Geography Reference
In-Depth Information
Parkfield area had done little to prepare for the
predicted tremors.
Could the Parkfield Prediction Experiment thus be
judged a failure? Not exactly: a national working group that
was convened to evaluate it argued that Parkfield remains
the best place in the country to 'trap' an earthquake and
monitor any possible precursors. Although the short-term
results had not been encouraging, the long term promises
a rich harvest of useful scientific data. However, the lesson
of 1992-3 is that communication between scientists and
the public needs to be improved, as too much emphasis
was given to short-term prediction, and the public had little
appreciation of the wider goals of the experiment.
Moreover, the impact of false alarms was underestimated.
In conclusion, short-term earthquake prediction
remains an elusive goal, as most earthquake faulting
mechanisms are unique, complex and, of course,
hidden deep in the ground. Yet the January 1995
Great Hanshin earthquake at Kobe in Japan produced
anomalies in strain rate, groundwater discharge, and
the radon and chlorine content of groundwater, which
began three months before the earthquake and were
manifest at four locations 20-50 km from the
epicentre. It is an open question as to whether they
could have been recognised before the main shock
and interpreted in such a way as to protect the
population.
minutes at the regional scale. This is insufficient at
the local scale, but it is compensated for by
THRUST, which has achieved a seventeen-second
average response time (Bernard 1991).
Geological Survey has developed a portable
observation network and associated training
scheme that can be used anywhere in the world
where volcanic hazards are serious. Be this as it
may, when a volcanic emergency occurs, it may
last for months and require long-term evacuation
(UNDRO/UNESCO 1985), as was the case with
the Campi Flegrei bradyseismic activity (i.e.
coastal vulcanism without surface eruption,
composed mostly of a dome-like uplift) which
affected the city of Pozzuoli, west of Naples, over
1983-5 and necessitated the permanent
evacuation of 55,000 people (Zelinsky and
Kosinski 1991).
MANAGEMENT OF VOLCANIC AND
SEISMIC EMERGENCIES
Although earthquakes cannot be predicted
accurately in the short term, much effort has been
invested in monitoring their precursors (Rikitake
1984), and thus the methodology and technology
have improved progressively. If precursors are
insufficiently clear for prior warning of major
earthquakes, then it is possible that once the
shaking begins sensitive equipment, including
computer systems and fast trains (Nakamura and
Tucker 1988), can be shut down before damage is
done.
The prediction of volcanic eruptions is more
feasible and relies on a variety of instruments,
including infrared sensors for monitoring heat
emissions, which are mainly the preserve of
satellite remote sensing (Rothery 1992). However,
Tilling and Lipman (1993) argued that
vulcanologists rely too much on pattern
recognition (the empirical approach to predicting
eruptions) and not enough on the understanding
of source mechanisms. On the other hand,
Martinelli (1991) suggested that seismic signals are
the best guide to impending eruption, especially if
they can be related to other precursors. As
monitoring is the key to prediction, the US
SOCIAL RESEARCH
Geographers have played a minor but significant
role in the process of planning for volcanic
emergencies (Chester 1993), and they have played
a more considerable part in planning against
earthquakes. This has involved both hazard analysis
(Beatley and Berke 1990) and policy studies
(Berke and Beatley 1992). The latter have included
analyses of the 'window of opportunity' for policy
formulation and implementation that opens when
a disaster has occurred recently and public opinion
demands that something be done (Solecki and
Michaels 1994). In the absence of such events,
planning can be based on a synthetic form of
'reality' by using scenarios that prefigure the
damage and casualties that will arise from an
earthquake disaster. Thus Borchardt (1991) used
