Memorandum submitted by the Met Office
Note that the research at the Hadley Centre
on which this submission is based, is largely funded by the Global
Atmosphere Division of Defra, with additional resources from the
Ministry of Defence and the European Commission. The scientific
results have been published in the peer reviewed literature (eg
Nature) or are being prepared for publication.
1. The terms of reference of this inquiry
are concerned with the policies required to keep the UK on track
in reducing greenhouse gas emissions and the government's role
as chair of the G8 and as President of the EC in driving forward
the Kyoto and post-Kyoto agendas. On the first point, the scientific
work of the Hadley Centre underpins the UK government's commitment
to reduce greenhouse gas emissions. This submission details some
of the latest results, strengthening the evidence, on which the
case for a reduction is based. On the second point, there are
a number of key scientific results which highlight the regional
and global consequences of a range of emissions scenarios which
include a range of mitigation options. The main topics covered
a. The fingerprint of the effect of human
activity on regional climate change.
b. The levels of greenhouse gases needed
to stabilise climate.
c. High impact events under climate change.
d. Changes in extreme events in the next
100 years and measures of uncertainty. With a focus on major cities
in G8 countries.
2. Key Hadley Centre findings on these topics
a. A considerable amount of evidence already
suggests that much of the observed 20th century global warming
has been driven by human activity. New evidence shows that human
activity has caused warming on regional scales too.
b. Stabilisation even at quite modest levels
of greenhouse gases may still require some adaptation, but early
action slows the rate at which adaptation will be required. Preliminary
calculations show that feedbacks between climate and the biosphere
accelerate climate change, implying that lower emissions are needed
to stabilise climate.
c. Even modest global warming in the next
100 years would lead to the melting of the Greenland ice sheet
on much longer timescales (of the order of millennia). On the
other hand, our results indicate that the Gulf Stream will probably
weaken but not collapse.
d. Changes in extreme events are the way
that most people will first experience climate change. Increases
in summer extreme high temperatures are likely to be much larger
than increases in mean temperature. For example in double CO2
conditions, mean summer temperatures are predicted to increase
between 2-7ºC deg in London, Paris, Toronto and Washington,
whereas extreme high temperatures are likely to increase by 10-15ºC.
Also, the percentage of dry days is predicted to increase by more
than 40% in many regions with associated decreases in summer mean
rainfall of over 40%.
We would be able to present some of these and
other new results at the forthcoming G8 scientific conference
at the Hadley Centre, recently announced by the Prime Minister.
A. THE FINGERPRINT
Figure 1. [Not printed, the information
is available at www.metoffice.gov.uk]
3. Observations of temperature show that
on average the globe has warmed substantially over the 20th century
but that there have been large regional variations in the amount
of warming. The Intergovernmental Panel on Climate Change (IPCC)
concluded in their Third Assessment that there is new and stronger
evidence that most of the warming observed over the last 50 years
is attributable to human activities(1). Previous analyses have
looked at temperature changes over the globe as a whole rather
than over individual continents, but recently the Hadley Centre
has examined the causes of 20th century temperature change on
the continental scale. Here, the focus is on the landmasses of
North America, Asia, South America, Africa, Australia and Europe.
The modelling study investigated the historic impact on the climate
greenhouse gases alone;
the combined effect of anthropogenic
sulphate aerosol, lower atmosphere and stratospheric ozone; and
the combined effect of volcanoes
and changes in the output of the sun.
The optimal detection method shows there is
a significant greenhouse gas warming signal in all of the continental
regions considered (Figure 1, left hand side). Temperature changes
from other anthropogenic and from natural factors are detected
in some but not all of the continental areas, since these forcings
are weaker and more uncertain than greenhouse gas forcing. Therefore,
we have more confidence in attributing a man-made greenhouse gas
component to continental scale temperature changes than in attributing
other factors. The right hand side of Figure 1 shows the temperature
changes in the model results compared to the observations for
each continental region. The increases in greenhouse gases caused
increased warming as the century progressed. This was balanced
to a greater or lesser degree, depending on region, by aerosol
cooling. In general there is good agreement between observed and
Footnote: (1) Page 158 of Climate Change 2001.
Synthesis Report. IPCC.
B. THE LEVELS
4. The ultimate objective of the United
Nations Framework Convention on Climate Change (UNFCCC) is to
achieve ". . . stabilisation of greenhouse gas concentrations
in the atmosphere at a level that would prevent dangerous interference
with the climate system." Even after greenhouse gas concentrations
are stabilised, it will take a considerable amount of time for
a balance to be reached between the incoming solar radiation and
the heat lost from the planet. If greenhouse gas concentrations
were stabilised today, which would require an immediate reduction
in emissions of around 60%, we would still expect the temperature
to eventually rise by another 1ºC due to the inertia of the
climate system. Sea level rise may be more of a problem due to
the long timescale for thermal expansion and potential melting
of ice sheets.
