Open access

# Aftershock Identification Through Genetic Fault-Plane Fitting

Written By

Submitted: 29 November 2010 Published: 01 August 2011

DOI: 10.5772/23733

From the Edited Volume

## Scientific and Engineering Applications Using MATLAB

Edited by Emilson Pereira Leite

Chapter metrics overview

View Full Metrics

## 1. Introduction

Since the earlier times of documented seismological observations, it was noticed that an earthquake (usually a large one) was followed by a sequence of many smaller earthquakes, originating in the epicentral region; the first, larger, earthquake is called the mainshock, or main shock or main event, and the following, smaller, earthquakes are called aftershocks. These sequences, and their spatial and temporal distributions, depend on the characteristics of the mainshock and on the physical properties and the state of stress, strain, temperature, etc., of the region of occurrence (Kisslinger, 1996).

The observation that in many aftershock sequences the magnitude of the largest aftershock is about Δ M 1.2 less than that of the mainshock is known as Båth’s law (Helmstetter & Sornette, 2003; Richter, 1958); it implies an energy ratio E ( M a f t ) / E ( M m a i n ) ~ 0.007 between largest aftershock and mainshock. An earthquake following some mainshock but not small enough to be considered an aftershock by the Båth’s law criterion is often considered to be a mainshock in its own right and to constitute, together with the mainshock, a multiple event or multiplet. A large event followed by an even larger one is demoted from mainshock to foreshock; a probabilistic calculation, based on the observation that aftershocks follow the Gutenberg-Richter relation (Gutenberg & Richter, 1954; Kisslinger & Jones, 1991; Shcherbakov et al., 2005), indicates that for Δ M = 1.2 the seismicity following a given mainshock has a ~ 6.3% probability of including a “daughter” event larger than the mainshock (Helmstetter et al., 2003; Holliday et al, 2008).

Thus, the simple definition of aftershock as an earthquake occurring after a mainshock and in its epicentral region, although implying some causal relation with the mainshock, is partly semantic and largely circumstantial. Indeed, smallish earthquakes that constitute the background seismicity occur all the time in a seismic region in the absence of large events, and continue occurring whether or not a large earthquake occurs, so that not all earthquakes occurring in the region after a mainshock are necessarily aftershocks.

If aftershocks are a result of the occurrence of the mainshock, then they should be related in a physical way with its rupture process. The aftershock-producing mechanism is not yet known, but it is conceivably related with adjustments to the post-mainshock stress field (Lay & Wallace, 1995) possibly through viscolelastic processes or through fluid flow (Nur & Booker, 1972); whichever the actual process, aftershocks should be related with the rupture plane of the mainshock. Kisslinger (1996) qualitatively defines three kinds of aftershocks: Class 1 is those occurring on the ruptured area of the fault plane or on a thin band around it. Class 2 is events that occur on the same fault but outside of the co-seismic ruptured area. Class 3 is events occurring elsewhere, on faults other than the one ruptured by the mainshock; these events, whether in the same region or not, will not be considered here as aftershocks, but rather will be classified as triggered earthquakes.

The number of aftershocks decreases with time after the mainshock according to the modified Omori relation (Utsu, 1961, 1969, 1970) as

N ( t ) = k ( t + c ) p E1

where t is time measured from that of the mainshock, and k, c, and p are positive parameters which vary with the lithologic, tectonic, and other conditions of the study area. Commonly, p ranges from 0.9 to 1.8, with most instances between 1.1 and 1.4 (Utsu, 1961); values which do not show dependence on the magnitude of the mainshock (Utsu, 1962).

Aftershocks occurring within 24 to 48 hours after a large earthquake locate mostly over the co-seismic rupture area, and provide a good estimation of it (Lay & Wallace, 1995), which indicates that seismicity at the time is mostly class 1; over longer times the aftershock area increases (Felzer & Brodsky, 2006; Helmstetter & Sornette, 2002; Mogi, 1968; Tajima y Kanamori, 1985; Valdés et al., 1982), including at the edges class 2 events.

