Документ взят из кэша поисковой машины. Адрес оригинального документа : http://star.arm.ac.uk/~meb/riskpap/riskpap03.doc Дата изменения: Fri Oct 22 14:29:40 2010 Дата индексирования: Tue Oct 2 03:41:13 2012 Кодировка: Поисковые слова: prominence

Risk and Natural Catastrophes: The Long View
Mark Bailey
Armagh Observatory

Summary

Natural catastrophes - rare, high-consequence events - present us with a
unique conjunction of problems so far as risk is concerned. First, they can
have an extremely long recurrence interval -, so long that the greatest may
not have occurred within living human memory. Secondly, the effects of
events with which we are all too familiar, for example earthquakes, floods,
volcanoes and storms, are easily trumped by the impacts of objects - comets
and asteroids - that reach Earth from outer space; and thirdly, the largest
of these events have a global reach, in principle threatening not just our
way of life but perhaps life on Earth itself. However, recognizing that
such events occur very rarely, should we 'make hay while the sun shines'
and ignore, ostrich-like, the significant actuarial risk; or should we seek
to understand the underlying phenomena and develop strategies to mitigate
the threat and perhaps technologies to avert it? Our individual response
often depends less on a purely rational assessment than on personal
circumstances and how we have been brought up, but in any case, the nature
of the risks, which are poorly understood, means that we must be prepared
to handle the law of unintended consequences (i.e. could our actions make
things worse?). We must also be prepared to explore what happens if,
perhaps inevitably, our current scientific understanding turns out to be
less certain than many experts believe.





Introduction

Rare, high-consequence events present society with exceptional difficulties
so far as risk assessment is concerned. The infrequency of the most
extreme events means that their causes are often among the most poorly
understood among environmental issues and their impacts are - fortunately -
poorly known from direct experience. In addition, natural catastrophes
may have an origin either entirely within the Earth-system or from outside:
from the Sun, solar system or wider Universe of which the Earth is a part.
It is widely accepted that the potentially devastating effects of cosmic
phenomena are likely, in the long term, to far outweigh any purely Earth-
based cataclysm.

In this paper, we first describe the canonical framework for risk
assessment in the context of natural catastrophes and major societal
events. There follows a broadening of the discussion to include the
implications of natural disasters ultimately triggered by events occurring
outside the Earth, in our solar system. These phenomena, coming at us from
a source in outer space, exceed in terms of magnitude the worst possible
hurricane or earthquake, and have the potential to cause very significant
loss of life - even mass-extinctions of species - and to change the course
of evolution of life on our planet. The final part of the paper presents a
more speculative aspect of this story, highlighting the fact that there are
more uncertainties, even in astronomy, than many experts are generally
prepared to acknowledge. This raises the uncomfortable question of the
limits to knowledge and how these systematic unknowns can be best planned
for.

The "Long View" of astronomy thus raises many questions for risk
assessment, for how we perceive ourselves, and for our views on the
development of society. For example, do we ignore rare, high-consequence
hazards that have an extremely low probability of occurrence in our
lifetime or even those of our children and grandchildren, palming as it
were the responsibility for managing the risk on to successive generations?
Similarly, how much should we in the developed world do in order to
quantify and mitigate potential global hazards, when we have relatively few
natural resources at our disposal and when our population is only a very
small fraction of that of the whole world. In short, who should take care
of global hazards: are they our responsibility because we recognize them,
or does the ultimate responsibility fall to others, perhaps those more
wealthy than ourselves or more at risk from the identified threats?

In dealing with these questions it is sometimes helpful to consider human
life and civilization as if it were an organism with a lifetime measured in
thousands or perhaps millions of years. Would such a creature, with a
memory of natural catastrophes occurring over long time-scales and with a
life expectancy up to millions of years, respond differently to long-term
risks than we currently do, whether as individuals, national governments or
international organisations?

Finally, having identified a risk, for example - as in astronomy - by
curiosity driven research, how should we allocate the necessary resources
to achieve the goal of fully understanding the risk and putting in place
appropriate mitigation measures. And if we choose to follow this path, how
do we ensure that risks implicit in the development of such countermeasures
are themselves controlled?


Natural catastrophes

First let us consider the term 'natural catastrophe'. This is usually
taken to mean a game-changer: a sudden, often unpredictable event outside
human control that totally changes the circumstances or environment in
which we live. Natural catastrophes are usually accompanied by significant
loss of life, although the loss can often be mitigated if appropriate
warning and/or countermeasures are put in place. Of course, the phrase
'appropriate warning' implies a deeper knowledge of the phenomenon or of
the factors that led to the critical event occurring.


**** FIGURE 1 HERE: North Sea flood of 31 January 1953 ***


Most of these events are short-lived, that is they occur on time-scales of
minutes, hours or at most days, and are (at least currently) beyond human
control and any purely technical fix. An obvious solution is not to live
in places where such events may occur - or alternatively to learn to live
with the risk (whatever that means) and make the most of the benefits that
may accrue, for example the availability of fertile land near volcanoes or
flood plains, spectacular scenery and so on. The acceptance of risk, and
the almost fatalistic acceptance by some people to 'take the hit' if and
when it occurs, perhaps on the assumption that the "Big One" (earthquake,
flood, volcanic eruption etc.) won't occur in their lifetimes, is a common
theme in our all too human response to rare, high-consequence events over
which we, as individuals, have little or no direct experience or control.