5. We present our first estimates of the
temperature and sea-level rise during the 22nd century for two
Proxy for stabilisation at 550 ppmIPCC
SRES B1 emissions and concentrations up to 2100 followed by constant
greenhouse gas and aerosol concentrations at year 2100 levels
for 100 years.
Proxy for stabilisation at 750 ppmIPCC
SRES A1B emissions and concentrations up to 2100 followed by constant
greenhouse gas and aerosol concentrations at year 2100 levels
for 100 years.
Note these scenarios imply unrealistically sharp
reductions in emissions at 2100, but are used here to give a general
impression of two approximate stabilisation concentration profiles.
The following results were calculated using a simple climate model
and will be repeated using our full coupled ocean-atmosphere model.
Figure 2 shows that for both stabilisation scenarios the global
mean temperatures continue to increase between 2100 and 2200,
but at a reduced rate.
6. The sea level also continues to rise
beyond the 2100 concentration stabilisation point. On this timescale
the major component of the sea-level rise is the thermal expansion
of the oceans. Because the oceans have a large thermal capacity
they take a long time to adjust to changes in atmospheric greenhouse
gas concentrations, and the commitment to future sea-level rise
beyond 2100 (expressed as a percentage of the 21st century rise)
is much greater than the commitment to temperature rise. The rate
of sea-level rise 100 years after stabilisation is only slightly
less than the rate when stabilisation occurred.
The effect of the carbon cycle on stabilisation
of atmospheric CO2 concentrations
7. The concentration of CO2 in
the atmosphere depends on the amount emittedfor instance,
from the burning of fossil fuels and changes in land useand
the strength of carbon sinks, such as the ocean and biosphere,
which remove CO2 from the atmosphere (Figure 3). As
the atmospheric concentration of CO2 increases, so
does the ability of vegetation to take up CO2 from
the atmosphere (the carbon fertilisation effect). However, the
increases in CO2 lead to changes in temperature and
rainfall, which can affect natural carbon sinks. Over land, climate
change can alter the geographical distribution of vegetation and
hence its ability to store CO2. In the Hadley Centre
coupled climate-carbon cycle model, we find that climate change
results in a dying-back of the vegetation in northern South America.
Climate change also affects the amount of CO2 emitted
by bacteria in the soil. In the ocean, changes in circulation
and mixing, which accompany climate change, alter the ocean's
ability to take up CO2 from the atmosphere. In addition,
the warmer oceans absorb less CO2. In order to include
all of these feedbacks, it is necessary to treat the carbon cycle
and vegetation as interactive elements in full global climate
modelling (GCM) experiments. This approach was pioneered by the
Hadley Centre and the first results were reported at CoP6 (Conference
of the Parties, 6).
8. In principle, there are an infinite number
of emissions pathways that lead to stabilisation of atmospheric
CO2 concentrations at a given level. IPCC technical
note 3 discussed two alternative emissions pathways which would
stabilise CO2 concentrations at a particular level:
the IPCC "S" emissions, and the emissions estimated
by Wigley, Richels and Edmonds, the "WRE" emissions.
These emissions were calculated using a simple carbon cycle model
that took account of the carbon fertilisation effect, but other
feedbackssuch as that associated with the change of vegetation
patterns or the oceans, due to climate changewere not included.
To investigate such effects we used a simple climate carbon-cycle
model, which includes the feedbacks from vegetation, soils and
the ocean. This reproduces the results of the full Hadley Centre
coupled climate-carbon cycle model, and was used to make new estimates
of the concentrations resulting from the WRE550 emissions scenario.
Figure 4 above shows the CO2 concentrations
that would result. Without including the feedbacks, the emissions
eventually lead to stabilisation of CO2 concentration
at around 550 ppm, as intended (below). However, when the more
comprehensive feedbacks are taken into account, the CO2
concentration rises much higherto 780 ppm by 2300. Thus,
the effect of carboncycle feedbacks is to allow a greater fraction
of CO2 emissions to remain in the atmosphere.
9. In summary, preliminary calculations
show that including the feedbacks between climate change and the
carbon cycle greatly reduces the "allowable" emissions
that lead to CO2 concentration stabilisation at a given
level. It does this by reducing the strength of carbon-dioxide
C. HIGH IMPACT
Will the Greenland ice-sheet melt?
