Time-dependent seismicity in China
Changyuan Qin*, E.E. Papadimitriou, B.C. Papazachos, G.F. Karakaisis
Laboratory of Geophysics, University of Thessaloniki, GR-54006, Thessaloniki, Greece
Abstract
Precise zonation of the territory of China has been performed based on the active known faults, type of faulting and seismicity level. One hundred and forty seven seismogenic regions were de®ned, forming 10 larger seismic areas, and the seismotectonic characteristics in each one of them were investigated in detail. After checking for data accuracy and completeness of the shallow earthquakes (h#60 km), the
regional time and magnitude predictable modelwas applied and the model parameters were estimated. Based on the model applicability in the studied area, probabilities for the occurrence of strong (M$6.0) earthquakes during the next 10 years were calculated for each seismogenic region. Statistical tests have been used proving the superiority of the model in comparison with the time independent one, as well as in comparison with the actual earthquake occurrence.q2001 Elsevier Science Ltd. All rights reserved.
Keywords: Time-dependent seismicity; Seismotectonics; Probabilities; China
1. Introduction
Many earthquake recurrence models have been proposed so far. However, only a few of them have been accepted that satisfactorily describe the seismic activity. The most promi-nent of them are seismic gaps (Imamura, 1928; Fedotov, 1965; Mogi, 1985) and time and slip-predictable models (Bufe et al., 1977; Shimazaki and Nakata, 1980).
Earlier this century, Omori (1907) and Imamura (1928) had explicitly stated the existence of seismic gaps. Fedotov (1965) introduced the seismic cycle concept, which means that the probability for the occurrence of a mainshock in a certain fault increases with the time elapsed since the occur-rence of the previous mainshock in this fault. The earlier long-term prediction was made by Kelleher et al. (1973) and McCann et al. (1979) based on the seismic gap concept. The 1978 Oaxaca earthquake in Mexico (M7.7, 16.5N 96.5W) was successfully forecasted by this pattern (Ohtake et al., 1981; Mogi, 1985). The seismic gaps, as Mogi (1985) de®ned later, can be divided into two categories. (a) If great earthquakes are phenomena whereby crustal stress gradu-ally increases over a wide range, strain is accumulated, a rupture occurs when the stress has reached the limit, and stress and strain are released all at once, then places where no great earthquake has occurred for a long time can be regarded as possible sites for the next great earthquake. (b) If the activity of small earthquakes falls off in the focal region prior to a large earthquake, the appearance of
such a temporary inactive period is a warning of the occur-rence of a large earthquake.
The time- and slip-predictable model originally proposed by Shimazaki and Nakata (1980) can be summarized as, (a) if the ®nal stress varies in time while the initial stress remains constant, the regularity makes it possible, in prin-ciple, to predict the occurrence time of the coming earth-quake (so-called ªtime-predictable recurrenceº), (b) if the ®nal stress remains constant and the initial stress varies, the coseismic slip produced by the coming earthquake can be determined prior to the event (so-called ªslip-predictable recurrenceº). Shimazaki and Nakata (1980) presented three example sequences of large thrust±fault earthquakes in Japan and found a regularity showing that the larger the earthquake, the longer the following quiet period. Bufe et al. (1977) presented a model of seismic slip and recurrence intervals for a segment fault in northern California and proposed that a given time interval is proportional to the amount of displacement in the preceding earthquake. Sykes and Quittmeyer (1981) found that data on the geome-try, seismic moment and repeat time of large shocks of both the strike±slip and convergent types agree better with the time-predictable model of earthquake recurrence than with the slip-predictable model.
The seismic gap concept can only give information on the spatial domain. It is insensitive to the time domain. The time-predictable model is related with the displacement produced by the former event, in a certain fault, which can be indicative for the expected time of the next one, since tectonic loading can be assessed. However, as of the late 1980's, the time-dependent character of the occurrence
1367-9120/01/$ - see front matterq2001 Elsevier Science Ltd. All rights reserved.
PII: S 1 3 6 7 - 9 1 2 0 ( 0 0 ) 0 0 0 1 9 - 5 * Corresponding author.
of earthquakes has quantitatively been studied and success-fully been used in long-term earthquake prediction (Papa-zachos, 1988a,b, 1989, 1992). The author explicitly expressed that the time interval elapsed from the last main-shock as well as the size of the following mainmain-shock depends on the magnitude of previous mainshock, introdu-cing the ªregional time and magnitude predictable modelº. This model holds for a ªseismogenic regionº, that is, for a relatively small part of the lithosphere which includes the rupture zone (fault, deformation volume) of the largest mainshock of this part of the lithosphere as well as second-ary faults where smaller mainshocks are generated.
Papazachos and Papaioannou (1993) further improved the model by suggesting the following two relations which give the occurrence time,Tt, and the surface wave
magni-tude, of the following mainshock,Mf, as a function of: the
magnitude of the smallest mainshock considered,Mmin, the
magnitude of the preceding mainshock,Mp, and the annual
moment rate, mo, for each seismogenic region. These two
relations have the following form:
logT bMmin 1cMp1dlogm01q
Mf BMmin 1CMp1Dlogm01m
1
whereb, c, d, q, B, C, D, mare parameters to be calculated.
Nowadays this model has been used worldwide (Karakaisis, 1993, 1994a,b,c; Panagiotopoulos, 1993, 1994, 1995; Papa-dimitriou, 1993, 1994a,b; Papadimitriou and Papazachos, 1994; Papazachos et al., 1994a,b,c, 1997a,b; Tsapanos and Papazachos, 1994; Papazachos and Papadimitriou, 1996).
Many seismologists also found that the seismic activity in China follows more or less the same pattern. So far as we know, the earlier time dependent seismic phenomena were used to analyze the seismic activity in large seismic region (Xiansuihe fault) in southwest China as early as 1983 (Han and Huang, 1983; Wen et al., 1988; Wen, 1989, 1990, 1993, 1995). Li et al. (1994) concluded that occurrences in groups, are the fundamental feature of the strong earthquakes in China. Tsapanos and Papazachos (1994) and Papazachos et al. (1997b) did a quantitative analysis of time dependent seismicity in China. Qin et al. (1999b) have showed the validation of the regional time- and magnitude-predictable model throughout China. Qin (2000) carried out a lot of tests to show priority of the time- and magnitude-predictable model over the time independent model. The present paper, with the help of a large amount of seismotectonic, geological, and seismic information, concerns the applica-tion of this model based on the relaapplica-tions shown in Eq. (1) to investigate the temporal dependence of the mainshock occurrence in China by applying the regional time and C. Qin et al. / Journal of Asian Earth Sciences 19 (2001) 97±128
magnitude predictable model. Moreover, based on the model applicability, probabilities for the occurrence of strong (M$6.0) mainshocks in each seismogenic region were estimated, contributing in that way to seismic hazard assessment in the studied area.
