Document Actions

ijege-13_bs-davies.pdf

background image
Italian Journal of Engineering Geology and Environment - Book Series (6) www.ijege.uniroma1.it © 2013 Sapienza Università
Editrice
229
DOI: 10.4408/IJEGE.2013-06.B-20
AVERTING PREDICTED LANDSLIDE CATASTROPHES:
WHAT CAN BE DONE?
T
im
DAVIES
University of Canterbury - Department of Geological Sciences - Private Bag 4800 - Christchurch, New Zealand
Email: tim.davies@canterbury.ac.nz
those at risk? Examples include the Indian Ocean
tsunami of 2004 (B
ilek
et alii, 2007), the lahar that
devastated Armero in 1987 (V
oighT
, 1990) and the
Huascaran disaster of 1970 (e
Vans
et alii, 2009); and
of course the Vajont disaster of 1963 (e.g. s
emenza
&
g
hiroTTi
, 2000). In each case, there was credible sci-
entific evidence that the catastrophe would occur, and
this evidence was provided to the civil authorities re-
sponsible for public safety; in each case no action was
taken prior to the occurrence of the predicted event,
with the result that thousands of people died. Herein
we attempt to answer the questions:
1. Would better science have been more credible to the
authorities and allowed them to take preventive
action (evacuation)?
2. Would better communication of science have pre-
vented the disaster? If so, how can science com-
munication be improved?
3. If the answers to these questions are negative, what
is the fundamental stumbling-block that prevents
science being effective in disaster management?
How can it be overcome?
We apply these questions to three current situ-
ations in New Zealand where science suggests that
coseismic landslides threaten the lives of hundreds
or thousands of people, yet there is little or no public
or governmental appetite for further investigation to
clarify the risks. In two cases mitigation by reloca-
tion is the only option, and evidently implies signifi-
cant costs, but could be affordably carried out over
ABSTRACT
Today landslide science is more complete than
at the time of the Vajont catastrophe, although some
way from being perfect. However, a factor is still
present that allowed the Vajont disaster, and a number
of other catastrophes since, to occur - organisational
and governmental reluctance to acknowledge the un-
acceptably high probability of a specific disaster and
take actions to avert it. At three tourism centres in
New Zealand there is substantive evidence that a large
landslide can be triggered by an earthquake; and that
each event can cause hundreds of millions of dollars
of damage and up to several hundred deaths (depend-
ing on the timing of the event). There can be no useful
warning of any of these events. Risk analyses suggest
that the risks are in all cases some orders of magni-
tude greater than acceptable societal levels. Further
investigation could clarify all these risks, but local and
national authorities are reluctant to pursue this. The
uptake of science by decision - and policy-makers is
constrained by short-term economic and political con-
siderations. The relevant science needs to published
and publicised in such a way that it can neither be mis-
understood nor ignored.
INTRODUCTION
Recent history contains many examples of natural
disasters that were foreseen and even predicted, and
subsequently occurred, with catastrophic consequenc-
es. The question arises, why was nothing done to warn
background image
T. DAVIES
230
International Conference Vajont 1963-2013. Thoughts and analyses after 50 years since the catastrophic landslide Padua, Italy - 8-10 October 2013
(one shown in Fig. 3). The hypothesis that postgla-
cial earthquake shaking has initiated a failure surface
in the upper part of the schist slope, which extends
deeper at every earthquake and is accompanied by
settling of the upper slope, can be neither proved nor
refuted prima facie, and neither can the corollary that
during a future earthquake a large rock avalanche will
occur from this slope. Evidently both hypothesis and
an extended time period; in the third, modification of
the natural situation appears feasible, at a cost that
is an order of magnitude smaller than by estimated
damage and loss of life.
CLEAR AND PRESENT DANGERS
FRANZ JOSEF GLACIER
The tourist township of Franz Josef Glacier lies
beneath the western range-front of the Southern Alps
in New Zealand’s South Island (Fig. 1). It is devel-
oping rapidly to support the growing tourism trade,
and during spring, summer and autumn may host 1-2
thousand tourists as well as several hundred perma-
nent residents. It lies on the Alpine fault, the active
boundary between the Australasian and Pacific tecton-
ic plates; this fault generates M~8 earthquakes every
2-300 years, the last of which was in 1717. The an-
nual probability of such an earthquake is ~1-2%. Im-
mediately behind and overlooking the township is a
hillslope with a distinctly stepped upper profile (Fig.