Among other reasons why aftershock identification is important, we can mention the following few examples. Aftershocks can give important information about the rupture area; also, from estimations of co-seismic slip on the fault plane by inversion of seismic waves, several authors have found that aftershocks are scarce in areas of maximum slip and concentrate around their edges (Dreger et al., 1994; Engdahl et al, 1989; Hauksson et al., 1994; Mendoza & Hartzell, 1988), so that aftershocks give information about the rupture process of the mainshock. Aftershocks can also yield information about the properties of the epicentral region (Knopoff et al., 1982; Kisslinger, 1996; Kisslinger & Hasegawa, 1991; Figueroa, 2009), and about possible triggering mechanisms (Roquemore & Simila, 1994).

Since aftershocks can be large enough to contribute to the damage (particularly after structures have been debilitated by a mainshock), it is important to estimate the hazard associated with aftershock activity (Felzer et al., 2003; Reasenberg & Jones, 1989).

It has been proposed that, since aftershock activity depends among other factors on the stress status of the region, there is research on whether some characteristics of the aftershock activity from intermediate-sized earthquakes can be useful as precursory data for large earthquake hazard estimation (e.g. Jones, 1994; Keilis-Borok et al., 1980; Wyss, 1986).

Finally, for some studies concerned with large earthquakes, aftershocks can be considered as noise, and have to be eliminated from the catalogs (e.g. Gardner. & Knopoff, 1974; Habermann & Wyss, 1984).

In order to use or eliminate aftershocks it is first necessary to identify them. Many methods have been used, ranging from visual inspection (Molchan & Dmitrieva, 1992) to sophisticated numerical techniques. A common method identifies aftershocks as those shocks locating within temporal and spatial windows having lengths which usually depend on the magnitude of the mainshock (Gardner & Knopoff, 1974; Keilis-Borok et al., 1980; Knopoff et al., 1982). More sophisticated methods identify aftershocks as belonging to a spatial cluster, consisting of events within a given distance of at least one other event belonging to the cluster, which includes the mainshock (Davis & Frohlich 1991a,b; Frohlich & Davis, 1985). A variation of the window method considers events from larger to smaller magnitudes with the size of the spatial windows a function of the magnitude, and density of events (Prozorov & Dziewonski, 1982; Prozorov, 1986). Other methods include recognizing some statistical property (e.g. Kagan & Knopoff 1976, 1982; Vere-Jones & Davies 1966) or interpreting the relations between events according to some statistical chain or branching model (Molchan & Dmitrieva, 1992; Reasenberg, 1985).

Our method includes some of the above mentioned techniques used to discard events which cannot be aftershocks, and then proceeds to identify aftershocks based on the physical model of a rupture plane and on recognized statistical relationships. An early unsophisticated application of the rupture plane model, which proved that this principle of aftershock identification was feasible, was part of an unpublished MSc. thesis (Granados, 2000).

## 2. The method

We work with seismic catalogs containing occurrence time (days), hypocentral x (East), y (North), and z (up), and, optionally, horizontal and vertical location uncertainties u h and u v , all these in kilometers. If location uncertainties are not included in the catalog, optional horizontal and vertical uncertainties are assigned equally to all events.

Any events occurring before the event with the largest magnitude M max , the mainshock, are eliminated. All spatial coordinates are then referred to those of the mainshock.

A rough time cutoff, eliminates events occurring after more than an optional cutoff time (default is n y r = 4 years ), because after a few years it is difficult to distinguish aftershock activity from background seismicity.

The extent of the aftershock area depends on the energy, i.e. on the magnitude, of the mainshock (Utsu, & Seki, 1955), as does the rupture area. A first rough spatial discrimination, based on an average of the empirical magnitude M vs. source-length r (km) relationships of Wells and Coppersmith (1994):

r f = 10 ( M max 5 ) / 1.22 E2

or (Kagan, 2004):

r = 20 10 ( M 6 ) / 2 E3

eliminates events farther away from the hypocenter than 1.5 times the r f length estimated by (2).

Next, a spatial clustering analysis, where events separated by no more than a given critical distance r, of the order of hundreds of meters to a few kilometers, depending on the spatial coverage of the catalog, are considered to be related, is used to eliminate events which do not relate to the mainshock or to other possible aftershocks.