2. Method
The precise zonation, the calculation of the seismic moment rate, and the declustering of the data, are prerequi-sites for the model application. The separation of a seismic area into seismogenic regions is based on certain seismo-tectonic and geomorphologic criteria. Such criteria are: spatial clustering of seismicity, topography variations (grabens, troughs, etc.), dimensions of rupture zones of large earthquakes (surface fault traces, distribution of after-shock volumes of recent events, tsunamigenic sources, well documented focal areas of historical earthquakes) and evidence for interactions between seismic events (Papaza-chos et al., 1997a). The seismic moment rate, expressed by the term logmo, which takes the tectonic loading into
consid-eration, is responsible in reducing the in¯uence of the tectonic distribution so that the model can be used in very complicated tectonic environment (Fig. 1).
2.1. Zonation
Although it was pointed out that the zonation is not criti-cal for the results of the model, accurate zonation will improve them (Papazachos et al., 1997a; Papazachos and Papadimitriou, 1997). The characteristic property of a
seis-mogenic region is the interaction among its faults during the important seismic excitations (redistribution of stress, etc.). Therefore, zonation in the present case is the procedure of de®ning, as accurately as possible, the boundaries of the seismogenic regions. The territory of China has been divided into 10 seismic areas, on the basis of the gross seismotectonic and geological properties and on and fault plane solutions' distribution.
Area A constitutes the large Himalayan thrust fault and is the frontier of the Eurasian plate which is pushed northward by the Indian plate (Fig. 2). Due to the northeastward intru-sion of the Indian plate, the Himalayan front in southwestern China is characterized completely by compression and is undergoing rapid uplift (Molnar and Lyon-Caen, 1989; Gao, 1996). This suggests that the fault planes dip north-wards. The overall pattern of fault plane solutions is compa-tible with the subduction of the Indian plate in front of the Himalaya Frontal Thrust. The meanPaxis of this area has a NNE direction almost normal to the Himalayan arc (Molnar and Lyon-Caen, 1989). The seismicity in this place mainly re¯ects the compressive stress between the two plates.
Area B is dominated by normal faulting going parallel to Area A. The area along the Indus±Tsangpo Suture is controlled by east±west extensive force. The mean T-axis has east±west direction. It is considered as a transition area between the thrust motion of the Himalayas and the strike± slip-motion of the central Qinghai±Tibet plateau. These normal faults are located at higher altitude (.5000 m), which gradually change to strike±slip faults with the decrease of the level northwards (Molnar and Lyon-Caen, 1989) (Fig. 3).
C. Qin et al. / Journal of Asian Earth Sciences 19 (2001) 97±128
Fig. 3. The normal faulting type area (from Qin et al., 2000).
In the Qinghai±Tibet plateau, three areas (C, D, and E) dominated by strike±slip motion (Fig. 4), constitute the special `creep' motion channel (Qin et al., 2000). The left-lateral Xianshuihe, Shanjiang (Anninghe, Zemuhe, Xiao-jiang) faults and the right-lateral Red River and Jinsajiang faults in Area C contrast with the other two regions. The crustal motion caused by the seismic activity also showed their difference. Area C moves in a NW-SE direction and at the same time rotates clockwise (Xu and Deng, 1996; England and Molnar, 1997; Qin et al., 2000). Area C behaves like a special ªriverº which ª¯owsº southeastward. The left-lateral Xiansuihe, Anninghe, Zemuhe, Xiaojiang faults, and the right-lateral Jinsajiang and Red River faults constitute the diamond clockwise rotating block (Fig. 1). The large component of the clockwise rotation makes it different from its adjacent areas. Area D, which is located in the heart of the Qinghai±Tibet plateau, mainly moves northeastward and is characterized by (right or left) lateral strike±slip faulting. The Altun fault marks its northern edge, whereas the Karakorum fault is the western border. The Kunlun fault passes through this region (Fig. 1). The northeast horizon-tal motion is the main reason for its separation from Area C. Area E, as the continuation of the special ªriverº, runs eastward
along the southernmost part of the southeast China. For the same reason mentioned for Area D, the eastward horizontal motion pattern in this area contrasts that of Area C (Fig. 4). The northern belt is supposed to be developed from the eastern boundary of the Tamir basin to the Tianshan thrust faulting area (Area H, Fig. 5) (Avouac et al., 1993). The whole thrust belt is considered as a resisting zone, which retards the Indian pushing force. These two areas are mainly distinguished by their seismic activity. The large Qilianshan thrust (Gaudemer et al., 1995) fault passes through Area F, which is characterized by intense seismic activity. The large Gulang (M7.7, 23 May 1927) and Changma (M7.5, 25 December 1932) earthquakes have occurred here. It is believed that it forms the northeastern bank of the special ªriverº, whereas Area H constitutes the thrust fault belt in the northern part of Qinghai±Tibet Plateau.
Area G is developed around the Ordos block in North China that is considered as one of the hardest blocks in China. Yinchuan, Hetao, Shanxi graben, and other second-ary seismotectonic elements compose this area. Area I is located in the northwesternmost part of China and is a less active seismic area compared with the Tianshan region.
Area J is in the easternmost part of China and goes along the big north south Tanlu fault.
These 10 seismic areas are further divided into several seismogenic regions. By seismogenic region, we mean a fault area with dimensions comparable to the fault of the maximum earthquake ever recorded there, which contains secondary faults as well. This de®nition is based on the principle of fault interaction in a speci®c fault system. The detailed division of the 10 areas into seismogenic sources is a tool for a more detailed investigation of the time depen-dent seismicity. By using the information of the seismic activity, the maximum magnitude earthquake, dimensions of focal areas, etc., 147 seismogenic regions were totally de®ned (Fig. 6) in the above 10 areas.
2.2. Seismic moment rate
The seismic moment rate, mo, is one of the important
parameters for the application of this model, since it expresses the tectonic loading exerted in the volume of each seismogenic region (Papazachos et al., 1997a). For the determination of the annual moment rate in the present case, a procedure suggested by Molnar (1979) was applied. According to this procedure, the following relation gives the number of events with seismic moment equal to or large thanMo:
N Mo G:M2oE 2
where
G10a1bk=r; E b
r 3
and a, b are the constants of the Gutenberg and Richter (1944) relation
logNa2bM 4
normalized for 1 year, and r, k the parameters of the moment±magnitude relation
logMorM1k: 5
For the present case, it is taken that r1.5 andk16.1 (Kanamori, 1977; Ekstrom and Dziewonski, 1988). The next relation gives the rate of seismic moment release,mo
(dyn.cm.year21),
mo G
12E:M 12E
o;max 6
whereMo,maxis the seismic moment released by the
maxi-mum earthquake in the region with magnitudeMmax.
