Geoscience Reference
In-Depth Information
fault length (1,200 km) was more than a third of the thickness of the earth's mantle and seismic
wavelengths substantially larger than major structural features.
(2) Similarly, the theory assumes a point source and neglects the signiicant interference
effects between the individual elements of an extended fault such as those capable of generat-
ing signiicant tsunamis.
Among the latter:
(3) A practical computation must be limited to a inite time window, the selection of which
requires a very subtle compromise between the necessary exclusion of later arriving seismic
phases and the use of an interval at least as long as the duration of the seismic rupture at the
source.
5
This compromise may not be possible for extremely long sources (eight minutes in the
case of the 2004 Sumatra earthquake).
(4) The algorithm implementing
M
wp
at the PTWC uses a distance correction adequate for
an ininite homogeneous elastic space rather than a more realistic description.
(5) Similarly, the operational algorithm elects to carry out the integrations in the time
domain, which assumes that the response of the seismic sensors is perfectly lat to ground
velocity. However, even the best seismic instrumentation (e.g.,
the Streckeisen STS-1) falls off
rapidly in its response at periods greater than 360 s, which leads to a systematic and predict-
able underestimation of the largest earthquakes, whose true size is expressed only at the
longest periods where the instrument begins to decline in response (see below in discussion of
W phase).
Such limitations in the operational aspects of
M
wp
were actually recognized by the opera-
tors of the centers during the development of their algorithms, through a comparison of their
results with published earthquake magnitudes obtained by the Harvard CMT project using
geophysical inverse procedures.
6
These authors documented that
M
wp
measurements become
increasingly deicient when the magnitude of the earthquake increases, and they proposed to
incorporate in the inal algorithm (i.e.,
the one presently used at the TWCs) a linear correction
for this effect, deined from an empirical regression of the misit of the values from their initial
dataset with respect to the published reference values.
In summary, the committee expresses concern that the TWCs rely on a single technique
applied without suficient attention to its limitations—both inherent in its concept and result-
ing from the particular algorithm chosen for its implementation. Furthermore, all magnitude
measures will saturate as the seismic moment increases and will not be able to quantify poten-
tial tsunami heights for great and mega-earthquakes.
INCORPORATING THE W PHASE
In 1992 during the Nicaragua tsunami earthquake, H. Kanamori identiied a “W” phase
wave: very-long period energy traveling along the earth's surface by multiple relections in the
mantle. Because of the ray-mode duality in wave propagation, the W phase is best thought