The parameters in the modified Omori’s law (1) are not known a priori; they are estimated from the statistics of the aftershocks (Davis & Frohlich, 1991b; Guo & Ogata, 1997; Ogata, 1983), which we do not yet have. However, this relation tells us that for long enough times after the occurrence of the mainshock the number of aftershocks decreases until seismic activity in the epicentral region returns to its background level (Ogata & Shimazaki, 1984). When aftershock occurrence shows gaps comparable to those characterizing the background seismicity, we can consider that the aftershock activity is, if not ended, at least scarce enough to be comparable to the background activity and can no longer be distinguished from it. The critical gap length depends on the region and the magnitude threshold of the observations; we use a default critical gap length t g a p = 10 days . The gap is measured as the average of a given number of inter-event times (default n g a p = 10 ). Events occurring after a critical gap are discarded from the possible aftershocks.

Weights, based on relation (1), are optionally assigned to the remaining events, using typical values for c (default c 2 ) and p (default p 1.0 ) and by setting k = c p N ( t = 0 ) = 1 , so that those events which have a large likelihood of being aftershocks have weights ~1, while later events have smaller weights:

w i = c p ( t i + c ) p E4

Next, plane fitting is carried out iteratively; at each iteration, a plane that passes through the mainshock hypocenter is fitted to all remaining events, through a genetic scheme described below, and fit outliers (events too far away from the plane) are discarded. Iteration continues until the goodness-of-fit criterion is met (successful fit) or until a preset maximum number of iterations is attained (unsuccessful fit).

The t g a p , c, and p values can be refined using the final results of a first, tentative aftershock determination, to do a second one.

The genetic plane search for the plane, characterized by its azimuth ϕ and dip δ , which minimizes the L1 norm of the perpendicular distances from the fault plane to all events, is as follows. An initial set of equispaced ϕ j and δ j values, covering the acceptable ranges, is built and the fit to the aftershock candidates is evaluated for each pair of values.

To estimate the error of fit for each candidate plane, for each azimuth ϕ j and dip δ j pair, event coordinates ( x , y , z ) are rotated as:

( x i y i z i ) = ( cos ϕ j sin δ j sin ϕ j sin δ j cos δ j sin ϕ j cos ϕ j 0 cos ϕ j cos δ j sin ϕ j cos δ j sin δ j ) ( x i y i z i ) E5

and in the ( x , y , z ) coordinate system the equation of the plane is x = 0 , so that the L1 fit error for the j’th parameter pair is easily computed as

ε j ε ( ϕ j , δ j ) = i = 1 N a | x i ( ϕ j , δ j ) w i | / i = 1 N a w i E6

where N a is the number of remaining aftershock candidates.

The parameter pairs corresponding to the best N p fits are chosen as the parents of the next generation; they are sorted by the absolute value of their respective errors, so that parent number 1 is the best fit and parent number N p has the largest error ε N p . The standard deviations of the parent’s parameters, σ ϕ and σ δ , are evaluated.

Next, N p ( N p 1 ) / 2 children are constructed with parameters which are averages of those of each pair of parents; other N c children are constructed for each parent j by randomly modifying one or the other of the parameter values, according to

ϕ j N p + k = N ( ϕ j , σ ϕ m ε j / ε N p ) δ j N p + k = N ( δ j , σ δ m ε j / ε N p ) k = 1 , , N c E7

where N designates the normal distribution, σ ϕ m = max { σ min , σ ϕ } , σ δ m = max { σ min , σ δ } , and σ min is a minimum allowable value that ensures significant variations.

Errors are computed for the children and the N p best fits among the whole population, parents plus children, are chosen as the parents for the next generation. The process is repeated until the goodness of fit criterion is met (and the process ends) or until a preset number of generations is attained.

For the current iteration the best fit is for the plane corresponding to parent number one, ϕ = ϕ 1 and δ = δ 1 ; using these values event coordinates are rotated as in (4) and the standard deviation, σ , of the x distances is evaluated. Those events with

x i f σ σ + u i E8

where f σ is a damping factor (default is 1.25) and u i is the location uncertainty, adjusted for the plane dip as

u i = ( u h i sin δ ) 2 + ( u v i cos δ ) 2 E9

are eliminated as outliers.