A further point is that taking the world as a whole ordinary or mundane
natural disasters occur with a sufficient frequency as to permit a level of
familiarity with the phenomena and the use of statistics to assess their
impacts. Many natural catastrophes are very localised, for example
confined to certain coastlines (Figure 1) or areas of known geological
activity; and unless one happens to live in such a region can often be
regarded with equanimity. Although we are familiar with many kinds of
natural catastrophe, most of us, certainly those who live in the UK, can
afford a sense of detachment, noting that when disasters occur they are
likely only to affect those unlucky enough to be caught in the action.

In summary, natural catastrophes are sudden, uncontrollable, high-impact
events, often occurring with little or no warning. They are mostly local in
scale, and very short-lived in duration compared with the mean time
intervals between them. Their global frequency means that they are
amenable to scientific analysis and direct observation, and in this way we
have opportunities to study the risks, receive warnings and anticipate
their occurrence sufficiently far in advance to remove populations away
from them if required.


*** FIGURE 2 HERE: Climate change. ***


There is of course another class of natural catastrophe, which I label
'unnatural' natural catastrophes. Here the signature of immediacy is
often missing; rather, the catastrophe - for it is no less a disaster for
those caught up in it - is the result of a slow acting process, and the
ultimate cause may be mankind's interaction with the environment. The most
obvious contemporary example is climate change (Figure 2), a phenomenon
that affects the whole globe (and may also have a significant
extraterrestrial vector). In this case the risk affects every creature and
organism on Earth, but because we have no recent experience of such an
event the risk is particularly difficult to quantify.

Unnatural natural catastrophes also elicit reactions such as denial ('maybe
it will not happen') or fatalism ('there is nothing we can do about it').
Again, we must rely on science to understand the risk, but there is always
much greater scientific uncertainty: partly because we have no recent
experience to calibrate our theories, and partly also because we are forced
by circumstance to extrapolate from rather limited knowledge and
understanding.

Here, perhaps history can be a useful guide. Archaeology and written
records can provide us with indirect evidence of events that took place
thousands of years ago. Deciphering the clues and aligning them with
projections from current theories may help significantly to advance
scientific knowledge so far as rare natural catastrophes are concerned.


Risk

Next, consider the first element of the title of this article, namely
"Risk". It is a word that has many different meanings in common parlance,
a fact that has had the rather unfortunate result that when people use the
word "risk" we, and they, don't necessarily know what they are talking
about! One has only to look at two dictionary definitions of the word to
see the confusion that exists. For example, whereas the Chambers Twentieth
Century Dictionary gets it more or less right, i.e. "hazard, damage, chance
of loss or injury; degree of probability of loss ."; the Compact Oxford
English Dictionary merely states that risk is "the possibility that
something unpleasant will happen .". The UK Treasury, which also takes a
keen interest in risk, defines it in the following way: "Uncertainty in
outcome, whether positive opportunity or negative threat, of actions and
events .".

Whatever the precise definition, the key is to recognize and clearly
separate the two elements that together make up our concept of "Risk",
namely the event component and the frequency of the event. The closest,
and in my view perhaps the best analogy is to adopt an insurance
perspective. Here, the precise nature of the event for which we seek
insurance is specified in the small print of the insurance policy; and the
insurance company would know or be able to estimate from its claims'
history how frequently such events occur and what their cost was likely to
be. Thus, we obtain an actuarial cost for the identified risk, which with
a bit of profit for the company, leads to the annual premium that you or I
would pay.

This provides a relatively dispassionate way to assess risk, one that in
the ideal case allows some events to be dismissed as lying below a 'de
minimis' threshold and others to be given correspondingly closer scrutiny.
A monetary proxy for risk allows different kinds of risk to be ranked and a
decision taken whether they are significant enough for us to be concerned.
In other words, it enables us to see the wood for the trees.

Risk assessments are notoriously subjective; and partly no doubt to provide
the illusion of mathematical precision estimates are usually presented in
the form of a 'risk matrix', with rows (or sometimes columns) representing
the perceived frequency of a given event, and columns (or sometimes rows)
representing the perceived impact of that event occurring. Businesses and
governments approve of this approach, which is illustrated in Figure 3 by
the risk matrix that the Armagh Observatory is obliged to use by
government.


*** FIGURE 3 HERE: Armagh Observatory risk matrix. ***


The rubric associated with Figure 3 indicates that the highest 'impact',
with a numerical value of five, might be caused by failure of key
Observatory or Departmental objectives, or lead to financial loss exceeding
several million pounds, or to significant public embarrassment to the
Department and/or National media coverage (though sometimes we seek that),
or attention from the Assembly or Public Accounts Committee . or even
Death. On the frequency axis, low-medium (i.e. a numerical value of two)
means that the event might conceivably occur at some time, which actually
means that it will probably occur once or twice; and in that sense the
product of the two risk factors, in this case 5 x 2 = 10, leads to a 'red'
or 'high' risk. Such a risk should always be mitigated, if possible.

In the context of natural catastrophes, which by their nature will almost
certainly lead to "death" and must "conceivably occur at some time", these
definitions are bound to place natural catastrophes in the red-risk 'high'
category, indicating that resources should always be provided to reduce the
risk to an acceptable level. However, while governments encourage us to
plan for the worst in the cases of pandemic influenza, storm or flood, much
less frequent, but potentially very much larger natural catastrophes, don't
figure on our risk register. By some sleight of hand, they are deemed not
'our' responsibility again raising the question, which of us is to look
after such risks?