2.3. Data used and declustering procedure
The data set used in the present study is taken from the catalogue that was published and distributed by the State Seismological Bureau (SSB) of P. R. China. This catalog was further checked and corrected accordingly with: (a) the catalogue of Pacheco and Sykes (1992) that gives informa-tion on all large earthquakes (M$7.0) which occurred during the present century; (b) The catalogue of Abe (1981) that covers the same time period but gives informa-tion on smaller events (M$6.5) as well since 1930; and (c) The ISC bulletins for the period 1966±1992 for smaller magnitude events. The data of the corrected catalogue
C. Qin et al. / Journal of Asian Earth Sciences 19 (2001) 97±128
have been found to be complete for the following time periods and magnitude cutoffs (Qin et al., 1999a):
Period Magnitude Cutoff 1950±1995 M$5.0
1900±1995 M$6.0 1800±1995 M$8.0
For the application of Eq. (1), only the main shocks, that is the largest earthquakes during the seismic cycle, and their occurrence times are needed. It is then necessary to de®ne them, as well as their ªpreshocksº and ªpostshocksº in the broad sense, as suggested by Papazachos et al. (1997a). These investigators, based on a large sample of observa-tions, proposed the following two relations for the calcula-tion of the total duracalcula-tion of preshock, tp, and postshock, ta,
activity:
tp3 years
log ta0:0610:13Mp
7
where Mpis the magnitude of the preceding mainshock. The
constant duration of the preshock activity and the increase of the duration of the aftershock activity with the size of the mainshock, is in accordance with the results of previous investigators (Mogi, 1985; Karakaisis et al., 1991). This declustering procedure proves to be a main procedure for the model.
2.4. Estimation of the model parameters
Recently, Papazachos et al. (1997a) have shown that the parameters b, c, d, B, C, and D of the empirical formulae are the same for every seismogenic region and seismotectonic environment. They have estimated these parameters by using a large sample of data for mainshocks of the conti-nental fracture system, that is, from seismogenic regions of Circum±Paci®c and of Alpine±Himalayan belts. These parameters were calculated from 1811 observations (T,
Mmin, Mp, Mf) from 274 seismogenic regions in 16 areas
worldwide (Papazachos et al., 1997a). Since this model has been used all over the world and proven to be stable (Papazachos et al., 1997a; Papazachos and Papadimitriou, 1997), the model parameters are kept the same as those found by Papazachos et al. (1997a), except for the constants qandmin Eq. (1). These two parameters must be calculated for every tectonic area. Finally, Eq. (1) takes the form:
logT 0:19Mmin 10:33Mp20:39 logm01q
Mf 0:73Mmin 20:28Mp10:40 logm01m
8
where q andmare parameters that differ among different seismic areas. Based on Eq. (8), the q andmvalues were estimated for the 10 areas and are given in Table 1. These two constants are considered as calibrating factors for each seismic area. The data sample used for the calculation consists of 665 sets (T, Mmin, Mp, and Mf) and concerns
147 seismogenic regions.
Fig. 7 illustrates the frequency distribution of log(T/Tt)
for the interevent times,T, of all 665 observations, together with the best-®t normal distribution which has a mean value of m0.06 and a standard deviation of s0.30. This standard deviation is usually attributed to the intrinsic limitations of the model as well as to the quality and quan-tity of input data and varies in different regions (Scholz, 1990). Papazachos et al. (1997a) found a standard deviation equal to 0.28 for the seismogenic sources of the continental fracture system. It is interesting to note that Nishenko and Buland (1987) found a value equal to 0.21 for the corre-sponding intrinsic standard deviation for mainshocks in plate boundaries. To compare the observed and the theore-tical distributions, the Kolmogorov±Smirnov test was applied. A signi®cance level of 0.91 was found, while the Dk-value, i.e., the largest absolute difference between the
obtained and the theoretical cumulative relative frequencies, is 0.02. The critical value ofDkat this level is equal to 0.05.
No. Area qa sqb Ma smb pa spb l10c SP10d
1 Front Himalaya area 7.74 0.25 26.44 0.42 6.16 0.31 3.56 5.78 2 Himalaya area 7.72 0.27 26.28 0.50 6.08 0.28 5.78 5.63 3 Xianshuihe River Red River area 7.67 0.28 26.15 0.46 6.11 0.29 6.44 7.29 4 Qinghai±Tibet area 7.57 0.33 26.15 0.42 5.97 0.30 4.00 5.66 5 Southeast China area 7.76 0.30 26.39 0.39 6.12 0.25 0.44 0.86 6 Qilianshan area 7.67 0.40 26.23 0.44 6.05 0.37 3.78 4.06
7 Ordos area 7.91 0.30 26.45 0.57 6.25 0.24 2.00 1.32
8 Tianshan area 7.65 0.25 26.16 0.50 6.08 0.26 4.44 4.10 9 Northwest China area 7.85 0.26 26.45 0.44 6.23 0.24 1.33 2.22 10 East China area 7.63 0.34 25.90 0.60 6.06 0.34 1.55 1.79
a q, mandpare the values of the parameters in the Eqs. (8) and (9). b sis the corresponding standard deviation.
c l
10is the average number of mainshocks withMs$6.0 per decade.
d SP
This means that the hypothesis of normal distribution of the quantity log(T/Tt) is valid.
Fig. 8 shows the frequency of the difference, MF2Mf, between the observed, MF, and the calculated
magnitude, Mf, by Eq. (8) for all the regions. The best
®t normal distribution with a mean value of m20.04 and a standard deviation equal to s0.48 is also shown. The corresponding value that Papazachos et al. (1997a) found for the continental fracture system was equal to 0.36.
2.5. Model probabilities
Although the time of the occurrence of the expected mainshock in a seismogenic region can be estimated directly by Eq. (8), it is better to give the probability of occurrence of a certain magnitude earthquake. This is due to the fact that there is considerable ¯uctuation of the observed repeat times, T, in respect to the corresponding repeat times,Tt, given by Eq. (8).
The probability density functions or distributions of the data used must be known in order to proceed in probability determination. Once the distribution function is de®ned, earthquake repeat time estimates can be presented in terms of a conditional probability, which describes the like-lihood of failure within a given time interval, t1dt, provided that the event has not occurred prior to timet.
The form of the current model leads to the assumption that the lognormal distribution appears to be more suitable for the data set, and this is evidenced by Fig. 7. Based on the above and taking into account the occurrence time of the previous mainshock, the model probabilities were calcu-lated in each seismogenic region.
2.6. Statistical tests
It must be noted that earthquake prediction hypotheses are dif®cult to test. The dif®culties concern the lack of a standard procedure for constructing a null hypothesis against which the predictions may be tested and that most predictions cover a long period and the hypotheses evolve before they can be tested (Kagan and Jackson, 1994). A question then arises to what extent the model is superior to a random occurrence model (time-independent). In order to address this question, the term cMp of the ®rst
relation of Eq. (1) is ignored. By doing this, all the main-shocks are ®tted without any limitation (i.e., randomly). This forms the Gutenberg±Richter relation for the main-shocks. Therefore, the empirical relation that describes the time independent model is as follows:
logT039Mmin 20:29 logmo1p 9
wherepis different from one area to another (Table 1). A direct way to check the superiority of the model against the classical time-independent model is to apply both models to a catalog and to compare the results (model prob-abilities) to real observations. This test can be made in an a posteriori way to a reliable and complete sample of data. Since the current data set is complete forM$6.0 during the time interval from 1900 to 1995 in China, these two meth-ods have been applied on this sample in the following way (Papazachos et al., 1997a).
