Memorandum submitted by Professor W P
Aspinall (SAGE 19)
I am responding to the committee's invitation
for written submissions on the issues identified in relation to
their inquiry into scientific advice and evidence in emergencies.
I am a consulting Chartered Scientist and Cabot
Professor in Natural Hazards and Risk Science at the University
of Bristol. My comments are informed by my experience as a member
of the Foreign Office Scientific Advisory Committee on Volcanic
Activity in Montserrat, as a member or the CSA's advisory group
SAGE on the recent Icelandic volcanic activity, and my other professional
experience in relation to hazard assessment and risk assessment
in safety critical industries, flight operational safety and medical
issues requiring expert judgment. These comments are made on a
personal basis and should not be construed as representing the
views of any organization, institution or committee with whom
I had been associated.
Before making some generic comments, I will
provide brief responses to the five points identified by the Committee
in relation to the four case studies they are considering. In
the main, my responses relate to the Icelandic volcanic ash eruption
in April 2010.
1. The hazard created by the volcanic eruption
was airborne ash, which entered UK and neighbouring airspace,
and the two main risks were (a) risks to flight safety (airworthiness),
and (b) operational disruption. Having been one of the people
called in to provide advice to SAGE, it was clear that the UK
government was little prepared for this emergency.
2. In this case, and in the case of the ongoing
eruption of the Soufriere Hills volcano on Montserrat, which started
in 1995, there was little if any preparation for such an emergency
and the government's response was, therefore, almost totally reactive.
For instance, in the case of the Montserrat crisis, the setting
up of a formal scientific advisory committee to provide advice
to the FCO was done at the instigation of scientists involved
in monitoring that crisis, and not as initiative of government
itself.
3. Unquestionably the government has sufficient
powers and sources to overcome obstacles, but an efficient and
unambiguous framework for obtaining reliable timely scientific
advice and evidence to inform policy decisions is not well established
within government. Moreover, in the case of the Icelandic ash
crisis this year, there was a signal absence of will to fund support
for urgent work to be done in the universities, in the British
Geological Survey, or by other outside specialists in industry
and consultancies with significant relevant experience. The imminence
of a general election seems a very questionable and lame excuse
for failing to properly mobilise and support vital hazard and
risk appraisal work for an emergency of such massive economic
impact, and one with a residual potential for mass casualties
when some airlines started operations while conditions were still
uncertain.
4. Strategic coordination between government
departments, public bodies, private bodies, sources of scientific
advice and the research basis has not been and is not fully effective
in preparing for or reacting to emergencies. A major problem,
which was manifest both in the recent Icelandic crisis and in
the Montserrat crisis, was the failure for one department in government
to take full ownership of the "problem"; in the case
of Montserrat, there was (and still is to some extent) a continuing
division of responsibility between the FCO and DFID, despite a
Select Committee recommending that one or the other take control
of the crisis after people were killed in 1997. In relation to
co-ordination between government and specialists other than those
in the government service or in universities, successful working
relationships are, not unnaturally, difficult to establish in
an emergency—prior liaison, breaking down silos and barriers,
invariably produces dividends, and the benefits can be attested
to from many other emergencies.
5. International coordination is very important
with respect to volcanic activity, and although the UK itself
has significant—possibly world-leading—volcanic risk
assessment and crisis management experience from the Montserrat
eruption, international collaboration is essential—such
activity can be very complex, is not confined to national borders,
and needs the best expertise for handling the many-faceted aspects
of the threats. This said, there have been some obstacles to strengthening
such collaboration imposed by the UK government: for instance,
an invitation was extended to me as a volcanic hazards specialist
to attend a major volcanic emergency exercise in Italy a few years
ago, but the invitation was not acted upon officially simply because
I was not a government employee—and this despite having
sat on the Montserrat SAC since its inception!
Moving on to more generic issues, any decisions
in relation to emergencies arising from the sorts of situations
mentioned in the Inquiry need to be "risk informed"
to the maximum extent possible. For the best and most defensible
policy decisions to be made where scientific uncertainty is a
key factor, comprehensive quantitative risk assessments (QRAs)
are needed. One thing that has struck me forcibly has been the
limitations in the extent and detail of application of good practice
QRA principles to the Montserrat and Iceland eruption scenarios,
when compared with the rigorous and exhaustive assessments undertaken
in this country for, say, external hazards to nuclear power stations,
for instance. [As well as participation in the two volcanic crises
at issue, I have been involved also in UK nuclear safety work
for more than 25 years, sit on various IAEA hazards committees,
and therefore have a wider than usual perspective on the issues].
