Layout 6 1 ANNALS OF GEOPHYSICS, 62, 5, SE568, 2019; doi: 10.4401/ag-8020 “SIGNIFICANCE OF FRACTAL CORRELATION DIMENSION AND SEISMIC B-VALUE VARIATION DUE TO 15TH JULY 2009, NEW ZEALAND EARTHQUAKE OF MW 7.8„ Saroj Kumar Mondal *,1, Paresh Nath Singha Roy2, Joshi K. Catherine1, Auchitya Kumar Pandey3 (1) CSIR−National Geophysical Research Institute, Hyderabad−500007, India (2) Department of Geology and Geophysics, IIT, Kharagpur 721302, India (3) Department of Applied Geophysics, IIT (ISM), Dhanbad−826004, India 1. INTRODUCTION The occurrence of high frequency earthquake in any region reveals the development of stress in the crust that leads to deformation. Such areas must be focused for understanding the mechanism of earthquakes and hence to find out the probable potential zone for fu- ture large size earthquakes. Several workers have ap- plied various methodologies to find out the precursory signature for future earthquakes, but a working model is yet to be achieved. Puzzling earth scientist has the quest to understand these complex tectonic patterns due to the heterogeneity of the crust and non-avail- ability of any such model for precursory study. How- ever, globally earthquakes rarely seem to be having surface outcrops of active faults. Most of the proposed models are of geological rather than actual physical base. Thus, with a view to have a more valuable per- ceptive of the physical model of earthquake occurrence, the fractal statistics have been used in which crustal Article history Receveid November 6, 2018; accepted April 12, 2019. Subject classification: Clustering, Coulomb stress, Earthquakes, Fractal correlation dimension, Seismic b-VALUE. ABSTRACT New Zealand earthquake that occurred on 15th July 2009 (Mw 7.8) was analysed using fractal correlation dimension (Dc) and seismic b-VALUE. We have analysed the earthquakes catalog of thirty-five years with a magnitude (mb ≥3.7), in order to observe a crucial in- formation in terms of Dc value fluctuation for the event. The event is preceded by fall and anomalous change in Dc value in the year 2007 about two years prior to the mainshock. A sudden decrease in Dc value with highly clustered events is observed before the main- shock. The low value of Dc is an indicator of clustering and it shows how intermediate size events correlate with one another in the pre- paration process of this event. Here the low Dc value may be the indicator for high stress developer along the fault to produce large size earthquake. Moreover, we also observed abnormal fluctuation in b-VALUE from 2003. The fractal clustering and scaling of earthquakes are indicated by b-VALUE change prior to strong earthquake as a harbinger of stress correlation in various scales. The event is also mar- ked for that occurred in the periphery of the positive Coulomb stress development, as obtained from three low Dc time windows' events. The drop in Dc value is not a single observation prior to this large event, but such pattern is also seen for other strong events in the study zone. One such well identified strong event is Mw 7.2 (2003) along with low Dc value prior to the event. Thus, stress correlation mea- sured along with these indirect statistical tools gives the clue of self-organization of long wavelength of stress, which was not measu- red earlier with classical approaches. This type of study may provide a very useful information for hazard mitigation. deformation pattern may be studied using fractal di- mension. Mandelbrot [1983] was the first to put the idea of fractal geometry and dimension to describe the scale invariance in natural phenomena. Latter, the concept of fractal structure has been applied by several researchers [Hirata, 1989; Feng and Seto, 1999; Dimitriu, et al., 2000; Kayal, et al., 2012; Roy and Mondal., 2012; Han, et al., 2015] to understand the complex mechanism of occurrence of earthquakes. Moreover, distribution of earthquakes and its pattern of evolution in different parts of the globe were also analysed [Legrand et al., 1996; Oncel and Wilson., 2002; Roy et al., 2012; Roy and Gupta, 2015; Mondal et al., 2016; Mondal and Roy, 2016]. Recently Michas, et al., [2015] have shown the significance of earthquake study using fractal analysis. Scientific community [Hirata et al., 1987] has also reported the laboratory experiment of decrease in frac- tal dimension with the development of rock fractur- ing process. Similarly the temporal variation of fractal dimension (Dc) of seismicity gives the idea of prepa- ration process for the large size earthquakes. Several large size earthquakes in different parts of the world have been studied statistically and have been found to be associated with some clustering of earthquakes be- fore the main shock earthquake (Roy and Nath., 2007; Roy and Mondal., 2010). The scaling nature in distri- bution of earthquake occurrences in space and time domain are the general characteristics, which are re- quired for understanding the physics of earthquake leading to seismic hazard assessment [Sobolev et al., 1991; Tomilin and Voinov, 1995; Knopoff et al., 1996; Telesca et al., 2003a, 2003b; Mukhopadhyay et al., 2004; Zamani and Agh-Atabai, 2009; Teotia and Kumar, 2011; Özturk., 2018]. The seismicity distribution in New Zealand (Figure 1) shows that the deformation is widespread, diffused and persistent. Here the scattered nature of seismicity