2; Barth, 2013), similar to those at two known large
prehistoric rock avalanches adjacent to the same fault
Fig. 1 - Outline map of New Zealand indicating potential
South Island landslide locations (K = Kaikoura;
FJ = Franz Josef; M = Milford Sound) and major
faults (AF = Alpine fault; HF = Hope fault)
Fig. 2 - Potential 10
7
m
3
rock avalanche (white dotted
line) above Franz Josef Glacier township. Grey
line = Alpine fault. Road bridge at lower right is
about 140 m long
Fig. 3 - The 0.6 km
3
coseismic Cascade rock avalanche,
South Westland (Barth, 2013). Dashed white line
is Alpine fault; dotted line outlines the 1 km wide
headscarp. Note stepped slope profile at far side
of headscarp, indicating partially-failed rock
mass. The 10 km
2
deposit is the hummocky area
at lower centre and right
background image
AVERTING PREDICTED LANDSLIDE CATASTROPHES: WHAT CAN BE DONE?
Italian Journal of Engineering Geology and Environment - Book Series (6) www.ijege.uniroma1.it © 2013 Sapienza Università
Editrice
231
walls. On its bed are the deposits of at about 20 rock
avalanches with minimum volumes ~10
6
-10
7
m
3
(Fig.
4; D
yksTra
, 2013), all of which have fallen since the
Fiord deglaciated about 15000 years ago. The rock
avalanches are probably coseismic, as the Alpine
fault runs just offshore about 16 km from the head of
the fiord, and there will be no warning of a rock ava-
lanche. Thus about once every thousand years a rock
avalanche generates a tsunami with an amplitude of
at least 4 m and a runup of ~17 m. It is estimated that
long-term this will give rise to about 0.5 deaths per
year at the present visitor rate (D
yksTra
, 2013), again
some orders of magnitude greater than societally ac-
ceptable; no mitigation is possible.
KAIKOURA
Some kilometres south of Kaikoura on the east
coast of the South Island (Fig. 1), a large sediment de-
posit is accumulating in the head of the Hikurangi Can-
yon due to northward drift of coastal sediments (Fig. 5).
There is evidence that about every 200 years this depos-
it (presently ~ 0.25 km
3
in volume) fails coseismically
corollary are of crucial significance to the future of the
town, and a detailed risk assessment needs to be car-
ried out, but no action has so far been taken by local
or national authorities. If the probability of the rock
avalanche occurring in the next earthquake is say 1%
(there have been about 60 earthquakes since deglacia-
tion at 15 ka), then the annual probability of the event
is about 1 in 10
-4
; if 100 people die then the annual
death risk is about 10
-2
, a very conservative estimate
and yet some orders of magnitude greater than the ac-
ceptable societal risk for such an event. The landslide
probability is, of course, additional to the much great-
er earthquake probability. There will be no effective
warning of the landslide, and no way of preventing it.
MILFORD SOUND
The fiord-head of Milford Sound (Fig. 1) is vis-
ited by thousands of people each day except in win-
ter when the access road is closed due to avalanche
danger. All visitors remain at sea-level for their whole
visit unless they take a scenic flight. Milford Sound
is a typical fiord with 1000-m high very steep rock
Fig. 4 - (Upper) View of Milford Sound - (Lower) Postglacial landslides on the bed of Milford Sound (D
ykstra
, 2013). Arrow
1 indicates line of sight of upper view. SM = landslide deposit; DF = density flow; L = lake. Shoreline in white
background image
T. DAVIES
232
International Conference Vajont 1963-2013. Thoughts and analyses after 50 years since the catastrophic landslide Padua, Italy - 8-10 October 2013
nami in Lake Geneva, 563 A.D.; k
remer
et alii, 2012).
