TIWARI_finaonline:Layout 6 ANNALS OF GEOPHYSICS, 57, 5, 2014, S0544; doi:10.4401/ag-6513 S0544 Analysis of soil radon data in earthquake precursory studies Hari Prasad Jaishi1, Sanjay Singh1, Raghavendra Prasad Tiwari2, Ramesh Chandra Tiwari1,* 1 Mizoram University, Department of Physics, Aizawl, India 2 Mizoram University, Department of Geology, Aizawl, India ABSTRACT Soil radon data were recorded at two selected sites along Mat fault in Mi- zoram (India), which lies in the highest seismic zone in India. The study was carried out during July 2011 to May 2013 using LR-115 Type II films. Precursory changes in radon concentration were observed prior to some earthquakes that occurred around the measuring sites. Positive correla- tion was found between the measured radon data and the seismic activ- ity in the region. Statistical analysis of the radon data together with the meteorological parameters was done using Multiple Regression Method. Results obtained show that the method employed was useful for removing the effect of meteorological parameters and to identify radon maxima pos- sibly caused by seismic activity. 1. Introduction Variation of radon concentration in soil can give ev- idence on tectonic disturbances in the Earth’s crust [King 1986]. However, the changes in radon concentration are not only controlled by seismic activity but also influenced by meteorological parameters like soil moisture, rainfall, temperature, barometric pressure and relative humidity [Kraner et al. 1964, Pearson 1967, Singh et al. 1988, King and Minissale 1994, Virk et al. 2000, Zmazek et al. 2003, Walia et al. 2005]. These parameters change the physical characteristics of the soil and thus affect the soil radon variation caused by geophysical processes. Therefore, the- oretical and empirical algorithm is necessary for remov- ing the effects of meteorological parameters [Zmazek et al. 2003, Ramola et al. 2008, Choubey et al. 2009]. Northeast India is seismically one of the six most ac- tive regions of the world, the other five being Mexico, Taiwan, California, Japan and Turkey. It lies at the junc- tion of Himalayan arc to the north and Burmese arc to the east [Jaishi et al. 2014a, Singh et al. 2014]. The high seismicity in this zone is due to the collision tectonics be- tween the Indian plate and the Eurasian plate in the north and Subduction tectonics along the Indo-Myanmar Range (IMR) in the east [Dewey and Bird 1970]. Based on the distribution of epicentres, fault plane solutions and geotectonic features, northeast region can be divided into five seismotectonic zones which include Eastern Hi- malayan collision zone, Indo-Myanmar subduction zone, Syntaxis zone of Himalayan arc and Burmese arc (Mishmi Hills), Plate boundary zone of the Shillong plateau and Assam valley, and Bengal basin and plate boundary zone of Tripura- Mizoram fold belt (Figure 1). The Indo- Myanmar subduction zone is highly seismic zone extending from the Bay of Bengal to the Mishmi Article history Received February 19, 2014; accepted September 15, 2014. Subject classification: Surveys, measurements and monitoring, Geochemical data, Statistical analysis. Figure 1. Seismo-tectonic map of North East India showing epi- centres of damaging earthquakes (modified after Raghukant and Dash [2009]). Tectonic zones (zones A, B, C, D and E) and major thrusts (MBT-Main Boundary Thrust; MCT-Main Central Thrust; NT-Naga Thrust; DT-Disang Thrust) are also shown. Thick blue line represents the International boundary of North Eastern part of India bordering Myanmar and Bangladesh (not to scale) and red lines represents faults and thrusts. block in a N-S, NNE-SSW and NE-SW directions. The main discontinuities in this zone are Naga thrust and the Disang thrust. The seismic activity in the Bengal basin may be related to an intra plate activity whereas Tripura- Mizoram fold belt to plate- boundary activity. The main tectonics domains in this region are E-W trending Dauki fault, N-E trending Sylhet fault, NE-SW Hail- Hakula lineament, NNW Tista fault, Mat fault and Tuipui fault. Mat fault is the most prominent fault in Mi- zoram state [Jaishi et al. 2014b]. It obliquely cuts across the general north-south trend of the Indo-Burmese arc. It trends NW-SE (Figure 2) and is traceable across entire Mizoram on the satellite as well as on the geological maps. Mat River crosses the Mat fault and follows it for considerable distance. Most pronounced part of this fault is in Serchhip district along Serchhip-Thenzawl road. Hence this part has been selected for monitoring of radon anomaly and their possible correlation with the occurrence of earthquakes. Authors have done pri- mary analysis of soil radon and thoron anomalies as a precursor to earthquakes [Jaishi et al. 2013, 2014a, 2014b, Singh et al. 2014]. The present paper describes the mul- tiple regression method to differentiate those radon anomalies which are solely caused by seismic events and not by meteorological parameters. 