This method has been implemented as a Matlab program, aftplane.m, which allows the user to interactively set most parameters, 3D plots the input hypocenters, identifies the aftershocks of one mainshock, optionally plots the aftershocks and/or independent shocks (in different colors), and optionally outputs the corresponding catalogs. A definite advantage of using Matlab for this algorithm is that both data and trial models are handled as matrices, so that rotating, sorting, and identifying values is done more efficiently and with less lines of code than would be possible in other programming languages like FORTRAN or BASIC.

A variation of this method is used as a function by program cleancat.m which identifies and eliminates the aftershocks of all mainshocks in a catalog. For each event not previously identified as an aftershock, aftershocks are identified following the same steps described above, except that, after fitting the plane a search is made for events with magnitudes M M m a i n Δ M , which will not be considered as aftershocks, and if any are found, the aftershock list is cut so as to exclude all events beginning with the first of these non-aftershocks. Thus, for mainshocks occurring at times t i , instead of the total Omori number of aftershocks (Utsu, 1970; Ogata, 1983)

N ( t ) = i H ( t t i ) K i ( t t i + c i ) p E10

where H ( t ) is the Heavyside function, we are actually evaluating

N ( t ) = i χ ( t i t k t i ) K i ( t t i + c ) p E11

where

χ ( t ) = Π ( t 1 / 2 ) = { t 0 t 1 0 other t E12

Π ( t ) is the boxcar function, and t k t i is the time of the succeeding event occurring along the same fault plane and cluster or the time when the gap criterion is met ( t k = if no such event or gap occur). An event at t k may “inherit” some of the aftershocks from the one at t i , but they will be identified as aftershocks and be duly eliminated. Båth’s law indicates that Δ M should be 1.2, but this criterion results, for most catalogs, in too many mainshocks; our default value is thus Δ M = 1.0 , which means that we accept as aftershocks those with energy ratio E ( M ) / E ( M m a i n ) ~ 0.016 , but this Δ M value can be easily changed by the user. The largest events in the catalog are plotted vs. time above the initial and resulting event densities, to illustrate which aftershocks were eliminated. The program iterates the whole process, as many times as needed, until no more aftershocks are found

In cleancat, parameters are not set interactively, but can be easily adjusted in a list of adjustable parameters at the beginning of the code.

The program optionally outputs a catalog excluding identified aftershocks.

## 3. Application

We will now show some examples of the application of the method. The aftplane program will be used to identify fault planes and aftershocks from three mainshock-aftershock sequences from two different parts of the world featuring different tectonic environments. The cleancat program will be used to clean the catalog of a fault system.

### 3.1. Aftplane: transcurrent regime, Joshua Tree and Landers earthquakes

In 1992, two large earthquakes, the Joshua Tree, April 23 M W 6.2, 33.9°N, 116.3°W and the Landers, June 28 M W 7.3, 34.2°N, 116.5°W, occurred close together on a line previously unrecognized as a potential throughgoing seismogenic fault (Nur et al., 1993). We chose these events as illustration because, although both events have mainly strike-slip mechanisms, they have slightly different strikes and dips, so that we wanted to test whether the method could identify these small differences.

Figure 1 shows the location of the study area in souther California, USA, and its recent seismicity; the faults ruptured during the Joshua Tree and Landers earthquakes are located within the red diamond.

For both events we used maximum allowable horizontal and vertical location uncertainties of u h h max = 0.2 km and u v max = 0.5 km, respectively; a priory cutoff times n y r = 4 years ; maximum distance for spatial clustering r = 1 km; maximum permissible time gap t g a p = 10 days estimated as the average of n g a p = 10 inter-event times; Omori weighting parameters c = 2.0 days and p = 1.0 ; aftershock magnitude criterion Δ M = 1.0 ; fit criterion maximum error ε max km.

### 3.2. Aftplane: Joshua Tree earthquake

The catalog for the Joshuea tree earthquake contained 5075 events spanning 66.30 days (~0.182 yr). Figure 2 (left) shows the 3497 remaining events after the first rough elimination by acceptable uncertainties and by an estimated expected fault length of ~7.97 km corresponding to a critical distance of 11.96 km. Figure 2 (right) shows as blue circles the 3379 shocks identified as clustering with the main event.

The main Joshua Tree shock and the identified 1094 aftershocks are shown in figure 3, both in a plan view (left) which clearly shows the resulting 171.6º faultplane azimuth, and a cross section along the fault plane azimuth (right) which shows the resulting 86.6º faultplane dip. The values found by aftplane agree extremely well with those estimated by Velasco et al. (1994) of strike 171º, dip 89°. Figure 4 shows a cross section parallel to the fault plane, illustrating aftershock concentrations.

