RUSSIAN JOURNAL OF EARTH SCIENCES VOL. 10, ES5002, doi:10.2205/2008ES000303, 2008
[15] Earthquake catalogue of Wadia Institute of Himalayan Geology includes 2628 seismic events occurred from 1552 till 30.01.2005. Seismological network was significantly improved in 1988. Therefore 1856 earthquakes from 01.01.1999 till 30.01.2005 were used for spatial modeling.
[16] These data were used for compiling the following three grid based models:
[17] To compile the models the seismic data are scanned with moving spatio-temporal window.
Seismic parameters are estimated by the earthquakes with the epicenters inside the window.
Evidently, the accuracy of the estimates depends on a number of the epicenters. It is supposed
that a set of epicenters inside the window is statistically uniform. For that the window must cover
homogeneous area in seismotectonic sense. Increasing the window size leads to doubtful
estimates because of mixing evidences from different probabilistic distributions. Decreasing the
window leads to uncertainties because of a small number of a sample set. A compromise consists
in adaptive fitting of the window size: the more density of seismic events the less the window
size.
[18] Moving window in our case is a cylinder with fixed height equal temporal interval from
01.01.1999 till 30.01.2005 and with radius which adaptively varies from
R1 to
R2. Minimal
number of seismic events in window
N min and nominal number of events
N are specified.
[19] The adaptive window algorithm is in following. Axis of the cylinder coincide with a
grid point
(l, j, t). If a number of events in cylinder of radius
R1 is
n N then
estimation is made. If number of events in cylinder of radius
R2 is
n
N min, then estimation is rejected. If
number of events in cylinder of radius
R2 is
N min
n
N, then estimation is made. If number of
events in cylinder of radius
R2 is
n > N, then the
N nearest events to the grid point
(l, j, t) is selected.
[20] Minimal representative magnitude of earthquakes m0 is the left boundary of the interval, in which Gutenberg-Richter low is hold. Value m0 depends on number of seismic stations, their spatial distribution, and sensitivity. The method of estimation of m0 is based on testing of statistical hypothesis. The method was developed by Pisarenko, [1989], algorithm was developed by Smirnov, [1995], and together with Smirnov was modified for GIS GeoTime [Gitis et al., 1994].
![]() |
Figure 2 |
[22] Maximum likelihood estimate of b-value is following:
where
b(l, j, t) = (1Nm0 mi
m0 (mi - m0(l, j, t)))-1,
Nm0 = Nm0(l, j, t) is a number of earthquakes with
m (m0(l, j, t), M max(l, j, t)).
![]() |
Figure 3 |
[24] Seismic activity
A(l, j, t) is defined as normalized according to spatio-temporal window, the
number of earthquakes adjusted to the magnitude
m (mA - dm2< mA + dm2)
![]() |
where
Nm0(l, j, t) is a number of earthquakes with the magnitudes
m (m0(l, j, t), M max(l, j, t)),
k=1/ST,
S is a spatial size of the window,
T is the temporal interval of the window.
![]() |
Figure 4 |
Citation: 2008), Analysis of seismicity in North India, Russ. J. Earth Sci., 10, ES5002, doi:10.2205/2008ES000303.
Copyright 2008 by the Russian Journal of Earth Sciences (Powered by TeXWeb (Win32, v.2.0).