2. Materials and methods For radon monitoring in soil, we have used LR- 115(II) cellulose nitrate detector manufactured by Kodak Pathe in France, a useful solid state nuclear track detector (SSNTD) for registering the alpha tracks gen- erating from radon gas. The LR-115 (II) films of size 3cm × 3cm was attached in the dosimeter with the sen- sitive side facing downwards for exposure. The dosime- ter has been developed at the Bhabha Atomic Research Centre (BARC), India. The dosimeters were buried in holes of depth of about 80 cm in the soil (Figure 3). The detector films were exposed for a period of 15 days, after that the exposed films were removed and rein- stalled with the new one in the dosimeters. The SSNTD’s are etched with 10% sodium hydroxide solu- tion in an etching bath at a temperature of 60°C for de- veloping the registered tracks. The etching time was set for 1 hour. The track etched method removes a bulk thickness of 4µm leaving a residual thickness of 8µm. As soon as the etching process is complete, films were washed in cold running water and then kept in distilled water for two hours. The detector films are pre-sparked at a voltage of 900V to fully develop the partially etched track holes using spark counter. The tracks are then counted at a voltage corresponding to plateau region of the counter (around 450V). 3. Results and discussion Radon concentration in soil gas have been meas- ured after every 15 days using LR-115(II) films during July 2011 to May 2013 together with the meteorologi- cal parameters viz.; rainfall, relative humidity, air tem- perature and barometric pressure at two locations (Mat Bridge and Tuichang) in Serchhip District, Mizoram (India). Numerous criteria have been used by different authors [Rannou 1990, Price et al. 1994, King et al. 1996, Fu et al. 2005] in order to calculate soil radon gas anom- alies which include mean plus ‘n’ (where n = 1, 2, 3,…) standard deviation (SD). Fleischer [1981] proposed the empirical relation between the magnitude (M) of the JAISHI ET AL. 2 Figure 3. Schematic description of the equipment used for radon measurements in soil.Figure 2. Location map of study area. 3 earthquake and the epicentre distance (D) within which the earthquake prediction zone can be manifested. (1) The details of the earthquakes (www.imd.gov.in) that occurred around the study areas fulfilling Equation (1) is summarised in Table 1. Multiple Linear Regression (MLR) Model Multiple regression analysis permits the study of several independent (predictor) variables for a given de- pendent (criterion) variable. For ‘n’ predictor variables the raw score regression equation is given by (2) where, is the predicted value of the dependent vari- able, a is the constant term and b1,b2,...,bn represents the regression coefficient of the independent variables X1,X2,...,Xn respectively. The drawback of the raw score regression is that the size of the b-coefficients is de- pendent on the scale of the predictor variables. There- fore, they cannot be compared directly with each other to determine which predictor variable has the most in- fluence on the criterion. By standardizing both the co- efficients and the variables we could make direct comparison among the coefficients [Yount 2006]. The standardized regression equation takes the form (3) where, b’s represents the standardized coefficients and Z’s represents the Z-scores. In the present investigation we perform MLR taking soil radon concentration (Rn) as the dependent variable and the meteorological pa- rameters viz. Rainfall (RF), relative humidity (RH), tem- perature (T) and pressure (P) as independent variables. The data used for analysis is reported in Table 2. Data analysis at Mat Bridge