Advances in Conception about Earthquakes - Earthquake Engineering for Structural Design - page 136

the P one being the first arriving to the site. Thus an early warning system for
earthquakes is possible only in case of earthquakes produced from deep sources
and excludes the ones of crustal fracture in the near source zones. The warning
time is rather small, ranging from seconds to a maximum of about one minute.
There are some examples of using this system. For Mexico City, situated about 320
km from the source (subduction of the Cocos plate beneath the North American
plate), the warning time is 60 to 75 seconds. For Bucharest, situated at 130 km
from the Vrancea epicenter, the warning time is about 25 seconds, while for Taipei
(Taiwan), about 10 to 17 seconds.
What is it possible to do during this very short warning time? The following
activities are possible: Evacuation of buildings (for time greater than 30 seconds)
or putting the inhabitants in a safer position, shutting-down the critical systems
(nuclear and chemical reactors), stopping the high-velocity train, putting the
strategic facilities into safe location, etc.
One must mention that, from the design point of view, only the long time
prediction, based on statistical analysis, can be considered in order to eliminate or
to limit the economical losses. The other predictions give the possibility to alarm
the authorities in order to prepare actions for limiting the loss of life.
4.6STATISTICALSEISMOLOGYANDTHETHEORYOFMULTI-
SOURCE DATA FUSION
Statistical seismology is a relatively new field, which applies statistical
methodologies to earthquake data in an attempt to raise new questions about
earthquake mechanisms and to make some progress towards earthquake
characteristic prediction (Vere-Jones, 2006, Vere-Jones et al, 2005). But the main
question to be agreed is: can the physics of earthquakes be a statistical problem
(Turcotte, 1999).
The physics of earthquakes covers a broad range of topics. Some of them are
quite well understood, but others not. Among the understood ones, the general
origin of earthquakes is well accepted as the displacement of tectonic plates along
the pre-existing faults. But many aspects still remain unsolved in the problem of
the dynamic rupture occurrence and of the prediction of earthquake characteristics,
because there are important differences between theoretical results and in-field
recorded values.
Considering the seismic data, there is a historical period until 1900 during
which only qualitative information was available and the last period of 50 years in
which seismic records have been available, being instrumentally monitored. For
this last period, there are many data for low seismic events, few values for
moderate earthquakes and, a very few for large earthquakes. Thus, a very
important question arises concerning the validity of the probabilistic methodology
in absence of sufficient valuable data. Due to this situation, the deterministic
methodologies have an important position in determining the seismic actions.
Supporters of the probabilistic approach defend its use, whereas the deterministic
followers deny the probabilistic way. The weakest aspect of the deterministic
approach is the tendency to give a definite numerical answer mainly based on the