#### 3.2.1. Aftplane: Landers earthquake

The catalog for the Landers earthquake contained 49,605 events spanning 4,932.52 days (~13.514 yr). Figure 5 (left) shows the 17,553 remaining events after the first rough elimination by acceptable uncertainties and by an estimated expected fault length of ~76.78 km corresponding to a critical distance of 115.17 km. Figure 5 (right) shows as blue circles the 12,834 shocks identified as clustering with the main event.

The main Landers shock and the identified 3,225 aftershocks are shown in figure 6, in a cross section seen along the determined 340.6º fault plane azimuth, which shows the resulting 70.1º faultplane dip. The values found by aftplane agree extremely well with those estimated by Velasco et al. (1994) of strike 341°º, dip 70°. Figure 7 shows a cross section parallel to the fault plane, illustrating aftershock concentrations.

### 3.3. Aftplane: subduction regime, Armería (Tecomán, Colima) earthquake

The Armería M W 7.4 earthquake, also known as the Tecomán or Colima 2003 earthquake, occurred on January 22, 2003, on the subduction zone along the boundary of the Northamerican and Pacific plates in central-western Mexico.

Figure 8 shows the location of the study area, the mainshock epicenter (red star) and the subsequent seismicity recorded and located by the Colima Seismic Network (RESCO).

Nuñez et al (2004) and Yagi et al (2003) estimated a fault plane with a 300° strike and a quite shallow 20° dip, which agrees with the 20° to 30° dip of the subduction zone determined by Andrews et al (2010).

For aftplane we used the RESCO catalog with the same parameter values mentioned above, except for horizontal and vertical location uncertainties of u h = 0.075 km and u v = 0.50 km, and maximum distance for spatial clustering r = 4 km;

The catalog for the Armería earthquake contained 11,475 events spanning 1,529.9 days (~4.192 yr). Figure 9 (left) shows the 10,275 remaining events after the first rough elimination by acceptable uncertainties and by an estimated expected fault length of ~92.73 km corresponding to a critical distance of 139.09 km. Figure 9 (right) shows as blue circles the 7,109 shocks identified as clustering with the main event.

The main Armería shock and the identified 460 aftershocks are shown in figure 10, in a cross section seen along the determined 86.2º fault plane azimuth, which shows the resulting 33.2º faultplane dip. The values found by aftplane agree extremely well with the above mentioned estimates strike 300° and 20° to 30° (Nuñez et al., 2004; Yagi et al., 2004; Andrews et al., 2010). Figure 11 shows a cross section parallel to the fault plane, illustrating aftershock concentrations.

### 3.4. Cleancat: whole Joshua Tree-Landers fault system

To illustrate the use of program cleancat we chose the catalog covering the whole Joshua Tree-Landers fault system (Figure 1), because this is a system with many close-lying, subparallel, faults, which gives scope to the iterative aftershock recognition scheme of the program.

The parameters used are u h h max = 0.2 km and u v max = 0.5 km; a priory cutoff times n y r = 4 years ; maximum distance for spatial clustering r = 2 km; maximum permissible time gap t g a p = 30 days estimated as the average of n g a p = 10 inter-event times; Omori weighting parameters c = 2.0 days and p = 1.0 ; aftershock magnitude criterion Δ M = 1 ; fit criterion maximum error ε max = 0.35 km.

Figure 11 shows all events in the catalog (black crosses), and identified aftershocks as yellow circles. Total processing consisted of 10 iterations which identified and eliminated 11,665, 4,212, 1,702, 86, 94, 30, 80, 18, 49, and 1 aftershocks, respectively, for a total of 17,937 aftershocks.

Figure 12 shows the occurrence times and magnitudes of the largest events in the catalog (top), and below them (middle) is plotted the ocurrence density (per Δ t = 46 day) vs. time (blue line); peaks in the occurrence rate after large shocks aftershocks are clearly seen. The bottom panel of figure 12 shows the occurrence density after aftershock elimination using Δ M = 1.0 (red line); although densities are much lower than before filtering, the peaks coinciding with the occurrence of the Joshua Tree and Landers earthquakes indicate that many aftershocks are not being identified.