At Mat Bridge Rn1 is regressed with RF, RH, T and P (taken from Table 2). All the predictor variables men- tioned may not be a significant predictor of the de- pendent variable. To deal with this problem and to build a better model we perform backward regression method using SPSS statistical package. The backward regression method starts with all the predictor vari- ables, test the significance of each predictor variable using a t-test and then delete the variable (if any) that improves the model the most by being deleted. This process is repeated until the maximum number of steps is reached or no more independent variable qualifies for removal [Hocking 1976]. Table 3 shows the coefficients of the backward regression output. The backward re- gression output of the data has four distinct models (Model 1, Model 2, Model 3 and Model 4). The first model includes all the predictor variables (RF, RH, T and P). To check the significance of the predictor vari- ables a t-test is performed at 95% confidence interval (i.e., the alpha value is set at 0.05) and its corresponding probability is shown in sig. column of the table (see Table 3). If we look at the significance of each predic- tor variable in Model 1, two predictor variables (T and P) are non-significant predictors of the dependent vari- able as the p-value (sig.) is more than a-value (0.05) in both the cases. The backward regression method elim- inates one predictor variable at a time (the predictor variable with the highest p-value is deleted first). So, temperature (T) having the highest p-value (0.757) is re- moved from Model 1 and a second model is con- structed with RF, RH and P as independent variables. In Model 2, P is a non significant predictor (since the p- value is more than 0.05). Therefore, P is removed from the model and a third model is built with RF and RH as predictor variables. Model 3 tells us that RF is a non sig- nificant predictor (since the p-value is much more than 0.05). RF is therefore removed from the model and a fourth model is built with RH as a single predictor vari- ...Z Z Z ZY X X Xn n1 1 2 2b b b= + + +l ...Y a b X b X b Xn n1 1 2 2= + + + +l Yl . . . exp D M km1 66 10 0 48 = RADON IN EARTHQUAKE PRECURSORY STUDIES Day of occurrence Magnitude (M) Distance of epicentre (km) from Mat Bridge and Tuichang Region of occurrence 65th day 6.8 673 689 Sikkim-Nepal Border region 129th day 5.8 323 320 India-Myanmar Border region 353rd day 5.8 323 324 Phek (Nagaland), India 381st day 6.0 147 132 Myanmar 486th day 6.7 327 304 Myanmar 545th day 5.9 314 312 India-Myanmar Border region 597th day 5.2 174 187 India-Bangladesh Border region Table 1. Details of earthquakes that occurred around the study area during the period of investigation (earthquake data obtained from: www.imd.gov.in). JAISHI ET AL. 4 Days Radon conc. at Mat Bridge Rn1 (Tr.cm -2) Radon conc. at Tuichang Rn2 (Tr.cm -2) RF (mm) RH (%) T (°C) P (mbar) 15 814 687 131.7 92.4 23.93 888.29 30 528 476 184 93.7 23.8 888.38 45 682 305 258.5 92.7 23.9 890.78 60 242 215 172 91.8 24.37 889.91 75 342 123 138 90.5 23.66 890.83 90 413 441 58.6 85.6 24.46 893.56 105 126 75 23.8 80.4 22.42 894.69 120 129 72 0 70 21.16 895.81 135 143 88 0 69.5 21.12 896.44 150 154 170 0 72.6 19.01 895.73 165 93 74 0 63.5 19.11 896.01 180 153 204 20.3 80.6 16.49 896.81 195 69 76 0 69.2 16.85 894.5 210 130 102 0 62.6 19.31 895.81 225 65 55 7.3 61.8 21.1 892.67 240 91 55 46.9 69.6 21.15 892.68 255 97 67 0 61.4 24.45 894.36 270 107 81 112.3 74.9 22.09 894.28 285 41 11 152.4 87 22.33 891.61 300 104 48 118.8 81.8 23.07 891.49 315 339 126 56.8 86.5 24.81 889.16 330 590 90 211.1 91.3 23.67 887.36 345 121 84 262.9 93.4 22.79 886.05 360 145 504 158.8 90.6 23.93 887.07 375 666 215 93.6 92.7 23.45 887.09 390 286 192 265.5 93.13 24.23 887.63 405 148 67 200.4 93.12 23.49 889.84 420 241 190 203.7 92.1 23.53 891 435 562 185 160.2 91 23.57 893.07 450 214 208 171.1 89.1 21.84 894.45 465 260 419 0 85.5 22.25 896.27 480 256 304 109.9 88.2 19.68 895.98 495 325 384 6.1 84.4 19.87 895.84 510 180 171 0 77.2 18.37 896.53 525 319 328 0 65.7 17.74 895.3 540 448 447 0 69.8 15.39 895.77 555 198 204 0 70.5 18.02 896.14 570 186 235 0 69 21.03 895.66 585 162 205 3.3 66 21.38 894.42 600 158 157 0 63.6 23.68 894.99 615 377 335 5.6 73.3 23.76 893.35 630 415 371 21.5 74.3 25.19 891.81 645 99 80 39.9 80.2 22.54 892.69 660 101 95 324.4 93.6 20.2 890.18 675 378 307 124 87.7 23.43 888.58 Table 2. Radon concentrations and the meteorological parameters during the period of observation at Mat Bridge and Tuichang. 