Dungeness in Danger
by Dennis LeggettAppendix C
IPCC and Munich Re Reports
The Intergovernmental Panel on Climate Change (IPCC) published their third and most recent report during January and February 2002. The report was written by 426 authors coordinating the work of many others around the world. It was considered by 440 government and expert reviewers.
The members of the Panel are divided into interest groups. Groups 1 and II reported that for the 20th century
• globally averaged surface air temperature rose 0.6 +/-0.2 °C. The rise was much faster in the latter part of the period
• previously predicted physical and biological alterations have occurred
• coastal defence strategies have moved from the hard protection of sea walls and groynes to the soft protection of beach nourishment and managed retreat. They note that "....coastal and marine management (is) most effective when incorporated with.... disaster mitigation plans and land use plans."
Summary for Policymakers Group II (1)They forecast that by 2100
• surface air temperatures could rise 1.40 °C min to 5.80 °C max. Most forecasts cluster around 2 °C i.e a two to threefold increase on the previous century
• globally averaged sea level would rise 0.09 to 0.88 metres .
• the intensity and frequency of extreme weather conditions would increase, although at present the risks cannot be quantified. "The extent and severity of storm impacts, including storm-surge floods and shore erosion, will increase as a result of climate change including sea-level rise. High-latitude coasts will experience added impacts related to higher wave energy and permafrost degradation."(1)
Munich Re is one of the two largest reinsurance companies in the world. It has formed its own geophysics research group, Geo Risks Research, to provide up to date knowledge of climate change and scenarios of the future. It's data is used to make risk estimates underlying its contracts with primary insurance companies.
Since early 1990, it has issued warnings, as has Swiss Re, about the potential for unmitigated enhancement of the Greenhouse effect to generate more intense and more frequent storms than those in the twentieth century.
Munich Re, in their own reports on property-catastrophe losses conclude that steepening meridional temperature gradients, will generate more severe storm tracks in Europe. In a press release (2) they state that storms in Europe are still an underestimated risk.
Statistical Uncertainties In all scientific enquiry it is recognised that all measures have a degree of uncertainty. Statistical treatments attempt to quantify the limits of those uncertainties. It is essential to realise that the use of any statistic requires at some point a subjective decision. Thus the I.P.P.C panels produce forecasts ranging in anticipation and the Environmental Agency choose what they consider to be the most likely forecast upon which to base their specifications for sea defence strengthening.
The pitfalls in subscribing definite qualities to statistical estimates are shown in the sea defences of Dungeness Power Stations.
In all the Environmental Agency's estimates for improvement of sea defences along the coast from Hythe in Kent to Pett Level in East Sussex a sea level rise of 6mm a year is chosen, 1 mm from geological sinking and 5mm from global warming. When one remembers the number of forcing agencies, both positive and negative, that are recognised by the I.P.C.C. panels, yet unused because of inadequate data, the uncertainties in the latter figure are immense. They are compounded by economic and political pressures on choice. But an estimate must be made somewhere or nothing would be done, except that error in terms of the nuclear power stations would lead to horrendous and probably irreparable human and environmental damage.
The statistical 'black box' used to generate the probabilities on which estimates are made, in itself has potential flaws dependent upon human choice. The statistical process, I am given to understand is of recent development. It involves choosing a mathematical model that best fits the spread and shape of data available. There is disagreement about which of the models is best for the job, although statistician-peer-group opinion elects for one.. Additionally, forecasts of extreme weather events, generated by the 'black box', depend on information the collection of which was started some fifty to sixty years ago. The data is not plentiful.lt is coupled with estimates of the effects of global warming. From this information estimates are made of events approaching the catastrophic. Thus for the immediate defences of Dungeness preparations have been made for events, with wave magnitude and frequency unspecified, only likely to occur once in ten thousand years.
Even given that the statistical models are sound, and the elements of human choice are accurate is it tenable to expect that what has happened in the last fifty years is applicable to the next ten thousand? In illustration satellite evidence of apparently randomly produced giant waves that can sink, without record, the largest cargo vessels, and 500 metre high tidal waves produced by large landslips, has only very recently come to light. This shows that the frequency, and magnitude of tidal waves is at present a matter of guesswork.