Use of Δ M = 0.9 (Fig.13) effectively diminishes the troublesome above mentioned peaks to background seismicity level; the seismicity rate peak around t ~ 7.25 10 5 days is associated with distributed seismicity with events about the same size. Thus, we see that Båth’s law is not appropriate for the mainshock-aftershock relationships in the seismicity of the Joshua Tree-Landers fault system; a small (0.1 unit) change in the magnitude criterion can make a large difference in the aftershock recognition capability.

## 4. Conclusions

We present a simple method for identification and/or elimination of aftershocks, based on the generally accepted assumption that aftershocks are related to the fault rupture of the mainshock. The method has been tried on various catalogs with good results and, when aftershocks are numerous enough, good estimates of rupture planes that agree very well with those reported in the literature.

A variation of the method used for eliminating all aftershocks from a seismicity catalog (“catalog cleaning”) uses, iteratively, a variation of the same principle. Using the seismicity occurrence time rate as illustration and criterion of the effectiveness of the method, indicates that the required difference between mainshock and aftershocks Δ M is a key parameter for correct aftershock identification, and that Δ M = 1.2 (Båth’s law) may be too strict for some geographic areas and/or seismo-tectonic settings.

## Acknowledgments

Many thanks to José Frez, Juan García A., and María Luisa Argote for useful criticism and comments. We are grateful to RESCO and Gabriel Reyes, and to the SCEC for the use of their catalogs. We also thank Mr Davor Vidic of Intech for the kind invitation to participate in the present book.