is evident on both regional and local scales (Figure 1). New Zealand is located along a zone of contact between Pacific and Australia plates. The motion of these plates leads to the major seismicity in this country. The 15th July 2009 earthquake was the largest in the past 80 years. This thrust-related earthquake occurred within the Fiordland subduction zone of southwestern South Island at 9:22 UTC. The Mw 7.8 earthquake rup- tured the plate boundary interface between the sub- ducting Australia plate and the overlying Pacific plate [Beavan et al., 2010]. Mahesh, et.al [2011] reported that this quake has significantly increased the Coulomb stress on the overriding plate particularly on the off- shore portion of the Alpine fault. Here we have applied two statistical parameters Dc and seismic b-VALUE fluctuation with respect to one an- other for more than thirty years to see the anomalous be- haviour before the New Zealand earthquake (Mw 7.8). Further, Coulomb stress change prior to event has been studied to show the significance of Dc and b-VALUE study. 2. GEO-TECTONIC SETTING OF THE STUDY AREA Intense seismicity is experienced by the people of New Zealand, due to its tectonic plate motion (Fig.1). Australia and Pacific plate’s boundary run along the Puysegur Fault, Puysegur Trench, the Alpine Fault and the Hikurangi Trench. The country is adjacent to the subducting oceanic crust of Pacific plate below the con- tinental crust of Australia plate to the northeast at Hiku- rangi margin. On the other hand, Australia plate is subducting below the Pacific plate to the southwest of the country, at Puysegur margin. A very long Alpine fault of 650 km crosses the country and link the two subduction margin. During the early Miocene, displace- ment of about 480 km was estimated due to strike slip deformation along the Alpine Fault. New Zealand”s Southern Alps is considered to be formed due to the es- timated shortening of about 70-90 km across the Alpine Fault, during the last 6.4 Myr. The plate motion is vary- ing between 35-40 km/Myr from northern Puysergur Trench to Southern Hikurangi margin [De Mets et al., 1994]. The intense seismic activity in the Fiordland re- gion was marked with both shallow and intermediate earthquakes. Shallow earthquakes of dip-slip and strike- slip focal mechanisms are found, distributed in a nar- row zone along the Fiordland margin [Smith and Davery, 1984; Reyners, 1989; Anderson and Webb, 1994]. Tectonically and geographically the region is very complex. 3. DATA SET For the fractal analysis of the New Zealand earth- quake, we prepared the catalog using the global catalog of USGS and catalog of Geo-Net, considering earth- quakes of (mb≥ 3.7) during the period 1973-2009. In our analysis epicentral distance based fractal analysis was considered but not the hypocentral location due to the presence of higher error values in hypocentre. The com- pleteness magnitude plays a vital role in finding out the significance of stress analysis. Considering the signifi- cance of statistical analysis, we merged the two catalogue MONDAL ET AL. 2 3 SIGNIFICANCE OF Dc AND b-VALUE ANALYSIS BEFORE LARGE EARTHQUAKE to make our data set complete for this type of analysis. We estimate the completeness magnitude using classical Gutenberg-Richter [1944] relation as given in Fig. 2a. As we know that the Earthquake scaling and frequency of occurrence relations requires that small earthquakes will be just as important as larger ones in redistributing the forces that drive relative displacements across active faults of any dimension, including plate boundaries [Hanks, 1992]. So we have taken care of that in our anal- ysis. While considering the catalogue of USGS and local catalogue of Geo-Net, for the entire periods from 1973- 2009, we felt that the earthquakes which are missing in one agencies are filled up by the other agency which are very much necessary to give better idea about the stress pattern / seismicity distribution. Here we focussed in finding the pattern of Dc value only before the main event of 15th July, 2009. So the data set after 2009 has not been taken for the analysis. The temporal variation of Dc and b-VALUE are estimated by taking successive 100 events” windows in the latitude range 40°S - 50°S and longitude range 162° E - 178° E (Figure1). We have taken all parameters from Global Centroid Moment Tensor (GCMT) catalog for Coulomb stress analysis. 4. METHODOLOGY 4.1 CORRELATION FRACTAL DIMENSION (Dc) Correlation integral method [Grassberger and Pro- caccia, 1983; Roy and Ram, 2006; Roy and Padhi, 2007; Roy and Mondal, 2012a, 2012b] is applied for this set of earthquakes to determine correlation dimension. Correlation function is defined as: (1) where N = total number of vector pairs, 𝑟𝑖𝑗 = distances between the points of a set, 𝐻 = Heaviside step function, 𝑟 = length scale. Fractal structure of the epicentral distribution is ob- tained by the relation given below: C(r) ~ rDC (2) 𝐶(𝑟) = � � 𝐻(𝑟–𝑟𝑖𝑗) 2 𝑁(𝑁–1) � � 𝑗=1 𝑖=𝑗+1 3 FIGURE 1. Tectonic map of study area with significant fault system and seismicity in and around the main event that occurred dur− ing the study period. MONDAL ET AL. 4 FIGURE 2. a) Cumulative Frequency magnitude distribution of seismicity for New Zealand earthquake for the present study. b) Plot for log C(r) versus log r is shown for low Dc value before the main event. c) Plot for log C(r) versus log r is shown for high Dc value for the window of main event a) b) c) 5 SIGNIFICANCE OF Dc AND b-VALUE ANALYSIS BEFORE LARGE EARTHQUAKE The plot of C(r) versus r in logarithmic coordinates is used to get correlation dimension (Dc).Two plots for log C(r) versus log r are shown in Fig. 2b for low Dc value before the main event and Fig. 2c for high Dc value for the window of main event. 