For each one of the seismogenic areas where at least one earthquake of M$6.0 occurred since 1901, and for each possible decade of the sample period, the model probability for the occurrence of a mainshock with magnitudeM$6.0
C. Qin et al. / Journal of Asian Earth Sciences 19 (2001) 97±128
Fig. 8. The normal distribution of the differenceM2Mf, whereMis the
magnitude of main shocks andMfis the magnitude calculated from the
second Eq. (7). Fig. 7. The lognormal distribution of the ratioT/TtwhereTis the interevent
time between two successive mainshocks, andTtis the time calculated from
was calculated. Each period starts just after the occurrence of the ®rst mainshock and ends 1 decade after the occurrence of the last mainshock (e.g. 1901±1910, 1902±1911,¼1987± 1996, when the ®rst mainshock occurred in 1901 and the last one in 1987 in this region). The model probabilities were estimated based on the model Eq. (8) and lognormal distribu-tion ofT/Tt. A corresponding `yes' or `no' was assigned to each
decade if such a mainshock had occurred or not, respectively. Then, the total range of the calculated model probabilities was divided in equal intervals (0.0±0.09, 0.1±0.19¼) and all decades were grouped in these intervals according to their model probabilities. Finally, the success ratio of the number of `yes' to the total number `yes1no' of decades was calcu-lated for each group. The same procedure was repeated by calculating probabilities on the basis of the time-independent seismicity model (Gutenberg±Richter law for distribution of magnitudes, Poisson law for distribution in time) that is expressed in Eq. (9).
The results of this test are shown in Fig. 9, where the success ratios are plotted vs probabilities,P10, for both the
time-dependent model (solid circles) and the time-indepen-dent model (squares with cross). The dashed line which means full success is also shown. This ®gure evidences the fact that the model probabilities are closer to the real earthquake occurrence and cover a larger probability range. In order to compare these results quantitatively, the linear regression analysis was applied in both cases. A slope equal to 1.04 (slope equal to 1 corresponds to the full success) was obtained for the model probabilities. The linear hypoth-esis is accepted at the 0.99 signi®cance level (t9.79.t7,0.0053.50). For the time independent model
probability a slope equal to 1.05 was found and is rejected even at 0.95 signi®cance level (t2.24,t2,0.052.92). It
is worth to note here the limited range of values obtained. Therefore, this test shows the effectiveness of theregional time predictable modeland its superiority with respect to the classical time-independent model for long interevent times (Papazachos et al., 1997a, 1997b).
A second test has also been applied to examine whether the calculated probabilities express the actual rate of earthquake occurrence. As an estimation of the expected number of main-shocks with M$6.0 in each seismic area during the next decade (1996±2005), the sum of the calculated probabilities,
SP10, can be considered for all seismogenic regions in each
seismic area. This sum can be compared with the average number, l10, of such observed earthquakes per decade, as
this number is deduced from the complete sample of data available. This comparison can indicate the level of expected seismic activity in the area, according to the model, with respect to the mean seismic activity. The estimated values are shown in Table 1 and plotted in Fig. 10, where the line is the bisector. From this Fig. we can see that the number of mainshocks predicted are close to the actual rates in almost all the 10 areas. Exceptions are areas A and D where the sum of the probabilities was found to be larger than the mean occur-rence rate. This is probably due to an expected high active period in these areas during the next decade.
3. Application of the model and results obtained
Based on Eq. (8) and the parametersq andmestimated separately for each seismic area, the probabilities for the occurrence of strong (M$6.0) earthquakes in the next 10 years, as well as the magnitude of the expected mainshock can be calculated. The results obtained are shown in Table 2.
