If you look at the Continent of Europe, the Netherlands
is blessed with having no serious flood problem at all as a result
of major investment since 1953; Germany has suffered serious flooding
on the Oder and to a lesser extent on the Rhine. There is not
a good database on the extent of flood problems across Europe,
except the database in Belgium on loss of life, and this shows
that in most cases most countries have a more serious loss of
life from flooding problem than does England and Wales. So, in
that respect, we are fortunate that many years of investment have
mitigated the effect of flooding, and that has been the consistent
policy of government right back to 1930. On the other hand, we
are now experiencing greater dislocation, we understand, of our
society as a result of the flood experience. This is a relatively
new phenomenon and some features of the flooding in 2007 were
certainly different from previous flood events. That does not
categorically answer your question, except to say that I do not
believe that Britain is amongst the foremost in flood affected
countries and, certainly, if you look further afield to South
East Asia, you will find countries such as Bangladesh where the
flood extent is greater and the flood severity is greater. We
do not even have in Britain the largest proportion of flood plain
in the country. That is, again, in the Netherlands or Hungary.
So, I would be cautious about saying we have a flood problem above
the average in Europe and certainly not above the average in the
rest of the world.
Again, to put it into context, you deal with flood risk management.
If we take the different types, you made an interesting observation
when you said that there were some different characteristics of
the flooding this summer, I think, particularly with the urban
dimension, because up to now most of the focus has been on coastal,
river, estuarial flooding, but we had some other things that occupied
our minds. In terms of order of priority, order of risk, the things
that worry you as an expert, looking at the spectrum of flooding
that we have in this country, what are the things that cause you
the most concern? If you lose sleep at night on these matters,
what might be the cause of that?
Professor Penning-Rowsell: I do
not lose a lot of sleep at night as a result of that kind of problem.
I presume you do not live in a flood plain then.
Professor Penning-Rowsell: Quite
close, actually. I think if I were to characterise the three most
serious flood problems that I foresee in this country, they will
be, firstly, a repetition of a 1953 type coastal flood event,
although I think it will be true to say that the effect of that
event will be far less than 1953, for reasons that I could go
into. The second problem would be a repeat of the 1947 type event
on the Thames, and there was a similar event in 1894 which is
perhaps more serious, although it differs in different locations
because there are large areas of the lower Thames catchment which
are not substantially protected from flooding, particularly between
Windsor and Teddington, and there a major event, such as a repeat
of 1947, would cause tens of thousands of properties to be flooded
and scores of thousands of people to have to be evacuated for
upwards of a week and perhaps as much as a month. That is the
second of my problems. The third of my problems would be a repeat
of the 2007 event, and perhaps less the 1998 event, over a very
large metropolitan city such as London: because if you had 100
millimetres of rainfall over London, as Gloucestershire had 100
millimetres of rainfall, you would have a much, much more serious
situation. Those are my three worst scenarios.
In your evidence at paragraph nine you say, "The Environment
Agency proposed, after the Boscastle flood, to identify the areas
of highest risk to life from flooding. This study is urgently
Have you got any idea why we have not seen any results of that?
Professor Penning-Rowsell: No,
you would have to ask the Environment Agency that. It is quite
a complicated problem, but the flood mapping programme that the
Environment Agency has undertaken was to have had a loss to life
component to it, and some research was done by Hydraulics Research
Wallingford in that direction. I do not believe it has been adequately
followed up by maps or other types of information which would
indicate areas where loss of life was more likely than elsewhere,
and I think that is needed as an element of research which would
help the Agency and particularly the emergency response organisations
to cope more satisfactorily with the kind of events such as 2007.
I wonder if you can help us. You have taken us, if you like, into
the forecasting and the probability side of flooding and we are
humble politicians, we are not statisticians and we are not experts.
You were talking about events almost on a probabilistic basis,
but one of the things which we were alerted to when we did our
report on the basis of the Foresight Report was the growing severity
of an event as well as the question of the probability of things
happening, and as lay people, we might be seduced by recent events
to think that, to put it crudely, there are more big floods happening.
In other words, they are happening with greater frequency and
they seem to have to deal with more water. That is in its very
crudest terms. Yet the concept of preparing for flooding on a
frequency basis is defined in terms of events that occur one in
100 years, one in 150, one in 200, but they do not seem to talk
about the severity of the event. In other words, I am not clear
who defines what a one in whatever number of years' event it is;
so how do you decide on the quotient of severity of the event
and are we, with climate change, seeing the degree of severity
starting to increase? Therefore, the third question that follows
from that is, if any of those first two premises are correct,
true or provable, what should we do about it?
Professor Penning-Rowsell: That
is a very complicated question.
That is why I put it to an expert.
