Faults which have seismogenic, stick-slip behavior can be characterized as
being “locked” (i.e., not slipping at all) during interseismic periods
of elastic strain accumulation, in a certain depth range.
The lower locking depth (deeper limit) will probably also be close to the lower
limit of future earthquake ruptures.
The upper locking depth (shallower limit) may not be a barrier to ruptures, but
instead defines a shallow limit above which the fault plane contributes no
stress-drop to drive the earthquake rupture.
In other cases, the fault patch that contributes stress-drop to drive the earthquake
rupture may extend right up to the surface; then, the “upper locking depth” is
considered to be zero.
Program NeoKinema requires information on the locking depths of each
fault.
NeoKinema uses the the lower locking depths to bound the dislocation patches
which it creates to estimate the “mean coseismic velocity” implied by the
current neotectonic model.
(Mean coseismic velocity of GPS benchmarks is then added to observed
interseismic velocity of those benchmarks to estimate their long-term-average velocity.)
Therefore, the fault-offset-rate data table that we began to build in Step 11 includes two columns for the upper (shallower; numerically smaller) locking depth and the lower (deeper; numerically larger) locking depth of each fault. These are Columns G and H, which are labelled “ULxKm” and “LLxKm”, respectively. They require your input of depths, in units of kilometers. The value in Column H should typically be larger than the value in Column G.
There are two primary sources of information about locking depths:
Microseismicity (when recorded by a high-quality seismic array, and
then precisely relocated) can define the upper and lower limits of the
seismogenic zone in the upper crust.
In practice, a “high-quality” array implies modern broadband seismometers,
which are digitally recorded by a network with high sampling rate.
“Precise relocation” implies a double-difference algorithm that corrects for
station-delay terms, and also for local variations in seismic velocities.
Even with these qualifications, it must be remembered that some residual error
remains in all hypocentral depths; therefore, some fairly-arbitrary
decision-rule is needed, such as estimating the shallower and deeper limits of
the seismogenic depth range at 5% and 95% (respectively) on the cumulative
distribution function (CDF) for hypocentral depths.
Nazareth and Hauksson [2004, Bull. Seismol. Soc. Am., 94(3),
940-960] present such a study of the seismogenic depth range in southern
California, and find that it typically extends to about 15±1.2 km, but that it
varies locally from <10 km to >25 km.
However, I would like to emphasize that this method only works when the
earthquakes are located using local instruments; the “hypocentral
depths” determined using remote seismometers, and listed in global
catalogs such as GCMT and PDE, are seriously misleading! (This problem is
documented in Bird
& Liu [2007].)
Geodetic profiles of interseismic velocity measured across the strike
of faults can be interpreted to give the lower locking depth.
In the case of a vertically-dipping, infinitely-long, and straight strike-slip
fault, it is well-known that the interseismic velocity profile has an
inverse-tangent form, and that its characteristic width is controlled by the
lower locking depth.
Other faults with dip-slip and/or non-vertical dip require more complex
dislocation solutions, but these are available in the literature.
Therefore, a common product of geodetic campaigns in neotectonic active areas
is estimates of the lower locking depth for the most important faults.
NeoKinema provides a convenient option: For any fault whose locking
depths are unknown, it is possible to enter negative numbers (e.g., “-1.0”) in either/both column(s)
of the table.
This is taken as a flag indicating that the corresponding locking depth limit
should be obtained from the NeoKinema parameter file, such as parameters_for_NeoKinema.nki.txt
which was introduced in Step 8.
In this parameter file, the only distinction made is between subduction zones
(usually treated as locked from 14~40 km depth) and all other faults (usually
treated as locked from 1~12 km depth).
For any fault that is creeping (and thus has Column F of the
fault-offset-rate data table set to “T”),
the upper and lower locking depths in Columns G (“ULxKm”) and H (“LLxKm”) will
not be used.
However, to prevent an error during the reading of this data table by NeoKinema,
two place-holder numbers must still appear!
This is probably best handled by entering “-1.0”
twice, since that means “unknown,” and values which are undefined are
certainly unknown!
In conclusion, a good way to begin filling in Columns G and H of the table is to enter “-1.0” everywhere, and then to overwrite these place-holders with specific information wherever available.