Fig. 10. The number of the mainshocks predicted vs the mean number of mainshocks in 1 decade for the 10 seismic areas marked by letters. The dash line means that the actual number is equal to the predicted one. Fig. 9. The success ratio vs the probability given by the time-dependent
C. Qin et al. / Journal of Asian Earth Sciences 19 (2001) 97±128
Table 2
Seismicity parameters (b, a, Mmax,mo) and expected strong mainshocks for each seismogenic region in the 10 areas of China territory.P10is the model
probability for the occurrence of a mainshock withMs$6.0 during the next 10 years andMfis the most probable magnitude of the expected mainshock. The
last two columns give the year of occurrence and the cumulative magnitudea,M, of the mainshocks occurred in each seismogenic region
Region Latitude Longitude a b Mmax logmo P10(%) Mf Year M
1 35.5 72.0 3.58 0.93 6.0 23.52 9 5.7 1981-09-12 6.0
36.4 73.3 34.5 75.5 33.5 74.0
2 33.5 74.0 4.22 0.93 8.3 25.47 76 6.5 1905-04-05 8.3
34.5 75.5 1967-02-20 6.0
31.7 78.1 1980-08-23 5.5
30.5 77.0 1986-04-26 5.3
3 31.7 78.1 3.84 0.93 7.0 24.35 1 5.7 1962-07-14 5.0
31.1 79.5 1969-06-22 5.1
30.8 80.0 1986-07-16 5.1
29.5 79.2 1991-10-20 7.0
30.5 77.0
4 30.8 80.0 4.39 0.93 7.1 24.96 50 6.1 1916-08-28 7.1
29.8 82.4 1945-06-04 6.5
28.4 81.8 1953-02-23 5.6
29.5 79.2 1958-12-28 6.7
1980-07-29 6.5
5 28.4 81.8 4.06 0.93 7.0 24.57 44 6.0 1936-05-27 7.0
29.8 82.4 1954-09-04 6.4
28.9 84.8 1973-10-16 5.1
27.5 84.2
6 27.5 84.2 4.09 0.93 8.0 25.17 68 6.3 1833-08-26 8.0
28.9 84.8 1956-07-03 5.7
28.4 86.8 1974-03-24 6.2
27.0 86.4
7 28.4 86.8 4.08 0.93 8.3 25.33 76 6.4 1934-01-15 8.3
28.4 89.7 1960-08-12 5.0
26.6 89.6 1965-01-12 5.7
27.0 86.4 1971-12-04 5.7
1980-11-20 6.1 1990-01-09 5.1
8 26.6 89.6 4.33 0.93 7.5 25.13 64 6.3 1931-09-13 6.5
28.4 89.7 1941-01-21 6.8
28.7 93.0 1950-08-17 5.8
27.0 93.2 1954-02-23 6.0
1964-04-13 6.2 1985-10-13 5.0 1995-02-17 5.7
9 28.7 93.0 4.43 0.93 7.3 25.11 53 6.1 1947-07-29 7.3
29.8 94.2 1964-10-22 6.7
28.2 95.5 1985-08-01 5.7
27.3 94.7 27.0 93.2
10 29.8 94.2 4.37 0.93 7.0 24.88 47 6.0 1938-11-21 6.0
30.8 95.3 1950-10-08 6.6
29.5 96.8 1965-06-15 5.0
28.2 95.5 1981-10-24 5.7
1988-01-25 5.6 1992-02-06 5.7
11 29.5 96.8 4.57 0.93 8.6 25.99 32 5.9 1950-08-15 8.6
27.8 97.4 1987-01-09 5.0
27.6 95.7 28.2 95.5
12 26.1 95.9 4.41 0.93 7.6 25.26 68 6.3 1908-12-12 7.6
27.6 95.7 1950-08-22 6.3
27.8 97.4 1959-02-15 5.7
27.2 97.5 1962-09-22 6.4
26.1 97.5 1976-08-13 6.3
Region Latitude Longitude a b Mmax logmo P10(%) Mf Year M
13 26.1 95.9 4.40 0.93 7.5 25.20 56 6.6 1931-01-28 7.5
26.1 97.5 1962-02-21 6.6
24.4 97.5 1971-05-31 6.2
24.4 95.9 1981-08-17 5.4
1994-01-11 6.1
14 35.3 74.6 3.93 0.93 6.7 24.27 13 6.0 1950-08-20 5.0
37.0 72.7 1964-02-02 6.1
38.0 73.2 1982-07-02 5.6
38.0 74.3 1990-03-25 6.3
36.4 76.0
15 35.3 74.6
36.4 76.0 33.5 79.0 32.4 77.5
16 33.5 79.0 4.04 0.93 6.8 24.44 24 6.0 1951-08-26 5.0
32.7 80.5 1955-06-27 5.7
31.1 79.5 1963-04-12 5.0
31.7 78.1 1966-08-05 5.1
32.4 77.5 1975-01-11 6.8
17 32.7 80.5 4.05 0.93 6.7 24.39 22 6.0 1911-10-15 6.7
31.7 83.1 1966-03-06 6.1
29.8 82.4 1982-01-24 6.6
30.8 80.0 31.1 79.5
18 29.8 82.4 4.35 0.93 6.7 24.69 39 6.1 1913-03-06 6.5
31.7 83.1 1944-10-18 6.9
30.4 85.9 1957-04-14 6.6
28.9 84.8 1971-05-03 5.4
19 30.4 85.9 4.09 0.93 6.8 24.49 4 6.0 1918-02-05 6.0
30.5 88.2 1951-05-28 5.7
28.4 88.5 1958-11-24 5.2
28.4 86.8 1964-11-10 5.0
28.9 84.8 1970-02-27 5.1
1974-09-27 5.6 1988-09-03 5.3 1993-03-20 6.8
20 28.4 88.5 4.44 0.93 8.0 25.52 48 6.4 1901-04-21 6.8
30.5 88.2 1924-10-09 6.5
30.9 89.3 1935-05-21 6.2
30.1 91.2 1955-03-27 6.3
28.6 91.4 1980-02-22 6.2
28.4 89.7 1989-02-04 6.0
1992-07-30 6.7
21 28.7 93.0 4.07 0.93 7.0 24.58 27 6.0 1915-12-03 7.0
28.6 91.4 1951-12-03 5.8
30.1 91.2 1959-02-22 5.7
31.2 92.7 1968-01-13 5.2
29.8 94.2 1992-08-17 5.5
22 31.2 92.7 3.46 0.93 5.5 23.12 ± ± 1959-11-18 5.2
32.0 93.8 1967-08-15 5.5
30.8 95.3 29.8 94.2
23 30.8 95.3 4.10 0.93 6.5 24.33 28 6.0 1950-08-26 5.2
32.0 93.8 1954-03-09 5.8
33.4 95.1 1961-12-04 5.9
32.2 96.7 1971-04-03 6.6
1985-01-16 5.1
24 32.2 96.7 3.74 0.93 6.5 23.97 24 5.9 1915-05-05 6.5
33.4 95.1 1977-12-16 5.3