5 able. Clearly, RH is a significant predictor (p-value <0.05) of Rn1. To check how much of the criterion variability is accounted for by the predictor and the viability of the model-as-a-whole we examine the squared multiple R value and a significant F-ratio (p<0.05) respectively, which is reported in Table 4. In the model summary of Table 4, R-square and adjusted R-square is re- ported. The adjusted R-square is the measure of the shrinkage if we were to apply this model to another sample (in other words the amount of predictive loss that we would observe). The value of adjusted R- square in model 4 is 0.198 i.e. 19.8% of the variation in radon concentration is accounted by RH. By reducing the number of predictors from 4 to 1 the adjusted R- square dropped from 0.240 to 0.198, a change of 0.042 (or a little over 4%) which is good. We do not lose much R-square by dropping 3 of the 4 predictor vari- ables. Also the standard error of estimates increase from 162.51 to 166.99 (only 2.68% increase) which is not much. To test the significance of the model-as-a- whole an F-ratio (MSreg./MSres.) is calculated for each model and the probability of the computed F-ratio is displayed in sig. column of Table 4. If we compare all the models, F-ratio is largest in Model 4 showing bet- ter ratio of regression to noise. Reducing the number of predictors from 4 to 1 reduces the p-value of F-ratio from 0.004 to 0.001 indicating a more significant and a better model. The raw score regression equation for Model 4 can be written as (4). . ( )Rn RH371 60 7 8651 =- +l RADON IN EARTHQUAKE PRECURSORY STUDIES Model Parameters Unstandardized coefficients Standardized t Sig. (p- value) B Std. error coefficients Beta 1 (Constant) 15506.483 12574.905 1.233 .225 RF -1.069 .513 -.538 -2.086 .043 RH 11.267 3.994 .666 2.821 .007 T 4.115 13.222 .054 .311 .757 P -18.091 13.854 -.309 -1.306 .199 2 (Constant) 17546.232 10612.758 1.653 .106 RF -1.090 .503 -.549 -2.169 .036 RH 11.414 3.922 .675 2.910 .006 P -20.287 11.791 -.346 -1.721 .093 3 (Constant) -707.813 284.934 -2.484 .017 RF -.708 .461 -.356 -1.535 .132 RH 12.804 3.926 .757 3.261 .002 4 (Constant) -371.602 185.026 -2.008 .051 RH 7.865 2.283 .465 3.445 .001 Model Sum of squares df Mean square F Sig. R square Adjusted R square Std. error of estimates 1 Regression 473551.25 4 118387.81 4.48 .004 .310 .240 162.512Residual 1056409.54 40 26410.23 Total 1529960.80 44 2 Regression 470993.53 3 156997.84 6.07 .002 .308 .257 160.712Residual 1058967.27 41 25828.47 Total 1529960.80 44 3 Regression 394529.07 2 197264.53 7.29 .002 .258 .223 164.420Residual 1135431.72 42 27034.089 Total 1529960.80 44 4 Regression 330863.50 1 330863.50 11.9 .001 .216 .198 166.990Residual 1199097.29 43 27885.98 Total 1529960.80 44 Table 4. Model summary and analysis of variance (Mat Bridge). Table 3. Coefficients of the Backward regression output at Mat Bridge. and the corresponding standardized equation would be (5) where ZRn'1 is the standardized value (Z-score) of Rn'1 and ZRH is the Z-score of RH. We calculate the Z- scores of the measured radon concentration (ZRn1) and relative humidity (ZRH) during the entire period of ob- servation and uses Equation (5) to find the standardized predicted value (ZRn'1). When the difference (i.e. the standardized residuals) between ZRn1 and ZRn'1 is more than 1SD, we denote it as a possible radon anomaly. The graphical representation of the standardized meas- ured radon concentration and residuals during the ob- servation period is shown in Figure 4. The radon maximum observed on 45th day of the observation period (Figure 4a) was ZRn1 = 2.263 Tracks/cm2. The difference (ZRes.) between the meas- ured and the predicted value was, 1.943 Tracks/cm2 (close to 2SD; Figure 4b). This radon maximum could be correlated to the earthquake of M6.8 which oc- curred on 65th day of the observation period (i.e. 20 days after the anomaly was observed). The second radon peak was observed on 330th day. The difference between the measured and the predicted value was recorded to be ZRes.