The risks that affect many people simultaneously are called "Societal
Risks", and owing to the large range both in the frequency of occurrence of
different events and their respective impacts are usually presented in
graphical form. Developing one of the points made earlier, namely that
natural catastrophes are mostly fairly localized and occur rather suddenly
and without warning, it is fortunate that we live in a world where the
greater the number of people affected, the longer the recurrence time
between events.


*** FIGURE 4 HERE Location on the Frequency-Impact plane of various high-
consequence risks facing the United Kingdom. ***


Figure 4 provides a graphical representation of risk, in which the relative
impact and the relative likelihood of different kinds of event combine to
indicate that electronic attacks, terrorist attacks on crowded places, and
pandemic flu are all quite likely - i.e. will occur within a human lifetime
- whereas major industrial accidents are far less likely but may have a
similarly high impact. In this way, we begin to see how risks can be
ranked according to the objectively assessed criteria of frequency and
impact.


*** FIGURE 5 HERE: F-N criteria for societal risk. ***


Figure 5, taken from the Health and Safety Executive's "Tolerability of
Risk" report (1992), presents what has now become a very common way to
assess the impact of very rare, high-consequence events. This provides
another way of looking at risk from an insurance perspective, but with the
unit of currency a human life. If, for example, everyone could agree on
the average value of a life (too often this is narrowly interpreted as the
'economic value' of a life), then together with the frequency of a given
risk - or to put it another way, the frequency of events that might lead to
a large number of fatalities - we could determine how much money we should
be prepared to pay to avoid, if possible, such an event. For example, if
a particular incident were to lead to 100 deaths every thousand years, on
average, and if this were regarded as the limit of intolerability (that is,
you could barely live with this outcome), then if the value of a human life
were ё1.5 million you should be prepared to install countermeasures costing
upwards of around ё150,000 a year in order to avoid that particular risk.







This approach has advantages, but raises many questions: not least what is
the value of a human life, and one's (or society's) risk appetite. For
example, if the events in question are very rare indeed, many decision
makers will take a pragmatic approach and 'gamble' that the risk will not
crystallise on their watch, though such an approach if continued will
likely end in disaster. Similarly, an individual's assessment of the
value of their own life or that of a family member is likely to be much
greater than ё1.5 million! There are many ethical issues, for example to
what extent can the value of a human life be measured in purely monetary
terms and be assumed a universal constant. Does it depend on the lives in
question; their income or employment; the good (or bad) that they do; their
place of birth, nationality, and so on; and how do you apply the actuarial
approach to situations of high risk involving large numbers of possibly
simultaneous fatalities?

Leaving these complications aside, the idea of a Frequency-Number or 'F-N'
diagram, showing the number of expected deaths versus the frequency of
occurrence of a given event, provides a useful way of ranking different
kinds of natural catastrophe. Risks can be deemed intolerable if the
frequency of N deaths is too high, and tolerable, in other words below the
negligible line, if the average number of lives at stake per year is close
to zero. Between the lines of Tolerable and Intolerable you seek to reduce
the risk to an acceptable level "as low as reasonably practicable" (ALARP).
And that's what we do in our daily lives so far as risk is concerned, as
also do governments in respect of their regulatory regimes on behalf of
society. Clearly, there are large uncertainties, but despite
shortcomings, an actuarial approach has the merit of allowing qualitatively
different types of risk to be assigned an objective cost that can be more
easily ranked and discussed.

Astronomy

We now leave the thin ice of ethics and move onto the apparently more
secure ground of astronomy. First, our violent Universe potentially trumps
any purely Earth-based cataclysm that you or I might care to name. One way
of illustrating this is through the death of the dinosaurs (Figure 6),
which occurred around 65 million years ago. The impact of a comet or an
asteroid on the Earth wiped out large numbers of species, changing the
course of evolution of life on Earth forever. In particular, it provided an
opportunity for mammals and eventually humans to appear - so from a purely
selfish point of view the impact was clearly a 'Good Thing'!


*** FIGURE 6 HERE: Death of the dinosaurs ***


Nowadays everybody is aware that the Earth - like the Moon - is a bombarded
planet. For every crater you see on the Moon, the Earth will have been
struck approximately twenty times as often owing to the relative size of
the two objects. The frequency of large impact craters on Earth is
illustrated in Figure 7. Some areas of the globe, such as some parts of
Africa, Australia and North America, are very, very old and still show
evidence of impacts that have occurred over geologic time. There are nearly
200 recognized large impact craters on the Earth, a figure consistent with
the Earth being hit by a 'Big One', i.e. a 'dinosaur-killer' capable of
causing a global mass-extinction of species, roughly once every 100 million
years or so.


***** FIGURE 7 HERE: Bombarded Earth.


In the distant past, soon after the Earth was formed, the rate of large-
body impacts was much higher, and as a consequence the newly formed Earth
was sterilized by the heat liberated by frequent impacts. Even today, 4.5
billion years after the formation of the Earth, occasional large impacts
still occur with the capacity to produce worldwide devastation. How are
we to consider such a natural catastrophe? In terms of Actuarial Risk,
such an event would formally have an infinite cost (i.e. it would be
intolerable in every sense of the word), and so - even though it is a very
rare event - the risk would be mathematically unbounded and we should do
everything possible to avoid its occurrence.