34.7 96.3 1986-04-04 5.1
33.4 98.0 1989-04-13 5.4
25 30.1 91.2 4.69 0.93 8.0 25.77 75 6.6 1921-10-15 6.2
C. Qin et al. / Journal of Asian Earth Sciences 19 (2001) 97±128
Table 2 (continued)
Region Latitude Longitude a b Mmax logmo P10(%) Mf Year M
32.0 91.0 1940-09-03 6.3
31.2 92.7 1951-11-18 8.1
1972-07-23 6.0 1977-05-28 6.3 1993-01-18 6.3
26 31.2 92.7 3.61 0.93 6.1 23.61 19 5.8 1955-08-04 5.0
32.0 91.0 1961-11-04 5.5
32.9 91.7 1971-05-23 6.1
32.0 93.8
27 32.9 91.7 4.25 0.93 6.5 24.48 5 6.0 1934-06-23 6.0
34.3 92.8 1959-04-27 6.0
33.4 95.1 1975-05-05 6.1
32.0 93.8 1983-06-15 5.7
1990-06-02 5.7 1994-06-03 6.6
28 33.4 95.1 3.71 0.93 6.5 23.94 24 5.9 1915-04-28 6.5
34.3 92.8 1950-09-19 5.3
35.5 93.7 1962-02-13 5.3
34.7 96.3 1992-02-03 5.2
29 30.3 98.3 3.65 0.93 7.0 24.16 29 6.0 1950-10-31 5.6
32.2 96.7 1966-03-14 5.1
33.4 98.0 1979-03-29 6.2
31.3 99.8
30 30.3 98.3 4.36 0.93 7.5 25.16 35 6.3 1923-10-20 6.5
31.3 99.8 1948-05-25 7.1
29.5 101.2 1960-05-03 5.4
28.8 101.0 1968-03-03 5.7
28.7 99.1 1979-11-06 5.0
1989-05-03 6.8
31 30.8 95.3 3.89 0.93 6.0 23.83 27 5.9 1951-03-17 6.2
32.2 96.7 1984-01-22 5.0
30.3 98.3 29.5 96.8
32 28.6 97.1 3.81 0.93 6.5 24.04 26 5.9 1921-05- 6.5
29.5 96.8 1960-09-02 5.7
30.3 98.3 1986-02-06 5.1
28.7 99.1
33 26.4 100.5 4.22 0.93 7.5 25.02 59 6.4 1925-10-15 6.0
26.3 99.2 1933-06-07 6.2
28.7 99.1 1948-06-27 6.4
28.8 101.0 1961-06-27 6.1
26.9 100.6 1966-09-28 6.1
1976-09-03 5.4 1982-07-03 5.4 1993-07-17 5.9
34 28.7 99.1 3.86 0.93 6.5 24.09 35 6.0 1911-07- 6.5
26.3 99.2 1950-08-22 6.1
26.1 97.5 27.2 97.5 27.8 97.4 28.6 97.1
35 24.1 99.7 4.11 0.93 7.0 24.62 53 6.4 1901-02-15 6.5
26.3 99.2 1925-03-16 7.0
26.4 100.5 1953-06-08 5.0
24.4 101.1 1963-04-23 6.0
1978-05-19 5.6 1986-03-13 5.5
36 21.5 100.5 4.54 0.93 6.7 24.88 26 6.4 1923-07-01 6.5
24.1 99.7 1942-02-01 6.7
24.4 101.1 1953-06-26 5.7
21.7 102.3 1965-07-03 6.1
Region Latitude Longitude a b Mmax logmo P10(%) Mf Year M
1981-09-19 6.0 1993-01-27 6.5
37 23.5 97.8 4.68 0.93 7.0 25.19 41 6.3 1929-10-17 6.9
24.4 97.5 1941-10-31 6.2
26.1 97.5 1955-03-22 6.2
26.3 99.2 1966-09-19 5.6
24.1 99.7 1970-02-05 5.7
1976-05-29 7.2 1991-07-22 5.3
38 21.5 98.6 4.81 0.93 7.3 25.49 28 6.3 1923-06-22 7.1
23.5 97.8 1938-05-14 6.0
24.1 99.7 1941-05-16 7.4
21.5 100.5 1952-06-19 6.5
1984-04-24 6.0 1988-11-06 7.6
39 29.5 101.2 4.28 0.93 7.5 25.08 52 6.4 1913-08- 6.0
28.8 103.0 1952-09-30 6.7
27.4 102.6 1962-02-27 5.6
26.9 101.9 1976-01-17 6.7
26.9 100.6 1988-01-10 5.7
40 24.4 101.1 4.25 0.93 6.9 24.70 34 6.4 1955-09-23 6.9
26.4 100.5 1964-11-20 5.2
26.9 100.6 1975-01-12 5.6
26.9 101.9 1993-08-14 6.1
24.7 102.3
41 24.7 102.3 4.17 0.93 8.0 25.25 68 6.6 1833-09-06 8.0
26.9 101.9 1927-03-15 6.2
27.4 102.6 1966-02-13 6.3
26.6 104.1 1985-04-18 6.3
25.0 104.0
42 23.2 101.6 4.50 0.93 7.3 25.18 40 6.3 1909-05-11 6.5
24.4 101.1 1913-12-21 7.1
25.0 104.0 1929-03-22 6.2
23.2 104.0 1940-04-06 6.0
1950-09-13 5.7 1965-05-24 5.0 1969-01-05 7.3 1980-06-18 5.6
43 38.3 77.8 3.04 0.93 7.0 23.55 ± ± 1975-02-12 5.1
36.9 75.4 38.0 74.3 38.2 74.5 39.0 76.6
44 35.2 77.3 3.87 0.93 6.7 24.21 33 6.2 1925-12-07 6.0
36.9 75.4 1967-05-28 5.9
37.6 76.6 1980-02-14 5.8
36.0 78.5 1993-04-08 5.5
45 35.2 77.3 3.79 0.93 6.5 24.02 29 6.1 1914-10-09 6.5
36.0 78.5 1926-08-07 6.2
34.3 80.2 1953-01-08 5.0
33.5 79.0 1968-02-12 5.3
46 32.4 81.4 3.32 0.93 5.5 22.98 ± ± 1950-08-17 5.3
32.7 80.5 1979-05-21 5.0
33.5 79.0 34.3 80.2 33.6 82.1
47 31.7 83.1 3.97 0.93 6.5 24.20 42 6.2 1955-01-29 6.6
32.4 81.4 1978-04-04 6.1
33.6 82.1 33.1 83.4 32.5 84.7
48 30.4 85.9 3.89 0.93 6.2 23.94 28 6.0 1947-02-10 6.4
C. Qin et al. / Journal of Asian Earth Sciences 19 (2001) 97±128
Table 2 (continued)
Region Latitude Longitude a b Mmax logmo P10(%) Mf Year M
32.5 84.7 31.8 86.3
49 30.4 85.9 4.00 0.93 6.5 24.23 27 6.2 1935-01-03 6.5
31.8 86.3 1952-10-08 5.4
31.8 88.1 1961-12-08 5.5
30.6 88.5 1974-03-03 5.7
30.5 88.2 1986-06-21 6.1
1994-07-24 6.0
50 30.6 88.5 4.16 0.93 7.1 24.73 29 6.1 1908-08-20 7.0
31.8 88.1 1934-12-15 7.1
32.6 89.2 1963-01-22 5.1
32.0 91.0 1977-03-15 5.4
30.9 89.3 1980-06-04 5.8
51 32.0 91.0 3.58 0.93 6.0 23.52 26 6.0 1938-12-03 6.0
32.6 89.2 33.7 89.8 32.9 91.7