There is very little room for doubt that the events out-
lined above are able to occur, and the estimated prob-
abilities lead to societal risks that are orders of magni-
tude greater than acceptable. The consequences in each
case are potentially catastrophic in both human and
economic terms, and would undoubtedly be severe in
political terms if the relevant authorities had been in-
formed of the situation. In the Franz Josef case, how-
ever, there remains some uncertainty about whether
the slope identified will ever collapse, and to resolve
this requires numerical dynamic stability analysis of
the slope - not a completely straightforward task due
to its difficult access, rainforest cover and location in
a national park, but certainly feasible. Given the gross
imbalance between uncertainties and potential conse-
quences, however, it is hard to believe that nothing has
been done to clarify or manage these situations.
MORE & BETTER SCIENCE?
To what extent is the unwillingness of authorities to
act on scientific information indicating likely future ca-
tastrophes the result of shortcomings in the quality and
quantity of the science? If this is indeed a factor, then
(l
ewis
& B
arnes
, 1999) and generates a local tsunami
with amplitude up to 10 m and runup of ~20 m (w
al
-
Ters
et alii, 2007). Tsunami deposits have recently been
identified ~15 m above sea-level close to the head of the
canyon. It has been estimated that this event can cause
up to 200 deaths and $200M of other costs; no effective
warning is possible due to the near-field nature of the
event. In this case, however, large-scale suction dredg-
ing of the head of the deposit could in principle increase
the stability of the deposit against coseismic failure, at
a cost of about $10M; this would subsequently need to
repeated at intervals of several decades. Considerable
analysis of the deposit stability would be required to de-
sign the mitigation dredging procedure and frequency,
and is certainly feasible. This hazard was first identified
in 1999, but apart from raising the awareness of the lo-
cal community little has been done.
None of these situations is in any way unusual
or extreme in geomorphological terms; the processes
involved are reasonably well-studied and well-under-
stood, and similar events have been recorded elsewhere
(e.g. recent coseismic rock avalanches in Alaska, Tai-
wan, Pakistan & China; catastrophic fiord tsunami at
Lituya Bay, Alaska, in 1958; submarine-landslide tsu-
Fig. 5 - Potential tsunamigenic submarine landslide south of Kaikoura
background image
AVERTING PREDICTED LANDSLIDE CATASTROPHES: WHAT CAN BE DONE?
Italian Journal of Engineering Geology and Environment - Book Series (6) www.ijege.uniroma1.it © 2013 Sapienza Università
Editrice
233
ly familiar with the processes and language of science,
nor with the personalities of scientists. Scientists are
also under ever-increasing pressures to justify their
existence by winning research grants and publish-
ing more papers than the next guy; and scientists are
generally not well-acquainted with local or national
government personnel or operations. Scientists see
politicians at all levels making decisions that in many
cases seem to go against scientific rationale, without
understanding that the soft factors in decision-making
can outweigh science, and they interpret this as lack
of rationality - to a scientist, the crowning sin. Politi-
cians see scientists coming at them with bewildering
demands for funding that they do not understand, and
that, if granted, may result in unwelcome information
becoming public; and with uninformed criticism of in-
stitutional decision-making that is far from endearing.
There is a lack of common ground and understand-
ing between scientists and government officials that,
bluntly, limits the trust each group has in the other.
This suggests that improving the communication of
science is a more complex task than at first appears.
The first requirement of improving science com-
munication is to improve scientists’ understanding
of the political process. This doesn’t mean that sci-
entists have to like or approve of the process; it is
the system we have and we can’t easily change it,
so it must simply be accepted as a necessary and un-
alterable fact of life - like gravity, it’s a hindrance
some times. (Incidentally, I suspect many politicians
have parallel reservations about the system in which
they operate.) Communication of difficult informa-
tion between two people is only possible if they trust
each other - only then will they be prepared to put in
the hard work needed to couch their own information
in carefully-designed terms, and to translate the oth-
er’s information so that it becomes acceptable into
one’s own world-view. Likewise, politicians need to
grasp the world in which scientists work, in order
to appreciate that science is (in principle) different
to policy or dogma - it should act as a real-world
constraint on policy and dogma, so that these do not
conflict with the behaviour of nature.
This is evidently not going to be a rapid or
straightforward process. It is to some extent recog-
nised by the occasional (and perhaps token) appoint-
ment of science advisors to high-level politicians,
and the presence of scientists in local government
we should expect the situation to improve over time
as science grows in both quantity and sophistication.