4.2 SEISMIC b-VALUE The b-VALUE has been computed using the classical Guttenberg Richter relation [Gutenberg and Richter, 1944], and the equation is given as: Log10 N(m) = a‐bm (3) Where N(m) is the number of earthquakes with mag- nitude larger than or equal to m, “b” is scaling parame- ter and “a” is a constant. The plot for data set is shown in Fig. 2a and the power law is valid. Self-similar prop- erty of earthquakes occurrence is considered in this law. The constant “a” depends on the overall seismicity of the region. The b-VALUE is normally 1, but it varies from 0.5 to 1.5 depending on the seismicity and tectonic setting of a seismically active region [Reasenberg and Jones, 1989; Kisslinger and Jones, 1991]. Here we have considered sampling of 100 events for estimation of b-VALUE with standard deviation of ~ 0.1b. The error value increases as the number of sam- pling earthquakes decreases. We can estimate b-VALUE with minimum of 25 events having standard deviation as low as ~ 0.25b [Bernice, 1983]. The estimation of b-VALUE is more reliable if the sampling subset con- tains minimum of 100 earthquakes. The higher the size of subset more reliable it is [Utsu, 1965]. 4.3 COULOMB STRESS CHANGE Coulomb Stress change may give the idea of stress change in the earth crust due to occurrence of the earthquakes prior to large size event and hence the pat- tern of seismicity in the study region. Observation of increased stress due to an event may explain how other events may be induced by the previous one in the fault zone. The earthquakes that occur in the increased static stress are supposed to have been advanced toward fail- ure by the positive stress increment, where the negative stress region are considered to be more relaxed. Static stress change may be the cause of rupturing processes due to previous events. It has been seen that the small, medium and large aftershocks generally do occur in the region of “stressed-up” by a mainshock or damaging earthquakes and that generally do not occur in the “stress-shadowed” region [Ellsworth et al.,1981; Simp- sons and Reasenberg, 1994; Jaume and Sykes,1996]. Researchers have also shown the relation between stress change and seismicity rate using some studies on the Coulomb stress relative to some global large earth- quakes [Reasenberg and Simpon; 1992; Stein et al. 1992; King, et al., 1994; Stein, 1999]. Here, we have applied in this study to see how the stress is correlated with the low Dc value patches which are considered as the possible highly stressed area. Here, we have com- puted the stress changes for entire study region by con- sidering the 2009 earthquake (MW 7.8) as mainshock earthquake source, whose strike, dip and rake values are 26°, 24° and 140° respectively [Mahesh et al., 2011]. Coulomb failure stress, ΔCFS has been computed in the elastic half-space associated by mainshock on the as- sumption of Poisson’s ratio of 0.25 and shear modulus values of 32 GPa [Okada, 1992]. Furthermore, we have utilized the equation (4) to examine the coulomb stress changes given by King et al., (1994), which is related to the shear stress changes and normal stress changes [King et al., 1994; Stein, 1999; Shan et al., 2013; Liu et al., 2017]. ΔCFS = Δτ +µ′Δσ (4) ΔCFS > 0 fault plane loaded ΔCFS < 0 fault plane relaxed where, Δτ is shear stress changes, µ′ is the friction co- efficient and Δσ is normal stress changes (positive for unclamping). For ΔCFS calculation, the constant fric- tion coefficient value has been assumed for fault µ′ = 0.40 and also uniform slip model has been con- sidered to estimate the ΔCFS for the entire study region. We checked the result through two different tests for entire region by using different values of coefficient of friction 0.5 and 0.6 to check the reliability of results and it has been found that very small changes are ex- isting and that can be neglected. Previously, several au- thors [Ellis et al., 2006; Boulton et al., 2012; Barth et al., 2013] utilized the low coefficient of friction in the range of 0.12-0.37 but it was found that these values show the weak results. Moreover, the value of coeffi- cient of friction (µ′ = 0.40) has been considered as stan- dard value by several authors [Gahalaut et al., 2011; Toda and Enescu, 2011; Xiong et al., 2017; Pandey et al., 2019] for different tectonic setting to compute the Coulomb stress changes that’s why we have used the coefficient of friction value 0.4 for this study region. Generally, the coefficient of friction is dependent on the category of rocks [Byerlee and Brace, 1968; Byer- lee, 1978] therefore entire result is based on rock type. We have used this parameter to observe the change in stress in two different plots, one with three time win- dows events that produced low Dc value and second with the entire data set used in the study region to see the pattern of stress transfer. We have estimated Coulomb stress changes by utilizing all data available from GCMT catalogue for three low Dc time windows and the entire Dc time windows from GCMT catalog. 