*** FIGURE 8 HERE: Spacecraft image of a comet nucleus and an asteroid
(from Slide 12) ***


The objects that produce these kinds of impacts are known to astronomers as
Near-Earth Objects (NEOs), owing to the fact that their orbits come close
to or cross that of the Earth. They are shown in Figure 8. Briefly, NEOs
are any astronomically "small" body that can pass close to Earth. The size
range of these observed objects is very large - from a few tens of metres
up to tens of kilometres or more.

NEOs constitute a very diverse population, including both comets and
asteroids, and fragments thereof; 'dead' or devolatilized (or temporarily
inactive) comets; and occasional, but very rare 'giant' comets (i.e. comets
with diameters larger than about 100 km). At the lower end of the size
range the population merges into that of meteoroids and interplanetary
dust, the total mass influx of these small particles on Earth being about 1
kg per second, or some tens of thousands of tonnes of material per year.

Where do they come from? The asteroids by and large come from the main
asteroid belt, lying between Mars and Jupiter in the solar system. The
dynamical process by which this occurs is rather complicated and has only
recently begun to be elucidated, but it usually begins with a catastrophic
collision between two asteroids in the main belt. This is a random and
unpredictable event and the resulting fragments move onto new, possibly
unstable orbits, some of which may become planet crossing, eventually Earth-
crossing and possibly Earth-colliding. It is evident that this process,
which has the potential to change the course of evolution of life on Earth,
introduces a high degree of contingency in how life on any such bombarded
planet might evolve.

The comets have a variety of sources. The main one is the Oort cloud, a
vast swarm of dust-and-ice comet nuclei extending halfway to the nearest
star and yet still part of the solar system. An important secondary source
is the Edgeworth-Kuiper belt, a thousand times closer, in the region
immediately beyond Neptune. Comets too are subject to the vagaries of
occasional collisions as well as the effects of loss of volatile material
from the central nucleus. Their orbits are more uncertain in the long term
owing to perturbations by occasional 'random' close approaches of distant
stars and molecular clouds to our solar system, and by the systematic
gravitational perturbations of the Galaxy as a whole.

If one asks how many of these objects there are and why the comet and
asteroid impact hazard has recently comet to prominence, the answer is that
our knowledge has increased dramatically within the last twenty years or
so. In this time there has been a very rapid increase in the number of
known, i.e. discovered, Earth-crossing asteroids: from approximately 135 in
1990 to around 7000 at the time of writing. Among these NEOs the number
with average diameters larger than about a kilometre is thought to be of
the order of 1,000. There is of course some uncertainty in that number,
but it is now widely accepted that it is known to rather better than a
factor of two. This together with knowledge of the NEO orbits enables
astronomers to estimate the average frequency of collision of such bodies
with the Earth; the result is that one such impact occurs roughly every
200,000 years on average. In addition to these kilometre-size objects,
there is a further more poorly understood population of comets and extinct
comet nuclei and a very large number of smaller bodies, the total number
increasing roughly as the inverse square of the diameter down to fifty or a
hundred metres or so.

What effects would one of these objects produce were it to impact the
Earth? The size of the astronomically smallest objects of any real
significance so far as impacts are concerned is in the range 30-100 metres,
and the size of the largest observed NEOs extends up to that of a
terrestrial mountain (i.e. up to several tens of kilometres). At the
lower end of this size range the kinetic energies of impact are of the
order of 10-100 megatons, and if they reach the ground (which not all of
them will) the objects would make craters with sizes up to a kilometre
across. A 'rule of thumb' is that the size of the crater is approximately
10-20 times the size of the object. Other things (e.g. mass) being equal,
comets are more dangerous than asteroids because on average they have much
higher impact velocities than asteroids. On the other hand, comets are
widely believed to be less dense than asteroids, and so size-for-size the
effects of the two classes of object are thought to be similar.

For object diameters much larger than 30-100 metres most will reach the
ground, producing craters bigger than a few kilometres across. Sub-
kilometre-size objects would destroy areas the size of a large city or a
small state or province. Oceanic impacts would produce tsunamis, and as
one moves up in size the kinetic energies of impact become almost
unimaginable. The size of craters resulting from the impact of kilometre-
size bodies and larger range upwards from around 20 kilometres, and the
implied giant tsunamis reach ocean scales.

The frequency of comet or asteroid impacts with diameters greater than a
critical value of the order of 0.5 to 2 kilometres is a key parameter in
the evaluation of the risk. For sizes larger than this, the impact will
have global consequences wherever it hits. Smaller objects cause local
devastation, but - just as with 'ordinary' natural catastrophes - the parts
of the world not immediately affected will probably not be affected at all.
The largest known NEOs, with sizes ranging up to ten or more kilometres
would produce environmental effects of such a magnitude that they could
produce a mass extinction of life on Earth.

It is fortunate that the largest objects are relatively few in number and
the most infrequent among impactors; it is fortunate too that they are also
the easiest objects to discover in space. This suggests that the
evolution of life on Earth has benefitted from a favourable conjunction of
events, in particular the time since humans have evolved is only a few
million years, short compared with the average time interval between really
large impacts. During this time (in fact within the last hundred years)
we have developed the knowledge and technical capacity to discover
essentially all the most dangerous objects in near-Earth space.