Professor Penning-Rowsell: Howard
can deal with probability and I will have a few words on severity
towards the end.
Professor Wheater: I think the
concept of a return period is very misleading. It is just a shorthand
way of talking about probability. So, when we talk about a one
in 100 year event, what we mean is we have got a giant dice and
every year we throw that dice there is a one in 100 chance that
in a particular year we will have a flood of a certain level of
severity or greater. So, the scenarios for climate change we can
talk about, but under certain scenarios in certain locations we
expect the flood risk to increase, and that means that the chance
of a large event will increase with time.
Let me stop you there, because that bit I understand: the rolling
of the dice. That is, if you like, gambling basically. We have
got gambling; we understand that. In terms of the eventjust
focus on the eventwho defines the event? Who says what
a one in 100 year event is?
Professor Wheater: You can define
it with respect to various criteria. You can choose those criteria.
Who does choose it though?
Professor Wheater: It depends
on the purpose for which you are doing an analysis, but the most
common one would be the peak flow in the river, that is the highest
discharge in terms of the amount of water moving down the river,
so many cubic metres per second. That would be the most common
criteria. Then what you would do is collect long records of data
and look at the highest flow each year, and then you would look
at the statistics of that and you would say, "Okay, that
flow is a one in 50 and that flow is a one in 20 year event."
If you were designing some flood protection, you might be interested
in what happened when, say, flood walls were overtopped. In that
case you might be interested in a certain volume of water above
a certain threshold, and you could do the same analysis but your
criteria then would be the volume over that threshold.
The reason I asked those questions is that we covered this in
the short-term against a background of unprecedented amounts of
rain falling in certain locationsHull, Sheffield, Tewkesbury,
for examplein very short periods of time (two days), getting
effectively a month or more than a month's rainfall at one fell
swoop. That is a pretty extreme event by our weathering. The question
that had occurred to me is that, in terms of when you are modelling
flood risk maps and trying to determine the policy decisions,
Professor Wheater, to which you adverted a second ago, it comes
down to the question now, when you are defining the event, as
to how you should define it. So one of my questions is: taking
into account the unknown variable of climate change but the reality
of what we have had to deal with, do you think we need to fundamentally
redefine these events and, against that background, re-examine,
therefore, the effect that a redefinition of those events would
have in deciding which areas are "at risk"?
Professor Wheater: The simple
answer is, yes. We have, in practice, a range of tools for looking
at the statistics of floods. Traditionally those tools, which
have been, incidentally, funded by Defra and MAFF, rely on climate
stationarity, so they assume an unchanging climate. If we move
into a situation with climate change, then we need a slightly
different range of tools. One of the important contributing factors
to flooding is how wet a river basin is before we receive the
heavy rain. Typically, you find that flooding is most common in
the winter because, although we do not have the highest rainfall
intensity, we have the wettest catchments when the rain arrives.
We then need to look in future climates at the combination of
dry summers, let us say, and wet winters to see how the balance
will play out. In terms of projections of the future, those are
dependent on climate models, and those climate models are very
good at getting temperature right for a global average. They are
not very good at getting precipitation right around the globefor
example, they cannot get it to rain at the right time in the tropicsand
when you come down to particular regions they are even more uncertain.
So, you will find that if you look at a range of different climate
models you will see quite a significant scatter of future impacts
on rainfall. Basically, you need to recognise that uncertainty,
run it through some hydrological models and then you can look
at the range of outcomes. In fact, Defra and the Environment Agency
have been funding research in this area to allow those kinds of
rainfall scenarios to be developed and to provide the models that
will allow the results to be generated. If I can touch on the
issue of the observed data, it is very difficult to look for trends
in both rainfall and flow data, particularly flow data, because
there are lots of compounding factorscatchments change,
people build houses, the land use changes, and so onbut
there is really no evidence, to the best of my knowledge, from
current observed records that flood risk is as yet increasing
in the UK from observed flow data. If we look at rainfall, the
picture is a little bit more complicated. Certainly there is evidence
that over the past 30 years there has been an intensification
of certain aspects of rainfall, but, on the other hand, 30 years
ago it was a slightly drier than average period. So, within our
immediate memory we have seen intensification but it is not clear
that, if you look back one or 200 years, there is a real significant
change of intensities.
So I do not misunderstand, Professor Wheater, what you said earlier
on, you seem to agree with the view that we need to revise our
current modelling arrangements and from the evidence I have read
some people question whether we have sufficient data to make that
decision, particularly in terms of what is the impact of climate
change. What we have had is the Chief Scientist Foresight Report
which postulates a number of scenarios of more intense weather
conditions. So, that is a projection saying these are things that
could happen, what we have is a series of events that have happened
and we have, if you like, our short-term memory of look what happened
this summer. Therefore, from a policy-making point of view, would
you say that there is some solid ground to say, against that background,
some of conjecture some of events, that we should revise our modelling
and do we have the tools sufficiently well developed to make the
kind of predictions that will ultimately guide the policy-maker?