The fact that foreseeable landslide catastrophes are still
not taken seriously 50 years after Vajont perhaps in-
dicates that the available scientific information is not
a key issue - today we know much more about large
landslides and their causes and triggers than we did in
1962, and we have orders of magnitude more empirical
data on landslides that have been studied. Further, we
are much more capable of modelling the future behav-
iour of large slope failures on the basis of theory, and
can thus make quantitative predictions of failure much
more readily than was possible 50 years ago.
However, this improvement in current capabil-
ity does not mean the scientific investigation of a
potential catastrophe will be necessarily carried out
to this capability, because high-quality investigation
and modelling requires significant funding from the
relevant authority - which will not be forthcoming if
the authority is not seriously concerned. There is thus
a Catch-22; until the responsible authority is seriously
concerned by convincing information, convincing in-
formation cannot be provided.
This signals an interesting behavioural anomaly;
science continues to advance and improve the quantity
and quality of knowledge that can be provided about a
specific situation, and such improvements are substan-
tially funded by Governments (usually via Universi-
ties or Research Institutes) on the basis that the result-
ing improvements in disaster prevention will justify
the cost of the research
, but other government organi-
sations will not fund their utilisation in a specific case.
There is evidently a substantial contradiction here.
I therefore conclude that the availability of sci-
entific information is not a significant factor in the
reluctance of authorities to take action in respect of
potential catastrophes.
BETTER SCIENCE COMMUNICATION?
Another possible explanation for the lack of en-
thusiasm on the part of authorities to fund scientific
studies of potential catastrophes is that they don’t un-
derstand the situation; in other words, they are not
receiving information from scientists in a manner or
form that convinces them of the need to act. Govern-
ment officials are very busy people, under a wide
range of pressures (economic, political and cultural)
from both above and below; in addition, they are rare-
background image
T. DAVIES
234
International Conference Vajont 1963-2013. Thoughts and analyses after 50 years since the catastrophic landslide Padua, Italy - 8-10 October 2013
organisations - but the latter is increasingly rare as
organisational science is out-sourced. The fact is,
trust cannot be out-sourced: the personal acquaint-
ance between scientists and politicians (not to men-
tion planners and engineers) that used to occur in lo-
cal government offices is increasingly rare.
For these and several other reasons, better commu-
nication of science to decision-makers lies way beyond
simply rephrasing descriptions of what is in the minds
of scientists, so that decision-makers will comprehend
it. There are factors in comprehension that go far be-
yond agreement on the meanings of individual words
and phrases, and are not easily achievable. While sci-
entists generally do not communicate well, even within
the confines of their disciplines, the gains to be made
by teaching them to use simpler words are probably not
significant. Politicians are in general better at communi-
cating with the public at large - their future, after all, de-
pends on persuading people to vote for them - but their
willingness and ability to get to the heart of science and
be moved to act by it is constrained by a host of political,
logistical, cultural and personal factors
. And in this, I
believe, we get close to the kernel of the problem. Com-
munication is indeed the underlying issue, but we need
to dig deeper to understand the impediments to commu-
nication, before suggesting how it can be improved.
TIME-SCALES
Geologists think in geological time-scales, of up
to millions or billions of years. Earth scientists gen-
erally have shorter time-frames, but when thinking
about the intervals between large landslides the ap-
propriate time-unit is the century or millennium. By
contrast, society in general is accustomed to think in
human time-scales, which span generations (small
numbers of decades) at most. Societal decision-mak-
ing, however, is commonly carried out by elected
bodies whose duration is even shorter - seldom longer
than a small number of years; this time-frame is im-
posed by the need to stand for re-election at the end
of the period. Re-election requires societal popularity,
which is likely to be dependent of the attitude of the
populace to the current issues. In this context, re-elec-
tion is unlikely for someone standing for a policy of
restricting land-use because of a landslide that might
not happen for thousands of years. This is particularly
the case when local economic issues dominate the po-
litical landscape, and when immediate fiscal pressures
can evidently be relieved by increased development of
sites vulnerable to rare but catastrophic events.