10. The Greenland ice-sheet would melt faster
in a warmer climate and is likely to be eliminatedexcept
for residual glaciers in the mountainsif the annual average
temperature in Greenland increases by more than about 3ºC.
This could raise the global average sea-level by seven metres
over a period of 1,000 years or more. We show here that concentrations
of greenhouse gases will probably have reached levels before the
year 2100 that are sufficient to raise the temperature past this
11. At present, about half of the snow falling
on Greenland melts and runs off as water, and the remainder is
discharged in the form of icebergs. Climate change caused by higher
greenhouse-gas concentrations is expected to produce both higher
temperatures and greater precipitation, but most studies conclude
that the increase in melting will outweigh the increase in snowfall.
For an annual average warming of more than 2.7ºC, the melting
exceeds the snowfalla situation in which the ice-sheet
must contract, even if iceberg production is reduced to zero as
it retreats from the coast. For a warming of 3ºC, the ice-sheet
loses mass slowly and over millennia might approach a steady state
in a smaller inland form. For greater warming, mass is lost faster
and the ice-sheet is likely to melt away. Calculations of Greenland's
temperature have been done using different scenarios of carbon
dioxide increases followed by stabilisation over the next few
Figure 5. Predicted warming of Greenland over
the next few centuries. The dashed line is the 2.7ºC threshold.
[Not printed, information is available at http://www.metoffice.gov.uk/research/hadleycentre/pubs/brochures/].
12. Figure 5 shows that the 2.7ºC threshold
is passed in all but one of the 35 combinations of model and stabilisation
level. The lowest carbon dioxide concentration considered was
450 ppm Given that this level is exceeded before 2050 in all of
the IPCC report's emission scenarios, and that carbon dioxide
is not the only greenhouse gas, we conclude that the Greenland
ice-sheet is likely to be eliminated by anthropogenic climate
change unless much more substantial emission reductions are made
than those envisaged by the IPCC. This would mean a global average
sea-level rise of 7 metres on millennial timescales.
13. This study has been extended at the
Hadley Centre, by using its climate model to simulate and predict
the evolution of the Greenland ice sheet over several thousand
years. This experiment is novel in that changes in the ice sheet,
such as the height of the ice or whether the ground is covered
in reflective ice or dark soil, are fed back into the climate
14. Figure 6 shows that over the 3,000 years
following a quadrupling of atmospheric greenhouse gas concentrations,
the ice sheet recedes from most of Greenland. By the end of the
simulation, it exists only on the mountainous ground of the East.
The fresh water released from this loss would cause a sea level
rise of around 7 metres. Earlier results suggested that if the
ice sheet is removed in this way it would not recover, even if
greenhouse gas concentrations were significantly lowered. The
next task is to understand at what point the melt down of Greenland
Figure 6. The Greenland ice sheet will melt
over the course of 3,000 years. Red indicates thick ice while
blue indicates thin (or no) ice. [Not printed, information is
available at http://www.metoffice.gov.uk/research/hadleycentre/pubs/brochures/]
Will the Gulf Stream collapse?
Figure 7. [Not printed, information is available
Figure 7 Simulations using the HadCM3 climate
model of the strength of the Atlantic thermohaline circulation
from 1930 to 2000 (using historical variations of greenhouse gases,
sulphate aerosol, solar radiation and volcanic dust). The simulations
show a freshening of the Labrador Sea from 1950-2000, as has been
seen in observations, but this is associated with a slight strengthening
of the thermohaline circulation over the same period, rather than
a weakening as has sometimes been suggested. When the simulations
are extended forward from 2000 to 2080 (using a projection of
future greenhouse gases and aerosols), both trends are reversed,
with a salting in the Labrador Sea and a weakening thermohaline
15. A key question in climate research concerns
the stability of the thermohaline circulation, a system of large
scale currents including the Gulf Stream in the North Atlantic
Ocean, which carries heat from the tropics to higher latitudes
as cold salty water sinks near the poles, drawing warm water north-eastwards.
Recent observations have shown a reduction in the amount of salt
in the seawater deep in the northwest Atlantic, and this has been
interpreted by some as an early sign of a weakening thermohaline
16. The Hadley Centre climate model (Figure
7) shows that the observations are in fact consistent with a slight
strengthening of the thermohaline circulation since the 1960s.
Nevertheless, the model predicts that in future it will weaken
somewhat as a result of global warming. The model suggests a reduction
in the strength of the Gulf Stream by as much as a quarterbut
not a collapse. However, even with this reduction in the Gulf
Stream, the net result of climate change will be a warmer Europe.