=1.458 Tracks/cm2, which is greater than 1SD. The second peak may be correlated to the earthquake of M5.8 which occurred on 353rd day (23 days after the anomaly was observed). The third anomalous change in radon concentration was recorded on 375th day. The difference between the measured and the predicted value during this period was recorded to be 1.847 Tracks/cm2 (much greater than 1SD). An earthquake of M6.0 occurred only 6 days after the anomaly was observed. So, the third radon maximum may be correlated to the mentioned M6.0 quake. The fourth maximum was recorded on 435th day. The difference in the concentration was recorded to be 1.305 Tracks/cm2 ( just above 1SD level). Though an earthquake of M6.7 occurred several days after the anomaly was observed, it was quite difficult to directly correlate the radon maximum to the mentioned event as the time interval between the anomaly observed and the event was quite large (51 days). In other words, this radon maximum may be treated as a false anomaly . ( )ZRn ZRH0 4651 =l JAISHI ET AL. 6 Model Parameters Unstandardized coefficients Standardized t Sig. (p- value) B Std. error coefficients Beta 1 (Constant) 7180.246 10796.340 .665 .510 RF -1.212 .440 -.753 -2.755 .009 RH 10.783 3.429 .787 3.145 .003 T -3.601 11.351 -.059 -.317 .753 P -8.577 11.894 -.180 -.721 .475 2 (Constant) 5395.113 9112.143 .592 .557 RF -1.194 .431 -.742 -2.767 .008 RH 10.654 3.367 .777 3.164 .003 P -6.656 10.123 -.140 -.657 .515 3 (Constant) -593.630 237.506 -2.499 .016 RF -1.069 .384 -.664 -2.780 .008 RH 11.110 3.273 .810 3.395 .002 Table 5. Coefficients of the Backward regression output at Tuichang. Figure 4. Variation of (a) Standardized measured radon concentra- tion and (b) Standardized residuals during the observation period at Mat Bridge. The solid vertical lines represent the earthquakes along with their magnitudes (M). 7 which may have occurred due to variation in relative humidity. For the radon maximum that occurred on 540th day, the residual was 1.620 Tracks/cm2 which is well above the 1SD level. This may be treated as a pre- cursory signal before the event of M5.9 which occurred on 545th day (5 days after the anomaly was observed). Data analysis at Tuichang As in case of Mat Bridge we perform backward re- gression taking Rn2 (measured radon concentration at Tuichang) as dependent variable and RF, RH, T and P as independent variables (data taken from Table 2). The SPSS print out of the results (coefficients) is shown in Table 5. The output of the backward regression method produces three models namely Model 1, Model 2 and Model 3. Model 1 includes all the predictor variables. Temperature (T) is having a p-value of 0.753 which is greater than the a-value (0.05); therefore T is removed from Model 1 to construct Model 2 with RF, RH and P as independent variables. It is clear from the sig. column of Table 5 that P is a non-significant predictor (since the p- value >0.05). Consequently, P is removed from the Model and a third Model is established with two predic- tor variables (RF and RH). Both RF and RH are signifi- cant predictors since the p-value in both the cases is less than 0.05. The model summary and the analysis of vari- ance are given in Table 6. To verify the significance of the model-as-a-whole, first we look at the adjusted R- square value. The adjusted R-square value is 0.178 (i.e. 17.8 % of the variation in radon concentration is ex- plained by RF and RH). Reducing the number of param- eters from 4 to 2 does not reduce the adjusted R-square value. In fact, the value increases from 0.148 to 0.178 (i.e. 16.86% increase). Second, there is no increment in stan- dard error of estimates by reducing the predictor vari- ables which is a good indication that Model 3 is better. Third, the F-ratio is largest in Model 3 (i.e. a better ratio of regression to noise) and the p-value of F-ratio is least. In conclusion Model 3 is the most significant and a bet- ter model. The raw score and the Z-score regression equations for Model 3 can be written as where ZRn'2, ZRF and ZRH are the standardized value of Rn'2, RF and RH respectively in units shown in Table 2. The plot of standardized measured radon concen- tration and residuals during the observation period at Tuichang is shown in Figure 5. The first radon maximum was recorded on 90th day of the observation period (Fig- ure 5a). During this period the difference between the 593.63 1.069 ( ) 11.110 ( ) . ( ) . ( ) Rn RF RH ZRn ZRF