Here we clearly have a difficult problem; how to
balance short-term ‘needs’ (in fact these are desires,
because what is at stake is a particular and relatively
high standard of living) against the genuine need of
long-term security against loss of life. Ironically, in
many such cases long-term cost-benefit analyses show
clear long-term net economic benefit in avoiding the
vulnerable areas, because the costs are so high when
the disaster occurs; but this has little impact when the
short-term economics are so much better assuming the
event does not happen in the short term
. Statistically,
of course, the disaster is much more likely to occur
over the long term. Nevertheless, the event will cer-
tainly happen one day (we don’t know when). In due
course the event will happen in the short-term - and
due course might be now.
This I believe is a fundamental reason that hazards
science finds little fertile soil in the minds of decision-
makers. A further reason lies in the mind-set of our
leaders. These are likely to be people of forceful men-
tality, who are not risk-averse, so they are in fact will-
ing - perhaps more willing than most of their elector-
ates - to take short-term risks for long-term gain. Those
who are fortunate enough to be successful naturally
ascend to positions of power where, it can be argued,
they are more in control of their circumstances and less
likely to be unlucky. Empirically many of them are
wealthy, again indicating greater-than-average ability
to judge short-term risks well. It is unlikely that such
people will take so seriously the threat of a future cata-
strophic event that they will forego short-term profit.
Finally, most of the populace and its leaders have no
experience of the types of event the scientists are talk-
ing about, so simply cannot be deeply affected by the
prospect of their occurrence.
Thus I posit an intrinsic tendency, built into the
structure of democratic society, to underemphasise
science related to infrequent but catastrophic disasters
of any kind, including large landslides. Recognition is
the first step towards remediation; but is there in fact
any way to alter this situation?
A SOLUTION: ACCOUNTABILITY?
The Achilles Heel of societal decision-makers is
that they are held accountable by society, via their
legal responsibilities, when disasters occur due to
background image
AVERTING PREDICTED LANDSLIDE CATASTROPHES: WHAT CAN BE DONE?
Italian Journal of Engineering Geology and Environment - Book Series (6) www.ijege.uniroma1.it © 2013 Sapienza Università
Editrice
235
quires a detailed investigation of the rock mass
characteristics of the hillslope, which is logistical-
ly challenging but certainly technically feasible.
k
aikoura
- Further work is needed to better charac-
terise the tsunami deposits at Goose Bay and
elsewhere; and, again, a 3D dynamic stability
analysis of the sediment accumulation against sei-
smic shaking.
m
ilforD
s
ounD
- Here, all that is needed is that the
paper be written.
Publication of such papers may seem a daunting
task for an individual scientist, especially a young
one with much to lose by antagonising funding agen-
cies and governments. It may result in some resist-
ance from the decision-makers, and perhaps from
the public. There may also be hostility from within
the conservative science community, for “giving sci-
ence a bad name” by “scaremongering”; one hopes,
however, that there would also be support from other
parts of the science community.
The papers must be able to be understood by lay
people, because the electorate alone has the power
to persuade governments to change policy or be-
haviour. Then, in the case that the decision-makers
decide to do nothing, and the public disagrees with
them, the decision-makers can be replaced at the
next election. If the public agrees with them, and
agrees to take responsibility for the risks it knows
that it runs, that is perfectly acceptable.
The earth scientist’s responsibility to society in
such cases is to use all possible methods to put the rel-
evant scientific information in front of the public and
the decision-makers in such a way that it can neither
be misunderstood nor ignored.
SUMMARY
The risks to society posed by future coseismic land-
slides in New Zealand are societally unacceptable and
in most cases manageable only by relocation of assets;
this management strategy is supported by long-term
cost-benefit analyses. However, governments are intrin-
sically loath to take seriously the risks posed by infre-
quent but catastrophic geological events, because their
priorities are short-term and would be constrained by
consideration of long-term risks. This conflict can per-
haps best be resolved by public awareness of and access
to scientific information describing the situation, in the
context of the government’s duty of care to its citizens.
demonstrable and wilful lack of care on their part.
Referring to my personal experience: D
aVies
(2002)
published a peer-reviewed paper in the Journal of Hy-
drology (New Zealand), demonstrating that the risk of
loss of life in a holiday park from landslide-dambreak
flooding was (a) some orders of magnitude greater
than societally acceptable and (b) manageable only
by relocation of the park. This public domain paper
came to the attention of the Government at both local
and national levels and, aided by an unusually risk-
averse Chief Executive in a key Ministry, the Minister
appreciated that if the disaster occurred, the Minister
would be responsible if nothing were done because
the public-domain information had been ignored. The
park was eventually relocated.