To design the sea defences required, forecasts of tide periodicity and levels, and a number of meteorological inputs must be made. Tide periodicity is immutable, the remainder show variability and uncertainty, yet all must be included. To do so spatial statistical distributions of factors applicable to a site are integrated. The mathematics are beyond the writer, but from them are determined probabilities of repeat or return times for storms of specific values of wind velocity, wave characteristics, and the added height of sea levels created by storm surges and global warming.
These mathematical models are now available commercially. Writing of insurance risks of extreme weather events, the Munich Re Geoscience Research Group summarise the situation (3). "The models that are available on the market provide quite plausible results when seen in isolation, but these results may vary considerably from one model to the next"...... "What is the reason for the differences in results? In all cases, the basis of calculations is the data that has been input, the quality and plausibility of which must be examined very critically and verified. (My emphasis) The mathematical, statistical part is the real heart of the model and is therefore set up as a kind of "black box" in many commercial models. Its contents are not accessed by the user or only in part, which makes it that much more difficult to interpret the results."
A thought experiment lets us understand the procedures. Imagine a cloth bag containing hundreds of small balls, identical in every way except colour, some of which are red and the remainder black. There are too many to count and there is no surface flat enough nor large enough on which to place them. The question is what is the proportion of red to black balls? Is there a way of establishing the proportion with a reasonable degree of certainty?
Well sampling can tell us. Let us obtain a sample of ten balls drawn without looking into the bag. Suppose it shows five of each colour. The proportion may be 50/50 but we need more samples to be more certain. So, put the sample back because we want the proportions to remain constant. Shake up the sack thoroughly, otherwise we may pick that sample again.
Now repeat this process twenty times and record the number of red balls in each sample. They range between 2 and 8 with most results being between 4 and 6, five being most frequent. It looks as if our first guess of 50/50 is right. A statistician, brought in at this stage, would be able to work out the odds on us being right, if he assumes that only two colours are involved and in all other respects the balls are the same, and factors will not change in the future.
To illustrate that last point suppose the number of black balls is doubled and we are not told. Then the proportion of red to black changes to one to two (1/2 or 50/100). In future samples there are likely to be more black balls than red. The numbers will not be so neatly balanced as in the first experiment. The statistician says the distribution shape has changed.
Similar processes are involved in estimating the frequency of extreme storms. The coastal defence engineer assumes that there is a sufficient number of separate events (samples) recorded, that future environmental conditions can be estimated and he can be reasonably confident of the shape of the distribution. His statistical calculations may well be supportive but in the first place he is forced to make assumptions that may be wrong.
The data he can use is limited to a few decades but he is forecasting for thousands of years. As a comparison the climate scientists use evidence covering tens of thousands of years to make forecasts of global warming for the next hundred years. Even then they offer optimistic, middling and pessimistic forecasts.
He assumes that conditions will not dramatically change for the period of his forecasts yet the U.N. Intergovernmental Panel for Climate Change (IPCC) reports warn us that there can be pleasant and unpleasant surprises, and that there are sources of change they do not include, because they have no measurements for them. Additionally the Third Assessment Report of the IPCC shows that average global temperatures are likely to rise faster than anticipated in their previous report. It would seem that temperature and carbon dioxide rise in the future could follow ever steepening curves the slope of which is at present indeterminable. Extreme weather events are associated with these rises, so they are not positively relatable to the last fifty years.
There is one other point.. The forecasts underlying designs for sea wall strengthening are based on the anticipated frequency of extreme events. Dungeness defences are built to withstand an event that is forecast to happen only once in ten thousand years, something having the dimensions of a tidal wave. Fine except that the maintained gravel barriers conceivably could not be strong enough, and the once in ten thousand years could be tomorrow.
In assessing future trends there is a great probability of error. For Dungeness Power Stations we should be acting to mitigate the effects of misjudgement, by immediately removing as much radioactive material as possible from the site, and heavily encasing the remainder in concrete to prevent contact with our environment.
(1)1PCC Report. 2001 Summary for Policymakers Group II pg11 See Reference List No 14
(2) Jan 3rd 2002 Munich Re web site www.munichre.com (An English Version of their current report, "Winter Storms in Europe fll) is to be published at the end of January See Reference List No 15
(3) topics 2000 natural catastrophes - the current position Munich Re Group
Dec. 1999 Pg 116 See Reference List No 15