D. CHANGES IN
NEXT 100 YEARS
17. Good progress is being made in predicting
climate change on global and regional scales. Nevertheless, substantial
uncertainties remain which are a hindrance to estimating impacts
and hence to efficient adaptation. Uncertainty in emissions is
dealt with by using a range of emissions scenarios (eg see above).
The other major uncertainty is in modeling the climate system.
To address this, the Hadley Centre has developed a suite of physically
plausible models. So far an "ensemble" of 53 models
(soon to be increased to over 100) has been used to estimate of
the likelihood of different climate change events. We have extracted
temperature and precipitation information from these results for
major cities within the G8 countries.
18. Changes in extreme events are the way that
most people will first experience climate change. The Hadley Centre
has focused initially on temperature and precipitation as there
is most confidence in the results for these fields and they are
of direct relevance to such concerns as heat stress, flooding,
19. Figure 8 shows changes in summer
temperatures from 53 different climate models arising from doubling
atmospheric CO2 concentrations with respect to pre-industrial
The blue bars represent the distribution of the changes in mean
daily summer temperature by the different models. For example,
Berlin has most models showing changes in the mean of between
2 to 5ºC with one model showing 11ºC. The red bars represent
the change in the hottest
day in the summer with values for Berlin ranging from 3 to 19ºC
but with the majority of models predicting changes of approximately
9ºC. This figure indicates that for most of the cities selected,
changes in the extremes of summer temperature are likely to be
far greater than the changes in average summer temperatures. Such
changes in extremes will have significantly greater impacts than
the more modest (although still substantial) changes in the means.
Figure 8. Distribution of changes due to doubling
atmospheric CO2 concentration in daily maximum temperature
for the months June, July and August for 53 climate models. Grey
bars represent mean daily temperatures and black "hottest
day of season" (Average and "hottest day" defined
as 50th and 99th percentile respectively).
20. To understand the changes in daily weather
more fully it is useful to see how the whole distribution of possible
daily values changes with increased CO2. A useful statistical
measure to do this is percentile values. For example the 90th
percentile is the value below which 90% of all the daily values
reside. If we look at how increased CO2 changes each
percentile we can form a picture of how the whole distribution
is changing, both extremes and more normal values. Figure 9 shows
this for changes in daily wintertime (December, January, February)
precipitation due to doubling atmospheric concentration of CO2
for the same eight cities. The upper and lower lines represents
the spread of results given by the 53 different climate models
and so give a measure of the uncertainty in the future predictions.
To illustrate, for London the range of changes in the 50th percentile
span both negative but predominantly positive changes and with
a central estimate of 10% increase in daily winter precipitation,
so we would conclude that there is moderate confidence that typical
wet days will be getting moderately wetter. For more extremes
rainfall, as defined by the 80th percentile and above, all models
show more substantial changes (>20%) with the uncertainty range
significantly clear of 0% change, thus we would conclude that
there is high confidence that extreme daily rainfall will increase
substantially for London in winter.
Figure 9. Changes in the distribution of daily
wintertime (December, January, February) precipitation due to
doubling atmospheric concentration of CO2. Presented
as changes in all percentiles. Upper and lower lines represent
the central 80% of results from 53 climate models.
21. For northern hemisphere summer much
of the land surface shows an increase in the number of days without
rain (Figure 10a). This is particularly so for European countries,
Canada, central USA, Russia, Africa and the Pacific east coast.
Substantial reductions in seasonal rainfall are also seen for
many of these areas, with particularly severe drying over Europe
and northern South America with seasonal reductions of more than
40% (Figure 10b).
Figure 10(a) [Not printed, information is available
The average percentage change in days without rain in June, July
and August for 53 climate models due to doubling atmospheric CO2
Figure 10(b) [Not printed, information is available
The average change in mean June, July and August precipitation
for 53 climate models due to doubling atmospheric CO2
22. In order to develop effective strategies
for mitigation and adaptation the following areas need further
1. Better assessment of uncertainties.
2. More understanding of changes in extreme
events (eg of temperature, precipitation).
3. More detail on regional scales.
4. Better understanding of the likelihood
of high impact events.
5. A better link between work on predictions
of climate change and the impacts of climate change.
The Hadley Centre is involved in research on
all of these issues.
28 September 2004
2 Summer and winter are defined here as the months
June, July, August and December, January, February respectively. Back
All results presented here have given equal weighting to each
of the 53 climate models. In reality some models will simulate
current climate better than others and it would therefore be appropriate
to give greater weight to the future predictions of such models.
Although this will be incorporated in future analysis no such
weighting has been applied here. Back
"hottest" in this context is defined as the 99th percentile. Back