5. RESULTS AND DISCUSSION Fault systems for a seismically active region are mul- tifaceted natural arrangement unveiling scale-invariant or fractal deformation. The manifestations of earth- quakes follow the fractal distribution property both tem- porally and spatially. The distribution of fault systems and earthquakes manifestation may be quantified by a single value called fractal dimension (Dc) which follow power law distribution. The deformation in general obey power-law scaling of fault length or earthquake distri- bution. We can measure both the heterogeneity and the clustering of earthquakes evolution using fractal di- mension of earthquakes scattering [Oncel et al., 1996]. A highly stressed region is associated with dense cluster- ing of smaller and medium size earthquakes which in- dicates strain release of the accumulated energy due to tectonic plate motion. These highly clustered region of earthquakes may be the most possible weak zones for potential site for future large size earthquake. High clustering of events in a region has been ob- served with low fractal dimension value leading to in- formation of seismicity before large earthquakes [Telesca et al., 2001; Roy and Padhi, 2007; Roy and Mondal, 2010]. Significance of occurrences of low fractal dimen- sion associated with high clustering before the large earthquake have also been reported by many researchers (Hirata et al., 1987; Roy and Ram., 2006; Roy and Nath., 2007; Roy and Mondal., 2010 ). The degree of clustering of earthquakes is inversely proportional to the Dc value. So the lower value is associated with high clustering and vice-versa for a region [Hirata et al., 1987; Kagan et al., 2006]. Fractal correlation dimension (Dc) of seismicity of a region is utilised to find out the estimated cluster- ing of earthquakes. 6. CORRELATION FRACTAL DIMENSION (Dc) ANALYSIS The present study for the New Zealand earthquake (Mw 7.8) shows a very good and clear indication of highly clustered events before the mainshock. We ob- served densely clustered events both in the case of using entire data set for this study (Figure 1), as well as more prominently in the case of three low Dc value contribut- ing events (Figure 3). These low Dc values are observed before the mainshock (Figure 4). Diagnostic pattern is re- vealed with highest Dc value in the year 2003 and one lowest Dc value in the year 2007 prior to the 15th July 2009 earthquake of Mw 7.8 as shown in Fig. 4 and Table 1. This anomalous variation in Dc value occurred just few years before the event. In the process of the mainshock three consecutive low Dc value were observed, which are the most clustered pattern of seismicity (Figure 3). These windows have the Dc values viz. 0.98 during April 2006, 0.94 during March 2007 and the third being the lowest is 0.66 during September 2007 (Table 1). Dc value of 100 events windows clearly depicts the fluctuating nature of statistical value called Dc with time (Figure 4) and Table 1. However it may be noted that the low Dc value charac- teristics is unique with space and time. The reason be- hind is particularly due to the possible approach of algorithm to quantify actual relative clustering spa- tiotemporally. Further it is known to us that the physical or geological facets change with space and time. Espe- cially the geochemical, thermodynamic etc. processes controlling the elasto-dynamics of the entire seismogenic study zone. In fact in support of these we can see the how the earthquakes cluster may easily be observed with the data set as shown in Figure 3 and Figure 1. Similarly the Dc patches are plotted using the above data set and we find the lowest Dc patch happened to be close to the mainshock (Figure 5) and (Figure 6). The low Dc contour as observed in Figure 5 gives clear indication of highly stressed zones [Roy and Padhi, 2007; Oncel et al., 1996; Roy and Mondal, 2010] and the main event occurred close to it. Essentially the Coulomb stress study also shows that the main event has occurred in the “stressed- up” region (Figure 7). Here the clustering pattern of in- termediate size events prior to main event is an example of strain liberation in self-organized fashion. Stress wave- length especially the longer one i.e regional one tries to get correlated with the smaller one in local scale. But this correlation can be observed only by seeing the clustering of earthquakes in fractal scale invariance or statistically self-similar fashion. We observed highly anomalous vari- ation in Dc value for this large size earthquake. The study area has also experienced other strong events. The strong events in different time windows is listed in table 1. Relative low Dc value for the earth- quake Mw 7.2 (2003) is also observed in Fig. 4. The clus- tering of earthquakes prior to this event is also seen (Fig.8). The clustering of events prior to other strong earthquakes are also noticed (Figure 9). Other two strong earthquakes of magnitude Mw 7.1 (2004) and Mw 7.4 (2007) are observed with clustering pattern Figure 9 and MONDAL ET AL. 6 7 SIGNIFICANCE OF Dc AND b-VALUE ANALYSIS BEFORE LARGE EARTHQUAKE FIGURE 4. Temporal variation of Dc value for 100 events window with error bar. FIGURE 3. Spatial distribution of seismicity that contribute to three low Dc value. These events have