## References

1. 1. Andrews,V., Stock, J. Ramírez-Vázquez,C., & Reyes-Dávila, G. (2010). Double-difference Relocation of the Aftershocks of the Tecomán, Colima, Mexico Earthquake of 22 January 2003. Pageoph, DOI 10.1007/s00024-010-0203-0
2. 2. Davis S. D. Frohlich C. 1991a Single-link cluster analysis, synthetic earthquakes catalogues and aftershock identificationGeophys. J. Int. 104 289 306
3. 3. Davis S. D. Frohlich C. 1991bSingle-link cluster analysis of earthquake aftershock: decay laws and regional variations. J. Geophys. Res. 96 6335 6350
4. 4. Dieterich J. 1994 A constitutive law for rate of earthquake production and its application to earthquake clustering J. Geophys. Res. 99 2601 2618
5. 5. Dreger, D., Pasyanos, M., Loper, S., McKenzie, R., Gregor, N., Uhrhammer, B., & Romanowicz, B. (1994). Source process of the 17 January 1994 Northdridge earthquake. EOS, 75(16), 103. (abstract
6. 6. Engdahl, E. R., Billington, S., & Kisslinger, C. (1989). Telesismically recorded seismicity before and after the May 7, 1986, Andreanof Islands, Alaska, earthquake”. J. Geophys. Res. 94 15481 15498
7. 7. Felzer K. Abercrombie R. Ekstrom G. 2003 Secondary aftershocks and their importance for aftershock forecasting Bull. Seismol. Soc. Am. 93 1433 1448
8. 8. Felzer K. Abercrombie R. Ekstrom G. 2004 A common origin for aftershocks, foreshocks, and multiplets Bull. Seismol. Soc. Am. 94 88 98
9. 9. Felzer K. Brodsky E. 2006 Decay of aftershock density with distance indicates triggering by dynamic stress. Nature 441 735 738
10. 10. Figueroa A. 2009Analysis of inter-event time in aftershock sequences for identification stress relaxation status., MSc Thesis, UNAM, 89pp (in Spanish).
11. 11. Frohlich C. Davis S. D. 1985 Identification of aftershocks of deep earthquakes by a new ratios methodGeophys. Res. Lett. 12 713 716
12. 12. Gardner J. Knopoff L. 1974Is the sequence of earthquakes in Southern California with aftershocks removed Poissonian? Yes. Bull. Seismol. Soc. Am. 64 1363 1367
13. 13. Granados,J. 2000 Identification of seismic aftershocks in catalogs. MSc.Thesis in Earth Sciences, CICESE, 145pp (in Spanish).
14. 14. Guo Z. Ogata Y. 1997 Statistical relations between the parameters of aftershocks in time, space, and magnitude J.Geiphys.Res.102(B2), 2857 EOF
15. 15. Gutenberg B. Richter C. 1954 Seismicity of the Earth and associated phenomena.Princeton Univ. Press, 310pp.
16. 16. Habermann R. Wyss M. 1984Background seismicity rates and precursory seismic quiescence: Imperial Valley, California, Bull. Seismol. Soc. Am. 74 1743 1755
17. 17. Hauksson, E., Hutton, K., & Kanamori, H. (1994). The Mw 6.7 Northdridge, California, earthquake of January 17, 1994 and its aftershocks. Program Northdridge Abstr., 89th Annu. Meet. Seismol. Soc. Am., unpublished abstract
18. 18. Helmstetter A. Sornette D. 2002 Diffusion of epicenters of earthquake aftershocks, Omori’s law and generalized continuous-time random walk models. Phys.Rev.E 66, 061104 EOF
19. 19. Helmstetter, A. & Sornette, D. (2003). Båth’s law derived from the Gutenberg-Richter law and from aftershock properties. Geophys.Res.Lett.30, 2069
20. 20. Helmstetter, A., Sornette, D., & Grasso, J. (2003). Mainshocks are aftershocks of conditional foreshocks: How do foreshock statistical properties emerge from aftershock laws. J.Geophys.Res. 108(B1), 2046
21. 21. Holliday J. Turcotte D. Rundle J. 2008 A review of earthquake statistics: fault and seismicity-based models, ETAS and BASSPageoph 165 1003 1024
22. 22. Kagan Y. Knopoff L. 1976 Statistical search for non-random features of the seismicity of strong earthquakes Phys. Earth planet. Inter. 12 291 318
23. 23. Kagan,Y. 2004 Short-Term Properties of Earthquake Catalogs and Models of Earthquake Source Bull. Seismol. Soc. Am. 94 1207 1228
24. 24. Keilis-Borok V. Knopoff L. Rowain I. 1980 Bursts of aftershocks long term precursors of strong earthquakes. Nature 283 259 263p.
25. 25. Kisslinger C. 1996 Aftershocks and fault-zone properties Advances in geophysics 38 1 36
26. 26. Kisslinger C. Hasegawa A. 1991Seismotectonics of intermediate-depth earthquakes from properties of aftershocks sequences. Tectonophysics 197 27 40
27. 27. Kisslinger C. Jones L. M. 1991Properties of aftershocks sequences in southern California. J. Geophys. Res., 96 11947 11958
28. 28. Knopoff L. Kagan Y. Knopoff R. 1982b-values for foreshocks and aftershocks in real and simulated earthquake sequences. Bull. Seismol. Soc. Am. 72 1663 1675