We humans are therefore living at a very special time in Earth history: for
the first time in the history of life on Earth, a species has developed
that has the scientific and technical knowledge broadly to understand its
place in space and the wider Universe; and at the same time has the
technological capacity in principle to do something about the risk of rare,
massive impacts. However, the issue is not of particularly urgent
concern: we do not know of any large comet or asteroid that is currently on
an orbit destined to collide with the Earth within the next hundred years
or so. Moreover, if the mean interval between really big impacts is tens
or perhaps hundreds of millions of years we would have to be very unlucky
indeed to be living at such a time of crisis.


***** FIGURE 9 HERE: Impact of Comet Shoemaker-Levy 9 on Jupiter. ***


Of course, our understanding of the environmental effects of more frequent
but still massive impacts is still at a very rudimentary stage, and we have
- fortunately - not been able to test our theories in this regime.
Impacts of kilometre-size bodies on planets do occur, however, and a few
years ago we had the opportunity to witness a sequence of such impacts on
Jupiter, namely the impact of fragments of the tidally disrupted Comet
Shoemaker-Levy 9 in July 1994. Figure 9 illustrates the Earth-scale
atmospheric effects produced by these events, and it is partly as a result
of this experience that some commentators now consider the 'critical' size
of an impactor necessary to produce a devastating global effect on Earth so
far as the survival of civilization is concerned (for example through a
rapid impact-induced climate change) as lying closer to 0.5 km than 2.0 km.



Another well-known example is the case of the 20th century Tunguska event,
which in June 1908 flattened and destroyed a forested area roughly equal to
the area of Greater London. It had a blast equivalent of approximately 3-
10 megatons, and had it landed over London, it would clearly have destroyed
that whole area and much of the surrounding district. Leaving aside the
potential loss of life, the economic consequences of such an impact,
negligible in global terms, would influence the affected nation's
prosperity for years.


The left panel of Figure 10 shows another instance, this time the impact of
a rather smaller object, in the large-meteorite class, which encountered
Earth in 1947. It is a painting from the Russian Academy of Sciences
showing the Sikhote-Alin meteorite. At right is the totem pole erected
tongue in cheek by Russian scientists close to the Tunguska 'ground zero'.
The totem pole represents the Siberian fire god 'Agby' - he who brings fire
to the forest - and there is already a thriving modern mythology that if
one does not leave a trinket or personal possession at Agby's feet, one
will never return to this 'sacred' site. It is amusing to note that many
western scientists travel thousands of kilometres, as if in pilgrimage, to
the area in order to inspect the site and search for traces of the
impactor.


*** FIGURE 10 HERE: Sikhote-Alin meteorite and Tunguska totem pole. ***


Cost of impacts

With these examples in mind, let us now estimate the actuarial risk posed
by NEOs. To do this, we assume there to be roughly 1500 objects larger
than 1km in diameter, each capable of causing a global catastrophe, for
example rapid climate change and a subsequent devastating effect on global
civilization. The mean impact probability is of the order of one in
several hundred million per year per object. Thus, the average rate of
impacts is roughly one event every 200,000 years (for objects larger than 1
km in diameter); and if a quarter of the UK's population of 60 million is
assumed to die as a result of such an impact and each life is valued at
ё1.5 million, the actuarial risk - or cost - of NEO impacts to the UK alone
becomes approximately ё100 million per year. Note that a proportionate
cost is incurred by other nations, and the cost is incurred wherever the
impactor happens to land: the effects of impacts by objects larger than a
kilometre in diameter are global.

This raises a number of questions so far as risk and natural catastrophes
are concerned. Clearly the annual cost (less than the cost of a large
passenger jet) is affordable, and there are also many things we can begin
to do to address the problem (e.g. discover the potentially hazardous
asteroids in advance of any impact, predict where and when the next impact
will occur and/or relocate people who happen to live close to 'ground zero'
and so on). There is even the possibility of deflecting the asteroid in
space so that it misses the Earth entirely, although this raises the
question of who controls such asteroid deflection technology and whether
the same technology could perhaps be used for clandestine offensive, rather
than defensive purposes.

Another point is that the cost of this type of natural catastrophe is, in
fact, much higher than has already been calculated, i.e. the risk is
actually greater than estimated. This is because the largest impacts may
lead to extinction of the human race, and in this sense the risk is
unbounded. Placing the calculated NEO Risk on the 'F-N' diagram shown in
Figure 5 indicates that asteroids easily fall in the 'Intolerable' part of
the figure. Moreover, because the risk is larger than that, reason tells
us we should develop an action plan to reduce or eliminate it. By the same
token, reason also tells us that that the initial cost of developing such a
plan could range up to several ё100 million per year, for the UK alone, and
still provide good value for money.

Let me emphasize that this is not an argument for funding astronomy(!), but
it demonstrates the value of funding 'pure' research, not really knowing
what future benefits it will bring. In this case, astronomers
serendipitously discovered a significant NEO impact hazard to civilization
having an actuarial cost to the UK alone of the order of ё100 million per
year. What we do with the information is another question, and one that,
like most extreme cases, goes to the heart of the problem. An issue that
might be considered, for example, is the question of our risk appetite or
(some might say) our risk indigestion. Note that the chance of a
kilometre-size impact on Earth within our lifetimes is extremely remote,
far less than the probability of risks that we routinely accept as
individuals on a daily basis (for example, a person in their mid-fifties
has a daily risk of dying of around one in 200,000). A pragmatic response
- perhaps the reaction of most people - would be to ignore the lowest
probability high-consequence risks as lying below one's individual level of
concern.