This may be applicable more generally. If there is
peer-reviewed and published science that directly and
unequivocally
demonstrates a clearly unacceptable risk
to the public from a demonstrably possible disaster (i.e.
of a type and scale that has occurred in a comparable sit-
uation elsewhere in recorded history), then any Govern-
ment that neglects to take action is, by omission, directly
responsible for the consequences. The words in italics
are key; there must be no room for doubt, no wriggle-
space for a smooth-talking Minister. In the case of Va-
jont, this was perhaps not the case; the scientifically-
foreseen collapse of Mt Toc could not be communicated
by reference to any other such event in recorded history,
thus leaving room for the science to be dismissed as
scare-mongering, because the public and government
had no pictures in their minds of such an event. The
same applies to the Mt St Helens disaster in 1980; there
was no historic description of an event of this nature and
scale to which public and politicians could relate.
THE WAY FORWARD
On this basis, the way forward for the three New
Zealand cases outlined above becomes clearer. There
must be peer-reviewed science published, and publi-
cised, specifically related to these cases that leaves no
doubt that the events can (and in fact will) occur; that
the consequential risks are unacceptable by a large
margin; and that either no mitigation is possible, or
mitigation can and must be undertaken. In two cases
further work is needed:
f
ranz
J
osef
- A 3D dynamic stability analysis of the
rock slope is needed to ascertain the degree of
shaking required to cause it to collapse. This re-
background image
T. DAVIES
236
International Conference Vajont 1963-2013. Thoughts and analyses after 50 years since the catastrophic landslide Padua, Italy - 8-10 October 2013
REFERENCES
B
arTh
N.C. (2013) - A tectono-geomorphic study of the Alpine Fault, New Zealand. PhD thesis, University of Otago, New
Zealand. 303 pp.
B
ilek
, s.l., s
aTake
, k. & s
ieh
, k. (2007) - Introduction to the special issue on the 2004 Sumatra-Andaman earthquake and the
Indian Ocean Tsunami. Bulletin of the Seismological Society of America, 97 (1 A SUPPL.): S1-S5
D
aVies
T.R.H. (2002) - Landslide dambreak flood hazards at Franz Josef Glacier township, New Zealand: a risk assessment.
Journal of Hydrology (NZ), 41: 1-17.
D
yksTra
J.L. (2013) - The role of mass wasting and ice retreat in the post-LGM evolution of Milford Sound, Fiordland, New
Zealand. PhD thesis, University of Canterbury, New Zealand. 307 pp.
e
Vans
s.g., B
ishop
n.f., f
iDel
s
moll
l., V
alDerrama
m
urillo
p., D
elaney
k.B. & o
liVer
-s
miTh
a. (2009) - A re-examination
of the mechanism and human impact of catastrophic mass flows originating on Nevado Huascarán, Cordillera Blanca, Peru
in 1962 and 1970.
Engineering Geology, 108: 96-118.
k
remer
k., s
impson
g. & g
irarDclos
s. (2012) - Giant Lake Geneva tsunami in ad 563. Nature Geoscience, 5: 756-757
l
ewis
k.B. & B
arnes
p.M. (1999) - Kaikoura Canyon, New Zealand: active conduit from near-shore sediment zones to trench-
axis channel. Marine Geology, 162: 39-69.
s
emenza
e. & g
hiroTTi
m. (2000) - History of the 1963 Vaiont slide: the importance of geological factors. Bulletin of Engineering
Geology and the Environment, 59: 87-97.
V
oighT
B. (1990) - The 1985 Nevado del Ruiz volcano catastrophe: anatomy and retrospection. Journal of Volcanology and
Geothermal Research 44: 349-386.
w
alTers
r., B
arnes
p.m. l
ewis
k.B., g
off
J.r. & f
leming
J. (2007) - Locally generated tsunami along the Kaikoura coastal
margin: Part 2. Submarine landslides. New Zealand Journal of Marine and Freshwater Research, 2006. 40: 17-28. DOI:
0028-8330/06/4001-0017.
Statistics