29. 29. Jones L. M. 1994 Foreshocks, aftershocks, and earthquake probabilities: Accounting for the Landers earthquake Bull. Seismol. Soc. Am., 84 892 899p.
30. 30. Lay T. Wallace T. 1995 Modern global seismologyAcademic Press, Inc., 521pp.
31. 31. Lin, G., Shearer, P. M., & Hauksson, E. (2007). Applying a three-dimensional velocity model, waveform cross correlation, and cluster analysis to locate southern California seismicity from 1981 to 2005. J. Geophys. Res. 112, B12309, doi: 10.1029/2007JB004986
32. 32. Lomnitz C. 1966Magnitude stability in earthquake sequences. Bull. Seismol. Soc. Am. 56 247 249
33. 33. Mendoza C. Hartzell S. H. 1988Aftershock patterns and mainshock faulting. Bull. Seismol. Soc. Am. 78 1438 1449
34. 34. Mogi K. 1968Development of aftershock areas of great earthquakes. Bull. Earthquake Res. Insti., Tokyo Univ., 46 175 203
35. 35. Molchan G. M. Dmitrieva O. E. 1992 Aftershock identification: Methods and new approaches Geophys. J. Int. 109 501 516
36. 36. Núñez-Cornú, F. J., Reyes-Dávila, G. A., Rutz-López, M., Trejo-Gómez, E., Camarena-García, M. A., & Ramírez-Vazquez, C. A. (2004). The 2003 Armería, México Earthquake (M w 7.4): Mainshock and Early Aftershocks, Seism. Res. Lett. 75 734 743
37. 37. Nur A. Booker J. 1972Aftershocks caused by pore fluid flow? Science, 175 885 887
38. 38. Nur A. Ron H. Beroza G. 1993 The nature of the Landers-Mojave earthquake line Science 261 201 203
39. 39. Ogata, Y. & Shimazaki, K. (1984). Transition from aftershock to normal activity: the 1965 Rat Islands earthquake aftershock sequence. Bull. Seismol. Soc. Am. 74 1757 1765
40. 40. Ogata Y. 1983Estimation of the parameters in the modified Omori formula for aftershock frequencies by the maximum likelihood procedure. J. Phys. Earth, 31 115 124
41. 41. Prozorov A. Dziewonski A. 1982A method of studying variations in the clustering property of earthquakes: application to the analysis of global seismicity J. Geophys. Res. 87 2829 2839p.
42. 42. Prozorov A. 1986Dynamic algorithm for removing aftershocks from the world earthquake catalog. Comp. Seism. 19 58 62Eds. Keilis-Borok, V. & Levshin, A., Nauka, Moscow.
43. 43. Reasenberg P. A. Jones L. 1989 Earthquake Hazard after a mainshock in California Science 243 1173 1176
44. 44. Richter C. 1958 Elementary seismology.W.H. Freeman & Co., USA, 768 pp.
45. 45. Roquemore, G. R. & Simila, G. W. (1994). Aftershocks from the 28 June 1992 Landers earthquake: Northern Mojave Desert to the Coso Volcanic Field, California. Bull. Seismol. Soc. Am. 84 854 862
46. 46. Shcherbakov R. Turcotte D. Rundle J. 2005Aftershock statistics. Pageoph 162 1051 1076
47. 47. Tajima F. Kanamori H. 1985 Global survey of aftershock area expansion patterns Phys. Earth Planet. Int. 40 77 134
48. 48. Utsu T. Seki A. 1955Relation between the area of aftershock region and the energy of the mainshock. Zisin, 7 233 240
49. 49. Utsu T. 1957Magnitudes of earthquakes and occurrence of their aftershocks, Zisin Ser. 2, 10,: 35-45.
50. 50. Utsu T. 1961A statistical study on the occurrence of aftershocks, Geophys. Mag., 30 521 605
51. 51. Utsu T. 1969Aftershocks and earthquake statistics (I). J. Fac. Sci. Hokkaido Univ., Ser. VII, 2 129 195
52. 52. Utsu T. 1970Aftershocks and earthquake statistics (II)- Further investigation of aftershocks and other earthquake sequences based on a new classification of earthquake sequences. J. Fac. Sci. Hokkaido Univ., Ser. VII, 3 197 266
53. 53. Valdés C. Meyer R. Zuniga R. Havskov J. Singh K. S. 1982 Analysis of the Petatlan aftershocks: numbers, energy release and asperities. J. Geophys. Res. 87 8519 8527
54. 54. Velasco, A., Ammon, C., & Lay, T. (1994). Empirical green function deconvolution of broadband surface waves: Rupture directivity of the 1992 Landers, California (Mw = 7.3), earthquake. Bull. Seismol. Soc. Am. 84 735 750
55. 55. Vere-Jones D. Davies R. 1966 A statistical survey of earthquakes in the main seismic region of New ZealandPart II. Time series analysis. New Zealand J. Geol. Geophys 9 251 284
56. 56. Wells D. L. Coppersmith K. 1994 New Empirical Relationships among Magnitude, Rupture Length, Rupture Width, Rupture Area, and Surface Displacement Bull. Seismol. Soc. Am. 84 974 1002
57. 57. Wyss,M. (1986) Seismic quiescence precursor to the 1983 Kaoiki (MS = 6.6), Hawaii, earthquake. Bull.Seismol.Soc.Am. 76 785 800
58. 58. Yagi, Y., Mikumo, T., Pacheco, J., & Reyes, G. (2004). Source Rupture Process of the Tecomán, Colima, Mexico Earthquake of 22 January 2003, Determined by Joint Inversion of Teleseismic Body-Wave and Near-Source Data, Bull. Seism. Soc. Am. 94 1795 1807

Written By