induced low Dc value win− dows prior to main shock. MONDAL ET AL. 8 FIGURE 5. Three low Dc value contour for the New Zealand earthquake of hundred events window. The lowest Dc values patch rep− resents possibly stressed region and clustering of events. Mean Time (100 events windows) Dc value b-VALUE Magnitude (Maximum in each window and source of data) 92.22 1.21 0.66 7.2 (1981) (USGS) 209.50 1.07 0.66 6.5 (USGS) 250.90 1.29 0.69 7 (USGS) 295.10 1.15 0.69 6.6 (Geo-Net) 341.00 1.03 0.76 6.2 (USGS) 362.58 1.13 0.63 7.2 (2003) USGS 372.63 1.36 1.24 5.9 ( USGS) 372.70 1.65 1.29 5.4 (2003) (Geo-Net) 373.08 1.16 0.47 6 (Geo-Net) 374.71 1.20 0.39 6.4 (Geo-Net) 382.39 1.29 0.37 7.1(2004) (USGS) 393.84 1.15 0.88 5.7 (Geo-Net) 404.77 0.98 1.32 5.4 (USGS) 415.67 0.94 1.15 7.4(2007) (USGS) 422.01 0.66 0.76 6.8 (USGS) 427.56 1.21 0.71 6.1 (USGS) 438.24 1.22 0.82 7.8 (15th July 2009) (USGS) 444.43 0.87 0.66 6.2 (Geo-Net) 444.55 0.87 0.99 5.6 (Geo-Net) 445.80 1.04 1.05 6.1 (USGS) TABLE 1. Mean time of 100 events windows for determination of Dc and b−VALUES. Figure10 Mw 7.2 (2003) earthquake occurred after two decades of event Mw 7.2 (1981). On the other hand events Mw 7.1 (2004) and Mw 7.4 (2007) occurred dur- ing the first and third years respectively after the event of Mw 7.2 (2003). The occurrence of these two events of magnitude 7.1 and 7.4 within a very short period indi- 9 SIGNIFICANCE OF Dc AND b-VALUE ANALYSIS BEFORE LARGE EARTHQUAKE FIGURE 7. Coulomb stress change due three low Dc time windows’ events. FIGURE 6. Dc value contour for the New Zealand earthquake of hundred events window of all the data set used. The lowest Dc val− ues patch represents possibly highly stressed region and clustering events. MONDAL ET AL. 10 FIGURE 9. Spatial distribution of seismicity that contribute to three low Dc value for the earthquake of Mw 7.1 in 2004. FIGURE 8. Spatial distribution of seismicity that contribute to three low Dc value for the earthquake of Mw 7.2 in 2003. 11 SIGNIFICANCE OF Dc AND b-VALUE ANALYSIS BEFORE LARGE EARTHQUAKE FIGURE 11. Temporal variation of b−VALUE for 100 events windows. FIGURE 10. Spatial distribution of seismicity that contribute to three low Dc value for the earthquake of M 7.4 in 2007. MONDAL ET AL. 12 FIGURE 13. b−VALUE contour for the New Zealand earthquake of hundred events window of all the data set used. FIGURE 12. Three low b−VALUE contour for the New Zealand earthquake of hundred events window. cate the probable reason for the fracturing process of the main event of Mw 7.8 in 2009. We do not have the suf- ficient data to see the Dc value variation for the event of Mw 7.2 (1981). Low Dc value variation prior to strong events can be observed in Figure 4. 6.1 SEISMIC b-VALUE ANALYSIS Similar to the Dc value anomaly, we are able to see such significant changes in b-VALUE before the main- shock. Changes in the b-VALUE are seen in the plot of frequency versus magnitude as shown by Figure 11. The plot of b-VALUE clearly depicts the highly abnormal changes in the b-VALUE since August 2003 Figure 11 and Table 1. The event is observed to have occurred after the abnormal increase and decrease of b-VALUE. There were rapid variations in Dc and b-VALUE almost six years be- fore the mainshock. The b-VALUE of each 100 events win- dows were determined and listed in Table.1. The lowest b-VALUE is 0.37 and highest is 1.32. Highest Dc value of 1.65 corroborates with the b-VALUE of 1.29. This may be supported with the power law validity for both the meth- ods with the dispersed seismicity. Three low b-VALUE patches of Fig.12 show the anomalous changes near the main events, which is not reflected by b-VALUE of all the data set used (Fig.13). Further, Coulomb stress change for three low Dc time windows clearly indicate the event has occurred in the positive ΔCFS region (Figure 7), which is again not reflected in the entire data set (Fig.14) that used for Coulomb stress study. The change or variation in b- VALUE from usual value of 1 is dependent on other re- lated factors of the study region. High b-VALUE was reported by Mogi [1962] due to increase in the material heterogeneity or crack density. Similarly decrease in b- value was noticed due to increase in applied shear stress [Scholz, 1968a, 1968b], or an increase in effective stress [Wyss, 1973]. The b-VALUE gradually decreases before a major fracture, even under a constant load [Mogi, 1981]. So, there is a possibility of gradual decrease in the b- VALUE before a large size earthquake. We observed in- crease and decrease in b-VALUE prior to this event as in Fig.11. No matter whatever, the mechanism may be, it is generally recognized that there is a tendency for the b- value to decrease prior to a large earthquake. So un- doubtedly this provides a clue in foreshadowing large size earthquakes. But this does not necessarily mean that a large earthquake will occur when there is a decrease in the b-VALUE. Essentially this observation will help us to understand the crustal activities for future study. 