52 35.5 79.0 3.90 0.93 7.0 24.41 25 6.1 1902-08-31 6.8
37.6 76.6 1952-06-04 5.2
38.3 77.8 1956-03-05 5.7
36.6 80.0 1975-10-27 5.2
35.8 79.5 1993-06-14 5.1
53 34.3 80.2 3.83 0.93 6.2 23.88 29 6.1 1946-11-07 6.2
35.5 79.0 1973-11-22 5.1
35.8 79.5 1977-02-20 5.7
35.2 81.5 1986-07-07 6.1
33.9 81.2
54 35.2 81.5 4.17 0.93 6.3 24.28 30 6.2 1920-10-12 6.2
35.8 79.5 1948-02-13 6.2
36.6 80.0 1975-04-28 6.4
36.5 81.8 1987-04-10 5.7
1992-04-05 6.2
55 33.1 83.4 3.79 0.93 6.5 24.02 29 6.0 1952-04-02 5.3
33.9 81.2 1961-06-04 6.5
35.2 81.5 1977-01-14 5.2
35.0 82.5 1985-06-15 5.4
34.3 84.4
56 33.7 86.0
34.3 84.4
35.0 82.5 35.8 84.4
34.7 87.2
57 32.5 84.7 0.93 5.5
33.1 83.4 34.3 84.4 33.7 86.0
58 31.8 86.3 3.79 0.93 6.0 23.73 29 6.0 1960-01-31 5.0
32.5 84.7 1973-09-08 6.1
33.7 86.0 1985-10-04 5.3
34.7 87.2 34.2 88.5
59 34.2 88.5 4.00 0.93 6.5 24.23 31 6.1 1950-12-29 5.7
33.7 89.8 1965-06-18 5.8
32.6 89.2 1977-11-18 6.6
31.8 88.1 31.8 86.3
60 32.9 91.7 4.18 0.93 6.5 24.41 30 6.1 1952-09-14 5.8
33.7 89.8 1981-06-10 6.2
35.0 90.9 1986-08-21 6.7
34.3 92.8
61 35.0 82.5 3.67 0.93 5.8 23.50 ± ± 1980-10-07 5.8
35.2 81.5 1989-10-08 5.4
Region Latitude Longitude a b Mmax logmo P10(%) Mf Year M
36.8 82.6 35.8 84.4
62 33.7 89.8 4.11 0.93 6.2 24.16 25 6.2 1920-05-02 6.4
34.7 87.2 1946-02-20 6.0
35.8 88.5 1994-08-14 6.0
35.0 90.9
63 34.7 87.2 4.07 0.93 6.9 24.52 46 6.3 1951-10-11 5.0
35.8 84.4 1965-01-21 5.5
36.6 86.0 1973-07-14 6.9
35.8 88.5 1985-05-20 6.3
64 35.8 84.4 4.34 0.93 7.2 24.96 28 6.1 1924-07-12 7.4
36.8 82.6 1960-12-11 5.5
37.5 84.0 1991-06-17 5.3
36.6 86.0
65 36.6 86.0 2.95 0.93 5.0 22.32 ± ± 1963-05-07 5.0
37.5 84.0 38.2 85.5 37.4 87.9
66 35.8 88.5 3.88 0.93 6.0 23.82 34 6.1 1959-11-11 6.0
36.6 86.0 1966-10-14 6.0
37.4 87.9 36.7 89.8
67 35.0 90.9
35.8 88.5 36.7 89.8 36.2 91.6
68 34.3 92.8 3.74 0.93 5.7 23.51 ± ± 1980-03-07 5.6
35.0 90.9 1989-05-14 5.7
36.2 91.6 35.5 93.7
69 35.5 93.7 3.56 0.93 5.7 23.33 ± ± 1952-10-01 5.0
36.2 91.6 1980-07-13 5.7
37.2 92.5 1986-12-21 5.3
36.6 94.6
70 36.2 91.6 3.42 0.93 5.5 23.08 ± ± 1994-12-28 5.5
36.7 89.8 37.7 91.0 37.2 92.5
71 36.7 89.8 3.51 0.93 5.6 23.22 ± ± 1994-11-28 5.6
37.4 87.9 38.2 89.8 37.7 91.0
72 37.4 87.9 4.09 0.93 6.8 24.49 16 6.1 1933-09-26 6.7
38.2 85.5 1957-08-23 5.0
39.0 87.5 1963-08-12 5.1
38.2 89.8 1993-10-02 6.8
73 34.7 96.3 4.72 0.93 6.9 25.17 48 6.4 1902-11-04 6.9
35.5 93.7 1963-04-19 6.9
36.6 94.6 35.9 97.5
74 33.4 98.0 4.63 0.93 7.5 25.43 66 6.6 1936-01-07 7.5
34.7 96.3 1971-03-24 6.4
35.9 97.5 35.5 99.0 34.8 99.5
75 32.0 99.2 4.09 0.93 7.4 24.83 29 6.1 1947-03-17 7.4
33.4 98.0 1961-05-18 5.2
34.8 99.5 33.1 100.7
76 30.3 100.6 4.56 0.93 7.5 25.36 45 6.3 1904-08-30 7.5
32.0 99.2 1919-05-29 6.4
33.1 100.7 1923-03-24 7.1
C. Qin et al. / Journal of Asian Earth Sciences 19 (2001) 97±128
Table 2 (continued)
Region Latitude Longitude a b Mmax logmo P10(%) Mf Year M
1972-02-06 7.3
77 31.1 101.9 4.35 0.93 7.7 25.25 85 6.8 1932-03-07 6.0
29.4 103.2 1941-06-12 6.0
28.8 103.0 1955-04-14 7.4
29.5 101.2 1975-01-15 6.2
30.3 100.6 1989-06-09 5.6
78 27.4 102.6 4.55 0.93 7.3 25.23 58 6.6 1917-07-31 7.3
29.4 103.2 1936-04-27 6.9
29.4 104.6 1959-03-11 5.3
28.6 104.8 1966-10-11 5.1
26.6 104.1 1970-07-31 5.4
1974-05-11 6.8 1994-12-30 6.1
79 24.4 107.5 3.43 0.93 6.0 23.37 ± ± 1955-05-27 5.0
25.0 104.0 1973-08-02 5.2
26.6 104.1 1985-12-02 5.4
28.6 104.8
80 22.5 107.0 3.62 0.93 5.7 23.39 ± ± 1962-04-23 5.5
23.2 104.0 1982-10-27 5.7
25.0 104.0 24.4 107.5
81 22.5 107.0 3.86 0.93 6.7 24.20 22 6.0 1936-04-01 6.7
24.4 107.5 1958-09-25 5.7
23.6 111.0 1977-10-19 5.0
21.7 110.5
82 23.6 111.0 3.78 0.93 6.4 23.95 19 6.0 1911-05-15 6.0
24.0 113.8 1969-07-26 6.4
23.3 114.6 1986-01-28 5.2
22.2 115.6 21.6 113.0 21.7 110.5
83 24.0 113.8 3.70 0.93 6.1 23.70 21 6.0 1962-03-19 6.1
26.8 116.5 1982-02-25 5.0
25.6 118.0 1987-08-03 5.4
23.3 114.6
84 22.2 115.6 4.29 0.93 7.2 24.91 24 6.0 1906-03-28 6.2
23.3 114.6 1918-02-13 7.3
25.6 118.0 1968-04-01 5.2
24.2 119.5 1995-02-25 5.6
85 24.2 119.5 3.54 0.93 8.0 24.62 ± ± 1960-07-21 5.0
25.6 118.0 1978-08-10 5.2
27.8 121.6 1985-11-28 5.0
26.6 122.8 1992-02-18 5.6
86 38.2 89.8 3.77 0.93 5.7 23.54 ± ± 1960-11-17 5.4
39.0 87.5 1979-12-02 5.6
39.7 89.5 1991-01-06 5.1
