Current warming and likely future impacts - Climate Change: Biological and Human Aspects - page 328

Do not be surprised by extreme weather events in our greenhouse future. But
equally do not be quick to attribute them to climate change as there is a difference
between an anomaly within background noise and a statistical trend. To overcome this,
we need statistical analyses of what might be expected (1) with natural variability
assuming no carbon dioxide forcing and (2) with climate forcing from additional
anthropogenic greenhouse gases. Both scenarios need to be compared with reality as
revealed by environmental measurements.
In 2004 a small UK team led by Peter Stott performed such an analysis to examine
whether the 2003 western European heatwave was due to either (1) natural variability
or (2) global warming, as per the previous paragraph. These two scenarios were
derived using one of the more sophisticated climate models available at the time, the
HadCM3 climatic model developed by the UK's Meteorological Office at the Hadley
Centre for Climate Prediction and Research. The model took into account the impact
of actual volcanic eruptions and also solar variability specifically for the south and
western part of the European peninsula. It was run three separate times simulating
June-August temperatures as if there were human additions of greenhouse gases and
over the period from 1900 to 2003 and then on to 2100: these are type 2 scenarios as
per the previous paragraph. It produced three temperature lines with a typical annual
variation of about 1 K and this followed both actual meteorological measurements

and also a smoothed, few-year-filtered, line to 2000. The actual meteorological line
and, consequently, the smoothed line both lay within the model's annual variability
band through to the 1990s. Consequently, the smoothed line also rose, albeit erratically
due to actual annual, volcanic and solar variation, by about a degree along with the
simulated lines, so suggesting that the model effectively captured anthropogenic
greenhouse warming.
The model was run again, but under a scenario assuming no additional greenhouse
forcing over this time - that is, as per scenario 1 in the above paragraph - and so only
included anthropogenic greenhouse forcing up to 1900 (and not beyond that year).
This non-global warming line largely followed (intertwined with) both the natural
and smoothed natural lines for the 20th century's first three quarters. However, for
the century's final quarter this simulation continued horizontally, albeit erratically as
if with natural variation, so that by the century's end there was nearly a 1 K difference
between it and both the actual instrumental record as well as the anthropogenic
greenhouse simulation. The inference clearly was that this 1 K difference at the end
of the 20th century was due to anthropogenic emissions of greenhouse gas.
However, looking at the actual meteorological line, not only did it broadly follow
the model's global warming scenarios but summer 2003 stood out as being about
0.75 K warmer than any other previous June-August period on record and is thought
to have given parts of western Europe the warmest summer for more than 500 years.
Indeed, the next warmest summers have all taken place since the mid-1980s and this
too was reflected by the three global warming model scenarios. It suggests that the
model successfully captures the annual likelihood of an exceptional summer heatwave
in western and southern Europe even if it cannot predict the actual year in which
such a heatwave might take place.
Having established that the HadCM3 climatic model captures both global warming
and the annual likelihood of extreme events, what did the model predict for the 21st

±