6.2 COULOMB STRESS ANALYSIS We have calculated the Coulomb stress changes of 126 earthquakes dataset for the entire region of the study area, as depicted in the Figure 14. From this figure, it has been noticed that the highest stress changes are ob- 13 SIGNIFICANCE OF Dc AND b-VALUE ANALYSIS BEFORE LARGE EARTHQUAKE FIGURE 14. Coulomb stress change using the entire time windows’ events. tained in the northern part of the study region i.e. to- wards the Australia plate. On the other hand Pacific Plate region is dominated by low-stress changes. Low or av- erage stress changes have been observed in the rest of the study area. The Coulomb stress change ranges lie be- tween 0.05 to -0.05 bar. Fig.8 depicts the stress change due to three low Dc time windows. Low to high stress variation is noticed in different parts of the study region. Distributions of Coulomb stress estimated using the en- tire data set, which shows that the event Mw 7.8 has oc- curred in the negative values of ΔCFS region Figure 14. Then the area should be considered as stable due to low stress and hence no large earthquake should be expected. On the other hand, Coulomb stress obtained by using the three low Dc time windows clearly indicates that the event has occurred in the positive values of ΔCFS re- gion. (Figure 7). Thus the low Dc patches (Figure 5) which is considered the possible highly stressed region corroborates well with stressed region those obtained by Coulomb stress change, as depicted by Figure 7 for three low Dc time windows. Not only this, distribution of low b-VALUE patches showed the abnormal pattern of the events that has occurred in that region as given by Fig- ure 7 and Figure 12. We observed that the low Dc patches which are possibly highly stressed region found to correlate well with other seismotectonic parameters. 7. CONCLUSIONS The aim of this study is to observe how the variation of Dc and b-VALUE for New Zealand earthquake of Mw 7.8 event behave prior to mainshock and its possible correlation with Coulomb stress. We observed highly anomalous fluctuation of these two statistical parame- ters prior to the main shock. On the other hand, we don’t find such high fluctuation in the values of these parameters in the region with other strong events, though we have observed drop in Dc value. This may be the possible reason for instability in the area for large size earthquake. Correlation of Dc/b-VALUE with Coulomb stress shows that the Mw 7.8 event has oc- curred in the positive value of ΔCFS region as observed from three low Dc time windows, as shown by Figure 5, Figure 7 and Figure 12. This result also show that the possible highly stressed region, as seen in Figure 5 cor- roborated well with the Coulomb stress change Figure 7 for the considered Mw 7.8 event. These two seismotectonic parameters that have an important role to play in providing information of pat- tern of seismic activity of the region. We may come up with very good relation with GPS based crustal defor- mation observation for longer period, and hence segre- gate the highly hazardous zone for future large size earthquake. But the important thing is to be noted that for this type of analysis, the data set should be of large quantity and should be complete even for lower mag- nitude earthquakes. Authors would like to express their gratitude to Dr V. M. Tiwari, Director, CSIR-National Geophysical Research Institute, Hyderabad for his kind permission to publish the work. First author is deeply thankful to Dr V. K. Gahalaut, Di- rector, National Centre for Seismology, Ministry of Earth Sci- ences, India, for his valuable suggestions and kind help. We acknowledge Global CMT, USGS and Geo-Net for the earth- quakes catalog data. We thank the two anonymous reviewers and Editor, Barbara Orecchio for their constructive suggestions. REFERENCES Anderson, H. and T. Webb (1994). New Zealand seis- micity: Patterns revealed by the upgraded National Seismograph Network, New Zealand, J. Geol. Geophys., 37, 477-493. Barth, N. C., C. Boulton, B. M. Carpenter, G. E. Batt and V. G. Toy (2013). Slip localization on the southern Alpine Fault, New Zealand, Tectonics, 32, 1-21. Beavan, J., S. Samsonov, P. Denys, R. Sutherland, N. Palmer and M. Denham (2010). Oblique slip on the Puysegur subduction interface in the 2009 July Mw 7.8 Dusky Sound earthquake from GPS and InSAR observations: implications for the tectonics of southwestern New Zealand, Geophys. J. Int., 183, 1265-1286. Bernice, B. (1983). Maximum likelihood estimation of b-values for magnitude grouped data, Bull. Seism. Soc. Am., 73 (3), 831-851. Boulton, C., B. M. Carpenter, V. Toy and C. Marone (2012). Physical properties of surface outcrop cat- aclastic fault rocks, Alpine Fault, New Zealand, Geochem. Geophys. Geosys., 13, 1-13. Byerlee, J. D. and W. F. Brace (1968). Stick Slip, Stable sliding and earthquakes- effect of rock type, Pres- sure strain rate and stiffness, J. Geophys. Res., 73, 6031-6037. Byerlee, J. (1978). Friction of Rocks, Pure Appl. Geo- phys., 116, 615-626. De Mets, C., R. G. Gordon, D. F. Argus and S. Stein (1994). Effect of recent revisions to the geomag- netic reversal time scale on estimates of current plate motions, Geophys. Res. Lett., 21, 2191-2194. MONDAL ET AL. 14 Dimitriu, P. P., E. M. Scordilis and V. G. Karacostas (2000). Multifractal Analysis of the Arnea, Greece Seismicity with Potential Implications for Earth- quake Prediction, Nat. Hazards, 21, 277-295. Ellis, S., J. Beavan, D. Eberhart-Phillips and B Stockhert (2006). Simplified models of the Alpine Fault seis- mic cycle: stress transfer in the mid-crust, Geo- phys. J. Int., 66, 386-402. Ellsworth, W. L., A. G. Lindh, W. H. Prescott and D. G. Herd (1981). The 1906 San Francisco earthquake