ZRH0 664 0 810 2 2 =- - + =- + l l RADON IN EARTHQUAKE PRECURSORY STUDIES Model Sum of squares df Mean square F Sig. R square Adjusted R square Std. error of estimates 1 Regression 226690.92 4 56672.73 2.91 .033 .225 .148 139.526Residual 778710.32 40 19467.75 Total 1005401.24 44 2 Regression 224731.88 3 74910.62 3.93 .015 .224 .167 137.988Residual 780669.35 41 19040.71 Total 1005401.24 44 3 Regression 216501.64 2 108250.82 5.76 .006 .215 .178 137.052Residual 788899.60 42 18783.32 Total 1005401.24 44 Table 6. Model summary and analysis of variance (Tuichang). Figure 5. Variation of (a) Standardized measured radon concentra- tion and (b) Standardized residuals during the observation period at Tuichang. The solid vertical lines represent the earthquakes along with their magnitudes (M). (6) (7) measured and the predicted value of radon concentra- tion was, ZRes. = 1.066 Tracks/cm2 ( just above 1SD; Fig- ure 5b). This radon maximum was quite difficult to directly correlate to any seismic event (as the time inter- val between the radon maximum and the closest earth- quake M5.8 was 39 days). So, the first radon maximum may be treated as a false anomaly possibly may have caused by variation in meteorological parameters (RF and RH). The next increment in radon concentration was recorded on 360th day. The difference between the meas- ured and the predicted value was recorded to be 1.902 Tracks/cm2 (close to 2SD). This peak may be correlated to the event of M6.0 which occurred 21 days after the anomaly was observed. Third change in radon concen- tration was observed on 495th day. But it was quite diffi- cult to correlate this peak to the event of M6.7 as during this period the difference between the measured and the predicted value was quite low (ZRes. = 0.338 Tracks/cm2; less than 1SD). Next impulsive change was recorded on 540th day of the observation period. The difference be- tween the measured and predicted value during this pe- riod was 1.934 Tracks/cm2 (close to 2SD). The mentioned peak may be treated as a precursor signal before the earth- quake of M5.9 which occurred 5 days after the signal. 4. Conclusion We have performed statistical analysis of the radon data collected from July 2011 to May 2013 at two sites (Mat bridge and Tuichang) along Mat Fault, the most prominent fault in Mizoram. The multiple regression analysis of the radon data with the meteorological pa- rameters provides useful evidence to extract a radon anomaly possibly caused by seismic event. The results obtained shows that at Mat Bridge about 57% of the anomaly observed could be correlated to the seismic events whereas at Tuichang 28.6% could be correlated, indicating that the radon activity in soil gas is mostly controlled by geophysical process. The precursor time between the observed radon anomalies and the earth- quakes varies from 5-23 days. High percentage of cor- relation between radon anomalies and the earthquakes at Mat Bridge indicates that the area under investiga- tion is seismically active and would be useful to carry out further studies. Several spike-like radon anomalies observed prior to the earthquakes in the present inves- tigation may provide soil radon database and the pre- liminary trend of their response to earthquake activity in the region. The Backward regression method applied gives valuable information about the meteorological parameter(s) that have the strongest effect on radon concentration. However, long term and continuous data recording is necessary to confirm a relation be- tween soil radon activity and earthquakes. Acknowledgements. Authors are grateful to the Ministry of Earth Sciences, Govt. of India, New Delhi for their financial sup- port to this work in the form of a Major project vide Sanction order No. MoES/ P.O.