If this turns out to be humanity's considered reaction to the problem,
perhaps the lesson is that - as with many modern scientific discoveries -
our 'instinct' is letting us down, or to put it another way: our 'gut' has
not kept pace with our brains. We are culturally attuned to ignoring low-
probability hazards that might significantly affect us as individuals,
families or small groups, often preferring simply to get on with our lives
rather than making a possibly uncomfortable change to our lifestyle to
avoid the risk. Examples such as smoking, crossing the road, driving whilst
using a mobile phone, flying, and so on immediately come to mind. With
these ingrained habits there is a risk that we adopt the same response to
low-probability high-consequence societal risks, even when the actuarial
risk is high and when a particular incident could affect millions or
perhaps billions of people at the same time.

To conclude this section, let me briefly emphasize a further important
point about the NEO impact hazard, namely its singular nature. As well as
producing a potentially unbounded risk, impacts are highly predictable,
often years or decades in advance, provided we have sufficient knowledge of
the NEO ensemble; in short, NEO impacts are avoidable given enough warning.
One could, for example, remove the affected population from 'ground zero';
stockpile food supplies for the years when food would be scarce; or even
engage in a 'star wars' deflection of the NEO in space, so that it never
hits the Earth. Considering that the technology exists to rendezvous with
comets and asteroids in space, and even to fire objects into them, it is
clear that most of this risk - potentially the most serious risk we have
yet faced as a species - could be mitigated.

Broader astronomical context

Rather than dwell on the effects of comet or asteroid impacts, or the
possible effects on Earth of the accretion of interplanetary dust produced
by the evolution of comets or by collisions of asteroids in the main belt,
let me finally describe some of the continuing uncertainties underlying our
present understanding of the astronomical context in which NEO impacts
occur.

It is widely recognized that astronomy is currently experiencing a "Golden
Age", in which the separate strands of human endeavour motivating interest
in the subject, for example the 'cosmological or quasi-religious' strand,
the 'astrophysics' strand and the 'practical or spin-off' strand, have come
together to produce incredibly rapid progress and great leaps in
understanding. Astronomy plays an important cultural role as an
imagination driver, not just in science but also in the humanities, art and
history, stimulating the work worldwide of artists, poets, musicians and
philosophers.

In any subject that is advancing so fast, there are bound to be errors and
uncertainties. In fact, it is part of an astronomer's job to dismiss old
theories and replace them with new, hopefully better understanding, and
there is no reason to think this has finished! Up to now I have described
what might be called the 'standard' model of the NEO impact hazard; but if
we are intent on assessing the hazard in the long term, which involves time-
scales of hundreds or thousands of years, or more, we really must take a
more strategic view. This reflects the point famously made by Donald
Rumsfeld about the 'known knowns', the 'known unknowns', and the unknowns
that we do not even know exist. How do we fold such systematic scientific
uncertainties into a correct assessment of the long-term risk posed by
natural catastrophes?

Let me illustrate the point by describing some recent advances in solar-
system astronomy that may yet prove to be an important element in assessing
the extraterrestrial impact hazard. Astronomers have discovered that giant
comets exist as part of a very broad size distribution of planet-crossing
objects extending to dimensions much greater than 100 kilometres in
diameter. The orbits of these objects, their number in the inner solar
system and their detailed physical characteristics all vary on time-scales
of thousands of years, i.e. on time-scales of human concern. Such objects
also contribute to the fluctuating population of NEOs, which themselves are
only a relatively recently discovered population of Earth-interacting
bodies. And comets have the peculiar characteristic that their evolution
and decay leads to the production of narrow trails of debris, and therefore
to non-random impacts on Earth, evidenced by the familiar annual meteor
showers, the Shoemaker-Levy 9 event on Jupiter, and the series of almost
daily small-comet impacts on the Sun, originating from the hierarchical
break-up of a large Sun-grazing progenitor thousands of years ago.

Cometary evolution is highly chaotic and therefore unpredictable in the
long term. Moreover, as evidenced by the Shoemaker-Levy 9 event and
examples where comets have split into two or more pieces, comets are very
fragile, easily broken up in space and with short physical lifetimes of
perhaps only a few thousand years in short-period Earth-crossing orbits.
Thus, the comets we see are very different from those that our ancestors
would have seen 4,000 or 5,000 years ago.

An example that illustrates this unpredictability is an object called
Chiron, which is actually one of a class of newly discovered solar-system
objects called Centaurs. Its present orbit has a perihelion distance (the
closest point to the Sun) slightly inside the orbit of Saturn, and an
orbital period of around fifty years. If one models the evolution of a
large number of bodies with very similar orbits to the real Chiron, the
result illustrates the enormous uncertainty in predicting the evolution of
an individual object over very long time-scales. The orbits are chaotic,
which means that very small changes in the initial conditions of the orbit
or in the circumstances of the comet's orbital evolution, quickly lead to
gross differences in the objects' predicted future and past orbital
evolution. In the case of a Centaur such as Chiron, there is a chance
that it may have been a short-period Earth-crossing comet for some
thousands of years as recently as 75,000 years ago.