One would conjecture again that what you would have is that certain
locations would become more risky to flooding and, therefore,
the question is: what do you do about it?
Professor Wheater: That is quite
a complicated question. I think that there are significant uncertainties
in modelling future climate, as you will appreciate. There are
uncertainties about how the global community will respond in terms
of future emissions, and so on. So, that is why there are various
scenarios to try and scope out the range of possible futures and
then within those scenarios the best tools that we have are models
that try and represent the whole of the global climate system,
and that is a pretty tall order. They can get some things remarkably
right, so they do a great job in explaining what has happened
to temperature over the past hundreds of years, but, as I said,
rainfall is difficult and more complicated, and what you find
is that the estimates of global climate models for rainfall are
not very reliable. So, there are two ways forward to try and come
up with better estimates. One is to improve the density of your
simulation of climate over a region of interest, and so we have
things called "regional climate models", which are high
resolution. If you look at a global climate model, it averages
the whole of the climate and the land surface over areas of the
order of 300 by 300 kilometres, which is very coarse. You get
a few grid squares over the whole of the UK. If you can nest a
fine resolution model within that global model, you can come back
to a grid that is maybe of the order of 40 kilometres, 50 kilometres,
something that is still a bit big but more useful. There are limitations
to that, because you have to take the boundaries from the coarse
model, feed them into the fine model and within the fine model
you cannot represent all the feedbacks and the oceans, and so
on, but that is one set of tools that you have in your tool box.
The Hadley Centre has a range of regional climate models which
it is using for looking at future climate scenarios. The other
approach is to say, there are certain things in climate models
that we can believe more than othersthose are things like
temperature, pressure and humidityand we can take those
pieces of information and use them to drive statistical models
that will then allow us to generate possible rainfall sequences
for the future, and that is one of the lines that is being funded
in the UK.
Let me finally, before I pass on to Mr Drew, ask you this question.
You have analysed some of the challenges using predicted models.
What we saw this summer in two days was a month's rainfall. Do
we have models whereby you could say, very crudely, "Okay,
I am going to apply that scenario uniformly across the whole of
England and just see what it throws up in terms of flood risk"?
Could you do that?
Professor Wheater: It could be
done, but it would be difficult, in the sense that the way in
which the UK is modelled is on a region by region basis and to
inconsistent levels of detail, but it could in principle be done.
Professor Penning-Rowsell: The
Foresight Report was not too far away from what you have just
said, which in a sense dumped a whole belt of flooding over the
whole country, not related to catchments but related to grid squares,
and looked at the consequences.
I think there was an element of surprise in Sheffield, for example,
about what happened. In terms of trying to identify where serious
flooding might in theory occur and, therefore, what the implications
would be, certainly looked at from the layman's point of view,
I would say if you could apply these things universally, what
pops up? Bearing in mind you have got concentrations of population,
is that kind of approach any use whatsoever in helping you to
decide your future flood protection analysis, or is it just an
interesting statistical game but without any sort of reason to
Professor Penning-Rowsell: That
is exactly what the Environment Agency is doing through its catchment
flood management plans: a slew of programmes to look at each catchment
in turn to look at exactly what you are suggesting. Where are
the hot spots? Where are the real pinch points? Where would investment
be best directed? At a strategic catchment level. That is something
that has been done over the last five or six years and that has
given us a view of the country which we did not have before.
To conclude then, does that process need in any way further refinement
or further investment to improve the quality of its outcomes?
Professor Penning-Rowsell: Probably.
Professor Wheater: Yes, I think
that---. Well, we can think about modelling at different scales.
There is work at the moment being done, for example at the Centre
of Ecology and Hydrology, funded, I think, by Defra and the Agency,
which is looking at a set of small catchments over the UK, running
different climate scenarios and seeing what the responses are.
When we work with the bigger catchment systems, it is a more complicated
modelling task because we have to consider the interactions of
all the contributing areas and the way in which floods are routed.
At the moment in the current stage of catchment flood management
plans, that is being done in a relatively simple way which is
not very amenable to looking at climate change. We actually need
an improved generation of models to allow us to simulate these
responses continuously so we can incorporate the dry summers and
the wet winters and then determine the outcomes.
Lynne Jones: We had a submission
from Professor Knight, who was very critical of the Environment
Agency's expertise and also of their modelling.
Chairman: Would you like to remind
us who Professor Knight is?
