Earthquakes - Extreme Events: A Physical Reconstruction and Risk Assessment - page 161

theshorelines were uplifted during earthquakes that occurred approximately
500, 1500 and 3600--3900 years BP.

Point measurements of surface rupture

Earthquakes usually occur due to fault movements and often the magni-
tude of an earthquake is proportional to the extent of that fault movement. The
larger the fault movement and, therefore, displacement between the rock units
separated by that fault, the more intense the earthquake. Measurements of this
displacement therefore can be used as a guide to the earthquake magnitude.
Two types of displacement can be measured: the length of surface ruptures and
the extent ofsurface displacement. Often only the lengths of surface ruptures
are used for this purpose rather than measurements of surface displacements.
Surface rupture lengths have been commonly used because:

(1) most data for historic ruptures include rupture lengths that can be

compared and compiled to estimate magnitude;

(2) magnitude estimates are most reliably correlated to fault length in his-

toric data sets;

(3) measurements of fault length are easily acquired from geologic maps,

seismicity plots and air photos; and

(4) segmentation schemes can be used to break large faults into rupture

segments.

Hemphill-Haley and Weldon (1999)note,however, that there are several reasons
why surface rupture lengths may not be reliable. These are:

(1) rupture length estimates are based on identification of often subtle,

fragile geomorphic features that are easily eroded or buried, especially
along long-recurrence faults;

(2) recent earthquakes have demonstrated that surface ruptures can inte-

grate faults that were previously not known to be related;

(3) there are difficulties in assessing single event rupture segments on even

themost active, mature faults with relatively short recurrence intervals;
and

(4) segmentation schemes are generally not quantifiable in the sense that

one can assign an uncertainty to the choice of a segment boundary.
Often this last factor is handled with decision trees, where scenarios are
weighed, but this approach cannot result in true uncertainties because
only a few possibilities are considered and weights are assessed by expe-
rience that is often shared by the experts who weigh the scenarios.