The present selective scenario-based approach
for risk appraisal for national contingency management, such as
has followed on from SAGE deliberations over the Icelandic ash
crisis, is unable to capture all the intrinsic variabilities in
volcanic processes, and I think it fair to say that nearly all,
if not all, of the volcanologists involved in SAGE were dismayed
by the approach that has been taken. The point here is that identification
and quantification of uncertainties are absolutely key—uncertainty
in relation to the true strength of contributory factors and,
in particular, those which can be described as having heavy or
long statistical tails, can drive up the estimated level of hazard,
and hence risk. If these uncertainties are not adequately captured
or otherwise are under-stated, there is the prospect of events
happening that lead to major surprises, sometimes involving situations
falling outwith the scope of policy decisions that were intended
to deal with all eventualities.
In this regard, the approaches to the Soufriere
Hills volcano and Icelandic ash crises were ad hoc and
insufficiently formalized, in my view. Such events may be very
low probability but the consequences can be major, either in terms
of calamitous loss of life or disastrous economic costs (or, indeed,
both). The pathway from scientific uncertainty, through judgment
to advice and thence to crucial decision-making should be made
as explicit, auditable and traceable as possible. In essence,
a simplified approach was adopted in both cases, contrasting with
the sort of "scorched earth" risk assessments that are
customary for major safety-critical industries. In both cases,
the spectrum of expertise that was mobilized has been limited—for
the recent Icelandic ash episode, academics, government agency
staff and others acted on a pro bono basis, and a whole
range of other expert knowledge and professional judgment, that
exists in other domains, was effectively eschewed. In the Montserrat
case, the risk assessment work has also been done on a shoe-string
and, in many ways, it has been a matter of good fortune rather
than sound mitigation that no one there has been killed or injured
since the fatalities of June 1997 [ n.b. also there have been
at least three significant civil airliner ash encounters near
Montserrat during its fifteen years eruptive activity].
In the case of the Icelandic ash crisis, one
important consideration is the very high level of safety that
is achieved in normal operations by the civil aviation industry.
The occurrence of extraordinary and highly unpredictable scenarios—involving
extensive volcanological uncertainty and meteorological variability,
and contingent engineered system impact uncertainties, such as
ash-induced engine failure modes—can easily produce a situation
where margins of safety are (or might be) eroded to the extent
that, inevitably, such reduction, perhaps tolerated unwittingly,
could appear indefensible post hoc, if disaster ensues.
In this particular case, I was concerned in SAGE that there was
no comprehensive end-to-end uncertainty/hazard/risk assessment
with which to inform the urgent decisions being made; a detailed
and thorough assessment was required that needed much more attention
to detail and much more effort than was possible under the imposed
ad hoc conditions. As the economic costs of the flights
disruption were massive, even a major cost incurred in producing
better hazard and risk assessments would have been justified,
in this case at almost any imaginable cost—especially given
the possibility that such advice, if well-founded, could have
been used to sustain, at the earliest possible opportunity, a
restart of flights on a reasoned evidence-supported basis. The
impatience of some airlines to get flying again was not properly
evidence-based and hence not completely rational in the circumstances,
but this could happen because no one government department seemed
prepared to grasp the nettle and contemplate all the issues fully
and with funding support.
This latter point, about when an emergency can
be considered ended, is important: it is a common feature of many
unusual threats, such as a volcanic eruption, solar storm or pandemic,
that determining when to declare a crisis over is frequently much
more difficult than raising the alert in the first place, when
things are obviously escalating. Premature decisions in sounding
the "all clear" can incur unintended risks—having
successfully negotiated previous, on-going activity, declaring
a false dawn can be disastrous. Avoiding this pitfall requires
a significant measure of expert judgment and a judicious way of
acquiring it, under pressure.
Thus one element in the provision of sound scientific
advice to government in response to national emergencies, which
calls for some new thinking, is finding an optimal structured
procedure for the elucidation of expert opinion to derive a rational
consensus as the basis for decision support. My professional experience
in this regard includes facilitation of performance-based (ie
scored and differentially weighted) expert elicitations for forecasting
and mitigating volcanic eruption scenarios, for ranking and managing
flight operational safety issues, and for modelling emergent virus
and other medical challenges [eg SARS risks to health workers;
vCJD risks from blood products; emergence of the XMRV virus, a
gammaretrovirus first described only in 2006, with potential associations
to chronic fatigue syndrome and prostate cancer].
It seems to me that adopting this formalized
approach to expert judgement—when the circumstances are
appropriate—would be a valuable addition to the toolkit
of decision support techniques, especially in the matter of quantitative
risk assessment in the face of significant scientific uncertainty.
In addition, there are available nowadays useful (Bayesian) graphical
techniques for weighing and combining different strands of uncertain
scientific evidence, so that the import of such evidence can be
thoroughly and efficiently adduced for decision support. It also
seems clear to me that further work is needed to establish government-spanning
protocols and a decision framework for accessing scientific advice
and scientific evidence in emergencies, and for utilizing the
best techniques and methodologies—bearing in mind the latter
is an evolving domain that warrants those involved keeping current
with developments. Again, reliance on outmoded, inferior or casual
approaches can expose government to censure if things go pear-shaped.
Professor W P Aspinall
13 September 2010
|