(Seismo) /1(67)/2009, Dated 16-06-2010. Data and sharing resources The earthquake catalog used in this study was ob- tained from: http://www.imd.gov.in/section/seismo/dynamic/wel come.htm References Choubey, V.M., N. Kumar and B.R. Arora (2009). Pre- cursory signatures in the radon and geohydrologi- cal borehole data for M4.9 Kharsali earthquake of Gharwal Himalaya, Sci. Tot. Environ., 407, 5877- 5883. Dewey, J.F., and J.M. Bird (1970). Mountain belts and the new global tectonics, J. Geophys. Res., 75, 2625- 2647. Fleischer, R.L. (1981). Dislocation model for radon re- sponse to distant earthquakes, Geophys. Res. Lett., 84, 77-480. Fu, C.C., T.F. Yang, V. Walia and C.H. Cheng (2005). Reconnaissance of soil gas composition over the buried fault and fracture zone in Southern Taiwan, Geochem. J., 39, 427-439. Hocking, R.R. (1976). The analysis and selection of variables in Linear regression, Biometrics, 32, 1-49. Jaishi, H.P., S. Singh, R.P. Tiwari and R.C. Tiwari (2013). Radon and thoron anomalies along Mat fault in Mi- zoram, India, J. Earth Syst. Sci., 122, 1507-1513. Jaishi, H.P., S. Singh, R.P. Tiwari and R.C. Tiwari (2014a). Correlation of radon anomalies with seismic events along Mat fault in Serchhip District, Mizoram, India, Appl. Radiat. Isot., 86, 79-84. Jaishi, H.P., S. Singh, R.P. Tiwari and R.C. Tiwari (2014b). Temporal variation of soil radon and thoron con- centrations in Mizoram (India), associated with earthquakes, Nat. Hazards, 72, 443-454. King, C.Y. (1986). Gas Geo Chemistry Applied to Earth- quake Prediction: An Overview, J. Geophys. Res., 91, 12269-12281. King, C.Y., and A. Minissale (1994). Seasonal variability of soil-gas radon concentration in central California, Radiat. Meas., 23, 683-692. King, C.Y., B.S. King, W.C. Evans and W. Zhang (1996). Spatial radon anomalies on active faults in Califor- nia, Appl. Geochem., 11, 497-510. Kraner, H.W., G.L. Schroeder and R.D. Evans (1964). Measurements of the effects of atmospheric vari- ables on radon-222 flux and soil gas concentration, In: J.A.S and W.M. Adams (eds.), The Natural Radi- ation and Environment, Symposium Proceedings JAISHI ET AL. 8 9 (Houston, Texas, April 10-13, 1963), Lower, Univ. Chicago Press, Chicago, 191-215. Pearson, J.E. (1967). Natural Environmental Radioac- tivity from Radon-222, In: U.S Public Health Serv- ice Publication No. 999-RH-26, U.S. Dept. of Health, Education and Welfare, Washinton, D.C., 31 pp. Price, J.G., J.G. Rigby, L. Christensen, R. Hess, D.D. La- Pointe, A.R. Ramelli, M. Desilets, R.D. Hopper, T. Kluesner and S. Marshall (1994). Radon in outdoor in Nevada, Health Physics, 66, 433-438. Raghukanth, S.T.G., and S.K. Dash (2009). Determinis- tic seismic scenarios for North East India, J. Seis- mol., 14, 143-167. Ramola, R.C., Y. Prasad, G. Prasad, S. Kumar and V.M. Choubey (2008) Soil-gas radon as seismotectonic in- dicator in Garhwal Himalaya, Appl. Radiat. Isot., 66, 1523-1530. Rannou, A. (1990). The bare detector and results of in- door radon Survey in France, In: Proceedings of the International Workshop on Radon monitoring in Radioprotection, Environmental Radioactivity and Earth Sciences (ICTP, Trieste, Italy, April 3-14, 1989), World Scientific publication, 145-222. Singh, M., R.C. Ramola, N.P. Singh, S. Singh and H.S. Virk (1988). Influence of meterological parameters on soil gas radon, J. Assoc. Explor. Geophys., 9, 85-90. Singh, S., H.P. Jaishi, R.P. Tiwari and R.C. Tiwari (2014). Variations of soil radon concentrations along Chite Fault in Aizawl district, Mizoram, India, Radiat. Prot. Dosim.; doi:10.1093/rpd/ncu221. Virk, H.S., V. Walia, A.K. Sharma, N. Kumar and R. Kumar (2000). Correlation of radon anomalies with microseismic events in Kangra and Chamba valleys of N-W Himalaya, Geofis. Int., 39, 221-227. Walia, V., H.S. Virk, T.F. Yang, S. Mahajan, M. Walia and B.S. Bajwa (2005). Earthquake Prediction Studies Using Radon as a Precursor in N-W Himalayas, India: A Case Study, TAO, 16, 775-804. Yount, W.R. (2006). Research Design and Statistical analysis in Christian Ministry, 4th Edition, Forth Worth, Texas: Southwest Baptist Theological Sem- inary, 26.1-26.12. Zmazek, B., L. Todorovski, S. Dmeroski, J. Vaupotic and I. Kobal (2003). Application of decision trees to the analysis of soil radon data for earthquake pre- diction, Appl. Radiat. Isot., 58, 697-706. *Corresponding author: Ramesh Chandra Tiwari, Mizoram University, Department of Physics, Aizawl, India; email: ramesh_mzu@rediffmail.com. © 2014 by the Istituto Nazionale di Geofisica e Vulcanologia. All rights reserved. RADON IN EARTHQUAKE PRECURSORY STUDIES << /ASCII85EncodePages false /AllowTransparency false /AutoPositionEPSFiles false /AutoRotatePages /None /Binding /Left /CalGrayProfile (Dot Gain 20%) /CalRGBProfile (sRGB IEC61966-2.1) /CalCMYKProfile (U.S. Web Coated \050SWOP\051 v2) /sRGBProfile (sRGB IEC61966-2.1) /CannotEmbedFontPolicy /Warning /CompatibilityLevel 1.3 /CompressObjects /Tags /CompressPages true /ConvertImagesToIndexed true /PassThroughJPEGImages true /CreateJobTicket false /DefaultRenderingIntent /Default /DetectBlends true /DetectCurves 0.1000 /ColorConversionStrategy /LeaveColorUnchanged /DoThumbnails false /EmbedAllFonts true /EmbedOpenType false /ParseICCProfilesInComments true /EmbedJobOptions true /DSCReportingLevel 0 /EmitDSCWarnings false /EndPage -1 /ImageMemory 1048576 /LockDistillerParams true /MaxSubsetPct 100 /Optimize false /OPM 1 /ParseDSCComments true /ParseDSCCommentsForDocInfo true /PreserveCopyPage true /PreserveDICMYKValues true /PreserveEPSInfo true /PreserveFlatness true /PreserveHalftoneInfo false /PreserveOPIComments false /PreserveOverprintSettings true /StartPage 1 /SubsetFonts true /TransferFunctionInfo /Apply /UCRandBGInfo /Preserve /UsePrologue false /ColorSettingsFile (None) /AlwaysEmbed [ true /AndaleMono /Apple-Chancery /Arial-Black /Arial-BoldItalicMT /Arial-BoldMT /Arial-ItalicMT /ArialMT /CapitalsRegular /Charcoal /Chicago /ComicSansMS /ComicSansMS-Bold /Courier /Courier-Bold /CourierNewPS-BoldItalicMT /CourierNewPS-BoldMT /CourierNewPS-ItalicMT /CourierNewPSMT /GadgetRegular /Geneva /Georgia /Georgia-Bold /Georgia-BoldItalic /Georgia-Italic /Helvetica /Helvetica-Bold /HelveticaInserat-Roman /HoeflerText-Black /HoeflerText-BlackItalic /HoeflerText-Italic /HoeflerText-Ornaments /HoeflerText-Regular /Impact /Monaco /NewYork /Palatino-Bold /Palatino-BoldItalic /Palatino-Italic /Palatino-Roman /SandRegular /Skia-Regular /Symbol /TechnoRegular /TextileRegular /Times-Bold /Times-BoldItalic /Times-Italic /Times-Roman /TimesNewRomanPS-BoldItalicMT /TimesNewRomanPS-BoldMT /TimesNewRomanPS-ItalicMT /TimesNewRomanPSMT /Trebuchet-BoldItalic /TrebuchetMS /TrebuchetMS-Bold /TrebuchetMS-Italic /Verdana /Verdana-Bold /Verdana-BoldItalic /Verdana-Italic /Webdings ] /NeverEmbed [ true ] /AntiAliasColorImages false /CropColorImages true /ColorImageMinResolution 150 /ColorImageMinResolutionPolicy /OK /DownsampleColorImages true /ColorImageDownsampleType /Bicubic /ColorImageResolution 300 /ColorImageDepth -1 /ColorImageMinDownsampleDepth 1 /ColorImageDownsampleThreshold 1.10000 /EncodeColorImages true /ColorImageFilter /DCTEncode /AutoFilterColorImages true /ColorImageAutoFilterStrategy /JPEG /ColorACSImageDict << /QFactor 0.15 /HSamples [1 1 1 1] /VSamples [1 1 1 1] >> /ColorImageDict << /QFactor 0.15 /HSamples [1 1 1 1] /VSamples [1 1 1 1] >> /JPEG2000ColorACSImageDict << /TileWidth 256 /TileHeight 256 /Quality 30 >> /JPEG2000ColorImageDict << /TileWidth 256 /TileHeight 256 /Quality 30 >> /AntiAliasGrayImages false /CropGrayImages true /GrayImageMinResolution 150 /GrayImageMinResolutionPolicy /OK /DownsampleGrayImages true /GrayImageDownsampleType /Bicubic /GrayImageResolution 300 /GrayImageDepth -1 /GrayImageMinDownsampleDepth 2 /GrayImageDownsampleThreshold 1.10000 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages true /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict << /QFactor 0.15 /HSamples [1 1 1 1] /VSamples [1 1 1 1] >> /GrayImageDict << /QFactor 0.15 /HSamples [1 1 1 1] /VSamples [1 1 1 1] >> /JPEG2000GrayACSImageDict << /TileWidth 256 /TileHeight 256 /Quality 30 >> /JPEG2000GrayImageDict << /TileWidth 256 /TileHeight 256 /Quality 30 >> /AntiAliasMonoImages false /CropMonoImages true /MonoImageMinResolution 1200 /MonoImageMinResolutionPolicy /OK /DownsampleMonoImages true /MonoImageDownsampleType /Bicubic /MonoImageResolution 1200 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.08250 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict << /K -1 >> /AllowPSXObjects false /CheckCompliance [ /None ] /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox true /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile (None) /PDFXOutputConditionIdentifier () /PDFXOutputCondition () /PDFXRegistryName (http://www.color.org) /PDFXTrapped /Unknown /CreateJDFFile false /SyntheticBoldness 1.000000 /Description << /ENU (Use these settings to create PDF documents with higher image resolution for high quality pre-press printing. The PDF documents can be opened with Acrobat and Reader 5.0 and later. These settings require font embedding.) /JPN /FRA /DEU /PTB /DAN /NLD /ESP /SUO /NOR /SVE /KOR /CHS /CHT /ITA >> >> setdistillerparams << /HWResolution [2400 2400] /PageSize [595.000 842.000] >> setpagedevice