What would such an object have looked like? A Centaur such as Chiron is a
very massive object, with a diameter of the order of 200 kilometres, and
the amount of dust that such an object could deposit in the inner solar
system during a period of evolution as a short-period comet is huge. It
is possible that it might shed fragments of ordinary comet size as well as
dust, and that these might have contributed to a heightened space density
of solid objects in the inner solar system and to a temporary enhancement
in the rate of accretion of such bodies - and hence the impact hazard - on
Earth.

Such a scenario is simply not accounted for in the present 'standard'
picture of the risks posed by the current NEO population, but this is
precisely the sort of bigger picture that must be developed if we are to
obtain a full understanding of our place, and that of the Earth, in the
cosmos.

It is here that arguments from history and perhaps archaeology too may help
to inform astronomers and constrain our understanding of the types of
phenomena that humans may have witnessed, but of which we now have
essentially no knowledge. There are many puzzling 'mysteries' concerning
the earliest Greek descriptions of the cosmos, for example Anaximander's
view of 'stars'; Aristotle's description of the Milky Way as lying in the
sublunary zone and being an accumulation product of the disintegration of
many comets; and other authors' identification of the Milky Way as the
former path of the Sun. None of these views can readily be reconciled with
our understanding of the solar system unless the 'sky' is those days was
somehow different, perhaps more active in terms of cometary and meteoric
phenomena than it is now, leading to a possible 'confusion' between what
was once a much brighter zodiacal light and the present Milky Way. In this
case, modern astronomy would have a lot to teach historians as well.

Conclusions

Studies of natural catastrophes caused by extraterrestrial impacts show
that these phenomena constitute a unique risk. The extraterrestrial impact
hazard provides a conjunction of difficulties for conventional risk
analysis. We do not have any recent experience of such impacts, except
perhaps through historical records, and their potentially unbounded
consequences and global reach would be intolerable except that they
represent very low-probability events. The existence of such a threat
raises questions such as which one of us as nations has the responsibility
to mitigate the threat, and who controls the resulting knowledge and
technology.

The actuarial approach provides a rational way to rank risks of very
diverse types and character, and in principle can be applied to all risks
so long as one can agree the costs. As with any market, the perceived
costs will vary with fashion and time (perhaps that is a strength rather
than a weakness), facilitating a focus on risks that are objectively
assessed as most important. Our present understanding of the majority of
high-consequence, low-probability risks remains very uncertain, and options
to mitigate the majority of natural catastrophes are very limited.

The foreseeable extraterrestrial impact hazard, however, is unique in that
in most cases it should be possible to predict with precision the time and
location of the large-body impact, and it is likely that the largest such
impactors would be discoverable decades, if not centuries in advance of any
impact. The biggest impactors have implications for the survival of
civilization and the human race, and perhaps also for the future evolution
of life on Earth. If humanity were to develop technology to mitigate this
threat, then we - uniquely among all life forms that have ever existed on
the planet - would have largely inoculated ourselves against a major
external driver of evolutionary change.

The analysis also shows that as a result of undertaking curiosity-driven
research we live at a 'special time' in the history of life on Earth: we
recognize Earth's place in the Universe; we recognize that Earth is a
bombarded planet; and that Earth is an 'open' system in touch with its near-
space environment. This understanding represents a significant paradigm
shift, though one that is difficult to see because we are living through
it. We also recognize that controlling impacts holds the key to the long-
term survival of civilization, even to the long-term survival of life on
Earth; and a species - namely us - has the knowledge to compute and assess
the risk.

So, for the first time in the 3.8 billion-year history of life on Earth
these facts are broadly known. The historian Stephen Toulmin in his book
"The Return to Cosmology" captured this sense of our position when he
wrote: "Human beings are the beneficiaries of history . our fate within
this historical scheme depends . on the adaptiveness of our behaviour .
[and] on the use that we make of our intelligence in dealing with our place
in Nature." It will be interesting to see whether we rise to the
challenge of 'the long-view' and take the steps necessary to mitigate the
potentially unbounded risks to life on Earth posed by extraterrestrial
impacts.

Acknowledgements. I thank the organizers of the Darwin Lecture series for
the invitation to provide this lecture and for their help in producing this
manuscript. Astronomy at Armagh Observatory is funded by the Northern
Ireland Department of Culture, Arts and Leisure.


FURTHER READING

Articles and reports concerning risk in general:

1. Government Policy on the Management of Risk, House of Lords Select
Committee on Economic Affairs, 5th Report of Session 2005-06, HL Paper 183-
I, 2006. Available online at:
http://www.publications.parliament.uk/pa/ld200506/ldselect/ldeconaf/183/183i
.pdf

2. Safety in Numbers?, Parliamentary Office of Science and Technology
Report No. 81, 1996. Available online at:
http://www.parliament.uk/documents/post/pn081.pdf

3. The Tolerability of Risk from Nuclear Power Stations, Health and Safety
Executive, 1992. Available online at:
http://www.hse.gov.uk/nuclear/tolerability.pdf

4. National Risk Register of Civil Emergencies, Cabinet Office, 2010.
Available online at:
http://www.cabinetoffice.gov.uk/media/348986/nationalriskregister-2010.pdf


Articles and reports concerning risk and NEOs:

1. Near Earth Objects - NEOs, Parliamentary Office of Science and
Technology Report No. 126, 1999. Available online at:
http://www.parliament.uk/documents/post/pn126.pdf

2. Bailey, M.E., Clube, S.V.M., Hahn, G., Napier, W.M., Valsecchi, G.B.,
1994. Hazards Due to Comets: Climate and Short-Term Catastrophism,
in Hazards Due to Comets and Asteroids, Gehrels, T., loc. cit., 479-533.