and the seismic cycle, in Earthquake Prediction: An International Review, Maurice Ewing Ser, edited by D. W. Simpson and P. G. Richards, AGU, Washington, D.C, 4, 126- 140. Feng, X. T. and M. Seto (1999). Fractal structure of the time distribution of micro-fracturing in rocks, Geophys. J. Int., 136, 275-285. Gahalaut, V. K., S. Rajput and B. Kundu (2011). Low seis- micity in the Bhutan Himalaya and the stress shadow of the 1897 Shillong Plateau earthquake, Phys. Earth Planet. In., 186, 97-102. Grassberger, P. and Procaccia (1983). Characterizations of stranger attractors, Phys. Rev. Lett., 50, 346-349. Gutenberg, B. and C. F. Richter (1944). Frequency of earthquakes in California, Bull. Seism. Soc. Am., 34, 185-188. Hanks, T. C. (1992). Small Earthquakes, Tectonic Forces, Science, 256, 1430-1432. Han, Q., A. Carpinteri, G. Lacidogna and J. Xu (2015). Fractal analysis and yule statistics for seismic pre- diction based on 2009 L’Aquila earthquake in Italy, Arabian J. Geosci., 8, 2457- 2465. Hirata, T. (1989). A correlation between the b-value and the fractal dimension of earthquakes, J. Geophys. Res.: Solid Earth, 94 (B6), 7507-7514. Hirata, T., T. Satoh and K. Ito (1987). Fractal structure of spatial distribution of micro fracturing in rock, Geophys. J. Int., 90, 369-374. Jaume, S. C. and L. R. Sykes (1996). Evolution of mod- erate seismicity in the San Francisco Bay region, 1850 to 1993: Seismicity changes related to the occurrence of large and great earthquakes, J. Geo- phys. Res., 101, 765-789. Kagan, Y. Y., D. D. Jackson and Y. F. Rong (2006). A new catalogue of southern California earth- quakes,1800-2005, Seism. Res. Lett., 77(1), 30-38. Kayal, J. R., V. Das and U. Ghose (2012). An Appraisal of the 2001 Bhuj Earthquake (Mw 7.7, India) Source Zone: Fractal Dimension and b-value Map- ping of the Aftershock Sequence, Pure Appl. Geo- phys., 169 (12), 2127-2138. King, G. C. P., R. S. Stein and J. Lin (1994). Static stress changes and the triggering of earthquakes, Bull. Seismol. Soc. Am., 84, 935-953. Kisslinger, C. and L. M. Jones (1991). Properties of af- tershock sequences in southern California, J. Geo- phys. Res., 96, 11,947-11,958. Knopoff, L., T. Levshina, V. I. Keilis-Borok and C. Mat- toni (1996). Increased long-range intermediate- magnitude earthquake activity prior to strong earthquakes in California, J. Geophys. Res.: Solid Earth, 101, 5779-5796. Legrand, D., A. Cisternas and L. Dorbath (1996). Multi- fractal analysis of the 1992 Erzincan aftershock sequence, Geophys. Res. Lett., 23(9), 933-936. Liu, C., P. Dong and Y. Shi (2017). Stress change from the 2015 Mw 7.8 Gorkha earthquake and increased hazard in the southern Tibetan Plateau, Phys. Earth Planet. In., 267, 1-8. Michas, G., P. Sammonds and F. Vallianatos (2015). Dy- namic Multifractality in Earthquake Time Series: Insights from the Corinth Rift, Greece, Pure Appl. Geophys., 172 (7), 1909-1921. Mahesh, P., B. Kundu, J. K. Catherine and V. K. Gahalaut (2011). Anatomy of the 2009 Fiordland Earthquake (Mw 7.8), South Island, New Zealand Geosci. Front., 2(1), 17-22. Mogi, K. (1962). Magnitude-frequency relation for elas- tic shocks accompanying fractures of various ma- terials and some related problems in earthquakes, Bulletin of the Earthquake Research Institute, Tokyo University, 40, 831-853. Mogi, K. (1981). Seismicity in western Japan and long- term earthquake forecasting. In "Earthquake Pre- diction") Maurice Ewing Series IV (D. W. Simpson and P. G. Richards, eds.) pp 43-51, American Geo- physical Union, Washington, D.C. Mondal, S. K., A. Borghi, P. N. S. Roy and A. Aoudia (2016). GPS, scaling exponent and past seismicity for seismic hazard assessment in Garhwal-Ku- maun, Himalayan region, Natural Hazards, 80 (2), 1349-1367. Mondal, S. K. and P. N. S. Roy (2016). Temporal multi- fractal pattern of seismicity in northwest Hi- malayan region, Journal of the Geological Society of India, 88 (5), 569-575. Mandelbrot, B. B. (1983). The Fractal Geometry of Na- ture, Freeman, New York. Mukhopadhyay, B., S. Dasgupta and S. Dasgupta (2004). Clustering of Earthquake Events in the Himalaya- Its Relevance to Regional Tectonic Set-up, Gond- wana Res., 7(4), 1242-1247. Okada, Y. (1992). Internal deformation due to shear and tensile faults in a half-space, Bull. Seismol. Soc. 15 SIGNIFICANCE OF Dc AND b-VALUE ANALYSIS BEFORE LARGE EARTHQUAKE Am., 82, 1018-1040. Oncel, A. O. and T. H. Wilson (2002). Space-Time corre- lations of seismotectonic parameters: Example from Japan and from Turkey preceding the Izmith earthquake, Bull. Seismol. Soc. Am., 92, 339-349. Oncel, A. O., I. Main, O. Alptekin and P. Cowie (1996). Temporal variations in the fractal properties of seismicity in the North Antolian fault zone be- tween 31 E and 41 E, Pure Appl. Geophys., 147, 147-159. Özturk, S. (2018). Earthquake hazard potential in the Eastern Anatolian Region of Turkey: seismotec- tonic b and Dc-values and precursory quiescence Z-value, Front. Earth Sci., 12(1), 215-236. Pandey, A. K., P. Chingtham, S. K. Prajapati, P. N. S. Roy and A. K. Gupta (2019). Recent seismicity rate fore- cast for North East India: An approach based on rate state friction law, J. Asian Earth Sci., 174, 167-176. Reyners, M. (1989). New Zealand seismicity 1964-87: An interpretation. N. Z. J. Geol. Geophys, 32, 307- 315. Roy, P.N.S. and A. Ram (2006). A correlation integral approach to the