Q115 Lynne Jones:
Professor Donald Knight from Birmingham University. He criticised
the over-reliance and the bias towards hydrology and the lack
of consideration of hydraulics and hydrodynamics. Maybe your response
is, "Well, he would, would he not", I do not know. He
also says that there is a lack of in-house expertise within the
EA and is critical of the fact that there is no chief engineer
and also says that the 20 per cent rule is very much an underestimate.
Would you care to comment?
Professor Wheater: Yes, there
are two aspects of flood generation really. First of all, how
much water is going to get into your river and then what is going
to happen to that water as it moves down the river system? The
area that Donald Knight has been referring to is the movement
of water down the river system, and the hydrological problem that
he refers to is how much water is going to run off from the land
into the river system. So, of course you need to understand both
problems, but the chances are that you can go very badly wrong
in estimating the amount of water that gets into the river and,
if that is the case, it does not matter how good your river routing
model is if you have got the wrong amount of water in it. For
me, speaking as a hydrologist who is interested in the run off,
I think the biggest question is how much water is going to arrive
in the river, and then the second question, how is that going
to transmit downstream? The question of expertise in that area:
I think that probably there is not a large amount of expertise
within the Agency on the particular problem of river hydrodynamics.
However, the Agency does a lot of work through consultants who
are well qualified in those areas and have models which are developed
to a high level and applied widely, not only in the UK but overseas.
So they have expertise that they can call on without having particular
scientific expertise on the nitty-gritty of the computer code
of those models.
Q116 Lynne Jones:
You think that is adequate, because certainly Professor Knight
does not. He criticises it for outsourcing all its hydrodynamic
Professor Wheater: I think that
there is a generic issue really about how much work an organisation
like the Environment Agency does in-house and how much it subcontracts.
Q117 Lynne Jones:
I think it is concerned with the lack of continuity that you can
have people involved on an ad hoc basis rather than having the
in-house expertise on a long-term basis?
Professor Wheater: Yes. I think
that it is difficult for people who are purely managing external
programmes to really keep in touch. In an ideal world one would
have some in-house expertise that is actually working on the problems
and, hence, has a real in-depth understanding of them, and I think
in this area that is not something that the Agency has at the
moment. I think the Agency, in terms of its science programme,
has been through quite a lot of changes. It used to be, going
back ten years or more, that science was organised in a fairly
ad hoc way, but then they developed in my mind a rather good programme
of developing centres of technical expertise, but in recent re-organisations
those centres have been disbanded and the expertise is managed
in a rather different way. I am not entirely sure what the current
status of the Agency science programme is because I know it is
undergoing fairly radical change at the moment.
Q118 Mr Drew:
Can I take Professor Penning-Rowsell back to his paper to start
with, but it does relate to what I want to ask you about. You
said post Boscastle the Environment Agency were going to identify
these areas of highest risk of flooding. Is not that what the
flood maps already do? I accept that Norwich Union would argue
their maps are better than the Environment Agency's, and there
may be some justification for believing that, but how much more
information can you give people without effectively either worrying
them to death or accepting that these areas have such a risk that
you would not want to situate additional development there?
Professor Penning-Rowsell: There
is a balance to be struck there. The comment on paragraph nine
is about information of risk to life from flooding, and that has
not been followed through by the Agency in a way that I think
needs to be done.
Q119 Mr Drew:
Can you do that?
Professor Penning-Rowsell: Oh,
yes. We are getting much more research and intelligence about
the factors that lead to loss of life in floods. In general terms
it is to do with the velocity of flood waters, it is to do with
the depth of flood waters and it is to do with the vulnerability
of people who might be exposed to flooding. So in areas of substantial
water depth where velocities are large, the risk of loss of life
is much more obvious, much more apparent, much more real than
right at the edge of the flood plain where velocities might be
quite low and the flood water might be quite shallow. So there
is gaining intelligence about the factors that lead to loss of
life. It is not a perfect science by any manner of means, because
sometimes loss of life from floods is a function of the behaviour
of the people involved rather than the flood itself, but the Agency
has commissioned research under the Defra, EA Research and Development
Programme to look at this and some algorithms have been produced
which could result in maps produced by the Agency within its flood
plain mapping programme, which is relatively unsophisticated in
the sense that it is for public consumption and it is not in very
great detail. It is the kind of thing you can look up on the web
for your own house and see whether it is blue or whether it is
turquoise, indicating a flood or a serious flood. It does not
go further than that, although it does differentiate between areas
that are defended or not, but it does not say this is a very serious
flood, with high velocities and lots of deep water around, as
it might do for Tewkesbury or for Oxford or for the coastal locations.
It does not do that. That is the point we are making under paragraph
1 Ev 35 Back
Ev 462 Back