3. Canavan, G.H., 1994. Cost and Benefit of Near-Earth Object Detection
and Interception, in Hazards Due to Comets and Asteroids, Gehrels, T., loc.
cit., 1157-1189.

4. Chapman, C.R., 2004. The Hazard of Near-Earth Asteroid Impacts on
Earth, Earth and Planetary Science Letters, 222, 1-15.

5. Chapman, C.R., Morrison D., 1994. Impacts on the Earth by Asteroids
and Comets: Assessing the Hazard, Nature, 367, 33-40.

6. Gehrels, T., 1994. Hazards Due to Comets and Asteroids, University of
Arizona Press, Tucson and London.

7. Remo, J.L., 1997. Near-Earth Objects: The United Nations International
Conference, Annals of the New York Academy of Sciences, New York, USA.


Books and articles describing natural catastrophes in a broader historical
context:

1. Bailey, M.E., Clube, S.V.M., Napier, W.M., 1990. The Origin of Comets,
Pergamon Press, Oxford.

2. Bailey, M.E., 1995. Recent Results in Cometary Astronomy: Implications
for the Ancient Sky, Vistas in Astronomy, 39, 647-671.

3. Greene, M.T., 1992. Natural Knowledge in Preclassical Antiquity, The
Johns Hopkins University Press, Baltimore and London.

4. Huggett, R., 1989. Cataclysms and Earth History: The Development of
Diluvialism, Clarendon Press, Oxford.

5. Nur, A., 2008. Apocalypse: Earthquakes, Archaeology, and the
Wrath of God, Princeton University Press, Princeton and Oxford.

6. Peiser, B.J., Palmer, T., Bailey, M.E., 1998. Natural Catastrophes
During Bronze Age Civilizations: Archaeological, Geological, Astronomical
and Cultural Perspectives, British Archaeological Reports, No. S728,
Archaeopress, Oxford.


FIGURE CAPTIONS

Figure 1: North Sea flood of 31 January 1953. The image shows the
devastation wrought by the flood at Oude-Tonge, on the island of Goeree-
Overflakkee in the South Netherlands. Image from Wikipedia Commons.

Figure 2: Climate change: an example of a slow acting `unnatural' natural
catastrophe. Illustrated by the record of mean annual temperatures
recorded at Armagh Observatory from 1796 to 2009. Image credit: Armagh
Observatory.

Figure 3: Armagh Observatory risk matrix. The darkest colour represents a
red 'high' risk; white represents a yellow 'medium' risk; and grey
represents a 'green' low risk. Image credit: Armagh Observatory.

Figure 4: Location on the Frequency-Impact plane of various high-
consequence
risks facing the United Kingdom. This is not the full range of possible
risks to the UK, but the location of each broad category of threat
nevertheless indicates the perceived risk that must be managed. Image
credit: Cabinet Office National Risk Register, ї Crown copyright 2008.

Figure 5: F-N criteria for societal risk. Image credit: Figure adapted
from Figure D1 of UK Health and Safety Executive (HSE) publication
'Tolerability of Risk' (1992). After Nigel Holloway (1997 Spaceguard
Meeting, Royal Greenwich Observatory, Cambridge).

Figure 6: Death of the dinosaurs. The painting by astronomer Don Davis
shows
pterosaurs at the instant of collision of a 10-km diameter asteroid with
the Earth 65 million years ago, a catastrophe that is widely believed to
have led to the mass extinction of species around this time, identified
with the Cretaceous-Tertiary ('K/T') geological boundary. Image credit:
Don Davis, NASA.

Figure 7: Bombarded Earth, showing sites of identified, highly probable and
probable impact craters as at 2010 May 16. Image credit: David Rajmon,
Impact Database 2010.1. Online at http://impacts.rajmon.cz.

Figure 8: Left: Composite image of the nucleus of Halley's comet taken
during the Comet Halley encounter of 1986 March 13-14. Halley's comet is
approximately 15 km long and 8 km wide. Image credit: H.U. Keller, Halley
Multicolour Camera, MPAe; ESA/Giotto. Right: Mosaic of the northern
hemisphere of the near-Earth asteroid Eros, taken by the NASA Near-Earth
Asteroid Rendezvous (NEAR-Shoemaker) spacecraft on 2000 February 29. Eros
is approximately 34 km long and 11 km wide. Image credit: Johns Hopkins
University, Applied Physics Laboratory, NASA/NEAR-Shoemaker.

Figure 9: Impact of Comet Shoemaker-Levy 9 on Jupiter. Comet D/1993 F2
(Shoemaker-Levy 9) broke into more than 20 fragments which collided with
Jupiter during the period 1994 July 16-22. The impacts produced long-
lived atmospheric 'scars' visible from Earth. This image of Jupiter with
the Hubble Space Telescope Planetary Camera shows five large impact sites
and three small ones, ranging in size from several hundred kilometres up to
Earth-size. Image credit: NASA/ESA Hubble Space Telescope.

Figure 10: Left: Fall of the Sikhote-Alin meteorite on 1947 February 12,
from the painting in the Russian Academy of Sciences. Image courtesy Yu. A.
Shukolyukov. Right: Totem pole erected close to the Tunguska 'ground
zero'. According to mythology, Agby is the Siberian 'god' who brings fire
to the forest.