study of 26 January 2001 Bhuj earthquake, Gujarat India, Journal of Geodynam- ics, 41, 385-399. Reasenberg, P. A. and L. M. Jones (1989). Earthquake hazard after a mainshock in California, Science, 243, 1173-1176. Reasenberg, P. A. and R. W. Simpson (1992). Response of regional seismicity to the static stress change produced by the Loma Prieta earthquake, Science, 255, 1687-1690. Roy, P.N.S. and A. Ram (2006). A correlation integral approach to the study of 26 January 2001 Bhuj earthquake. Gujarat, India, J. Geodyn., 41(4), 385- 399. Roy, P. N. S. and S. K. Nath (2007). Precursory correla- tion dimensions for three great earthquakes, Cur- rent Science, 93(11), 1522-1529. Roy, P. N. S. and Ami, Padhi (2007). Multifractal analy- sis of earthquakes in the south-eastern Iran Bam region, Pure Appl. Geophys., 167, 2271-2290. Roy, P. N. S. and S. K. Mondal (2010). Identification of Seismicity Pattern for Some Destructive Earth- quakes, AGU Chapman Conference on “Complex- ity and extreme events in Geosciences” held in Hyderabad, during 15-19 February, 2010. Roy, P. N. S., S. K. Mondal and M. Joshi (2012). Seismic Hazards Assessment of Kumaun Himalaya and ad- jacent region, Nat. Hazards, 64(1), 283-297. Roy, P. N. S. and S. K. Mondal (2012a). Identification of Active Seismicity by Fractal Analysis for Under- standing the Recent Geodynamics of Central Hi- malaya, J. Geol. Soc. India, 79, 353-360. Roy, P. N. S. and S. K. Mondal (2012b). Multifractal anal- ysis of earthquakes in Kumaun Himalaya and its surrounding region. J. Earth Sys. Sci., 121(4), 1033- 1047. Roy, P. N. S. and D. K. Gupta (2015). Fractal Analyzer: A MATLAB Application for Multifractal Seismicity Analysis, Seismol. Res. Lett., 86 (5), 1424-1431. Scholtz, C. H. (1968a). The frequency-magnitude rela- tion of micro fracturing in rock and its relation to earthquakes, Bull. Seismol. Soc. Am., 58, 399-415. Scholz, C. H. (1968b). Experimental study of the frac- turing process in the brittle rock, J. Geophys. Res., 73, 1447-1454. Shan, B., X. Xiong, R. Wang, Y. Zheng and S. Yang (2013). Coulomb stress evolution along Xian- shuihe-Xiaojiang Fault System since 1713 and its interaction with Wenchuan earthquake, May 12, 2008, Earth Planet. Sci. Lett., 377-378, 199-210. Simpson, R. W. and P. A. Reasenberg, (1994). Earth- quake-induced static stress changes on central Cal- ifornia faults, in The Loma Prieta, California Earthquake of October 17, 1989. Tectonic processes and models edited by R. W. Simpson U., S. Geol. Surv. Prof. Pap., 1550-F, F55-F89. Smith, E. C. C. and F. J. Davey (1984). Joint hypocentre determination of intermediate depth earthquakes in Fiordland region, New Zealand, Tectonophysics, 104, 127-144. Sobolev, G. A., T. L. Chelidze, A. D. Zavyalov, L. V. Slav- ina and V. E. Nikoladze (1991). Maps of Expected Earthquakes Based on a Combination of Parame- ters, Tectonophysics, 193, 255–265. Stein, R. S., G. C. P. King and J. Lin (1992). Change in failure stress on the southern San Andreas fault system caused by the 1992 magnitude = 7.4 Lan- ders earthquake, Science, 258, 1328-1332. Stein, R. S. (1999). The role of stress transfer in earth- quake occurrence, Nature, 402, 605-609. Tomilin, N. G. and K. A. Voinov (1995), Technique and results of the rock burst prediction Proc Interna- tional Conf. Mechanics of Jointed and Faulted Rock, (Rotterdam: Balkema,), 955-959. Telesca, L., V. Cuomo, V. Lapenna and M. Macchiato (2001). Identifying space-time clustering proper- ties of the 1983–1997 Irpinia-Basilicata (southern Italy) seismicity, Tectonophys, 330(1-2), 93-102. Telesca, L., V. Lapenna and M. Macchiato (2003a). In- vestigating the time clustering properties in seis- micity of Umberia-Marche region (Central Italy), Chaos Solit. & Fract., 18(2), 202-217. Telesca, L., V. Lapenna and M. Macchiato (2003b). Spa- MONDAL ET AL. 16 tial variability of the time correlated behavior in Italian Seismicity, Earth Planet. Sci, Lett., 212(3- 4), 279-290. Teotia, S. S. and D. Kumar (2011). Role of multifractal analysis in understanding the preparation zone for large size earthquake in the North-Western Hi- malaya region, Nonlinear Proc. Geophys., 18, 111- 118. Toda, S. and B. Enescu (2011). Rate/state Coulomb stress transfer model for the CSEP Japan seismicity fore- cast, Earth Planets Space, 63, 171-185. Utsu, T. (1965). A method for determining the value of b in a formula log N = a - bM showing the mag- nitude-frequency relation for earthquakes (in Japanese), Geophysics, Bulletin of Hokkaido Uni- versity, 13, 99-103. Wyss, M. (1973). Towards a physical understanding of the earthquake frequency distribution, Geophys. J. Royal Astron. Soc., 31, 341-359. Xiong, X., B. Shan, Y. M. Zhou, S. J. Wei, Y. D. Li, R. J. Wang and Y. Zheng (2017). Coulomb stress trans- fer and accumulation on the Sagaing Fault, Myan- mar, over the past 110 years and its implications for seismic hazard, Geophys. Res. Lett., 44, 4781- 4789. Zamani, A. and M. Agh-Atabai (2009). Temporal char- acteristics of seismicity in the Alborz and Zagros regions of Iran, using a multifractal approach, J. Geodyn., 47, 271-279. *CORRESPONDING AUTHOR: Saroj Kumar MONDAL, CSIR-National Geophysical Research Institute, Hyderabad-500007, India; email: sarojbum@gmail.com © 2019 the Istituto Nazionale di Geofisica e Vulcanologia. All rights reserved 17 SIGNIFICANCE OF Dc AND b-VALUE ANALYSIS BEFORE LARGE EARTHQUAKE