38.8 91.7 1994-09-07 5.7
87 37.2 92.5 3.93 0.93 6.3 24.04 22 6.1 1976-01-02 6.3
37.7 91.0 1990-01-14 6.1
38.2 89.8 38.8 91.7 38.4 93.3
88 36.6 94.6 3.21 0.86 6.0 23.52 19 6.0 1952-10-06 6.0
37.2 92.5 1993-09-05 5.5
38.4 93.3 37.7 95.6
89 36.6 94.6 3.88 0.86 6.9 24.77 26 6.1 1957-05-04 5.5
37.7 95.6 1962-05-21 6.9
37.0 98.5 1972-08-30 5.7
35.9 97.5 1977-01-19 5.9
1985-08-12 5.4 1991-09-02 5.5
Region Latitude Longitude a b Mmax logmo P10(%) Mf Year M
36.4 101.0 1990-04-26 7.2
35.5 100.5 35.5 99.0 35.9 97.5
91 33.4 100.5 3.07 0.86 5.1 22.80 ± ± 1950-06-18 5.0
35.5 99.0 1958-09-11 5.1
35.5 100.5 1978-02-21 5.1
34.0 101.8
92 33.4 100.5 3.48 0.86 6.0 23.79 22 6.0 1935-07-26 6.0
34.0 101.8 1952-11-01 6.0
32.8 102.8 1970-09-05 5.8
31.1 101.9
93 32.8 102.8 3.34 0.86 6.2 23.78 17 6.0 1974-09-23 5.6
35.5 100.5 1987-01-08 6.2
36.4 101.0 34.6 104.0 33.7 103.5
94 29.4 104.6 3.56 0.93 6.0 23.50 19 6.0 1962-07-01 5.1
29.4 103.2 1967-01-24 5.5
31.1 101.9 1970-02-24 6.0
31.0 104.5
95 31.1 101.9 4.67 0.93 8.0 25.75 56 6.6 1879-07-01 8.0
32.8 102.8 1933-08-25 7.3
33.7 103.5 1952-11-04 5.7
33.2 105.5 1960-11-09 6.8
31.0 104.5 1973-08-11 6.1
1976-08-16 7.1 1989-09-22 6.8
96 33.7 103.5 3.34 0.86 8.0 24.93 32 6.4 1936-08-01 6.0
34.6 104.0 1961-10-01 5.8
35.7 104.7 1987-10-25 5.5
35.5 106.6 33.2 105.5
97 34.6 104.0 3.41 0.86 7.0 24.36 20 6.0 1936-02-07 6.7
36.4 101.0 1957-07-18 5.1
37.2 102.2 1964-05-31 5.0
35.7 104.7 1968-12-22 5.4
1995-07-22 6.1
98 36.4 101.0 2.96 0.86 5.4 22.89 ± ± 1952-01-27 5.0
37.0 98.5 1986-09-17 5.4
38.2 99.5 37.2 102.2
99 37.2 102.2 3.78 0.86 7.7 25.18 49 6.4 1927-05-23 7.7
38.2 99.5 1955-05-04 5.0
39.1 100.3 1958-11-30 5.1
38.2 103.0 1978-08-16 5.0
1986-08-26 6.4
100 38.9 97.0 3.51 0.86 7.2 24.59 26 6.2 1962-08-01 5.4
40.1 98.0 1969-10-17 5.1
39.9 100.6 1975-01-04 5.3
39.1 100.3 1993-10-26 6.4
38.2 99.5
101 38.9 97.0 3.69 0.86 7.5 24.96 35 6.4 1932-12-25 7.5
39.7 94.2 1951-12-27 6.1
40.8 95.7 1989-09-21 5.3
40.1 98.0
102 37.7 95.6 3.41 0.86 6.5 24.04 23 6.1 1927-03-16 6.0
38.9 97.0 1930-07-14 6.5
38.2 99.5 1938-08-23 6.0
37.0 98.5 1957-03-23 5.0
1980-04-18 5.2
103 37.7 95.6 3.12 0.86 5.7 23.24 ± ± 1980-06-01 5.6
C. Qin et al. / Journal of Asian Earth Sciences 19 (2001) 97±128
Table 2 (continued)
Region Latitude Longitude a b Mmax logmo P10(%) Mf Year M
39.7 94.2 38.9 97.0
104 40.3 91.8 3.48 0.86 6.5 24.11 20 6.0 1922-10-17 6.5
39.7 94.2 1977-10-20 5.1
38.4 93.3 1987-02-26 6.4
38.8 91.7 39.7 89.5
105 38.2 103.0 3.46 0.86 7.0 24.41 14 5.7 1954-02-11 7.0
39.1 100.3 1989-09-04 5.1
39.9 100.6 39.8 103.2
106 35.7 104.7 3.16 0.86 6.2 23.60 1 5.6 1959-01-31 5.2
37.2 102.2 1990-10-20 6.2
38.2 103.0 37.7 104.8
107 35.7 104.7 3.90 0.86 8.4 25.75 17 5.9 1920-12-16 8.4
37.7 104.8 1959-01-31 5.2
37.6 106.5 1970-12-03 5.5
35.5 106.6 1982-04-14 5.7
1989-11-02 5.4
108 38.2 103.0 3.54 0.86 7.0 24.49 15 5.8 1954-07-31 7.0
39.8 103.2 39.6 104.8 37.7 104.8
109 37.7 104.8 3.56 0.86 8.0 25.15 47 6.3 1962-12-18 5.7
39.6 104.8 1971-06-28 5.1
39.3 106.7 1987-08-10 6.0
37.6 106.5
110 39.6 104.8 3.36 0.86 6.2 23.80 9 5.7 1959-10-06 5.0
40.3 105.0 1976-09-23 6.2
41.5 106.5 1991-01-13 5.8
40.3 107.6 39.3 106.7
111 40.3 91.8 3.39 0.86 6.4 23.96 11 6.0 1983-10-06 5.5
39.7 89.5 1987-12-22 6.4
41.3 88.1 42.3 90.3
112 41.3 88.1 3.35 0.86 5.6 23.40 ± ± 1953-11-29 5.0
42.7 86.9 1983-12-15 5.2
43.2 89.2 1991-06-06 5.8
42.3 90.3
113 43.2 89.2 3.83 0.86 6.6 24.52 45 6.2 1907-05-13 6.0
42.7 86.9 1934-08-07 6.0
44.2 85.6 1953-04-26 6.2
44.7 86.1 1965-11-13 6.6
45.1 88.2 1980-11-06 5.9
1987-10-06 5.0
114 42.7 86.9 4.09 0.93 7.1 24.66 42 6.2 1906-12-23 7.1
41.0 84.4 1927-09-23 6.7
42.8 84.0 1960-11-30 5.0
44.2 85.6 1966-07-22 5.0
1972-04-09 5.7 1988-05-26 5.2 1993-02-03 5.7
115 42.8 81.7 4.32 0.93 7.3 25.00 68 6.5 1916- - 6.0
42.8 84.0 1939-02-23 6.0
41.0 84.4 1949-02-24 7.3
40.7 81.2 1963-12-18 5.8
42.3 80.2 1970-11-29 5.0
Region Latitude Longitude a b Mmax logmo P10(%) Mf Year M
116 41.5 78.0 4.46 0.93 6.7 24.80 23 6.3 1915-12-17 6.5
42.3 80.2 1959-06-28 6.7
40.7 81.2 1969-02-12 6.7
39.8 78.9 1978-03-12 5.5
1987-01-24 6.4 1991-02-25 6.7
117 41.5 75.8 4.25 0.93 7.0 24.76 43 6.4 1938-06-21 6.7
43.0 75.8 1969-02-12 5.2
43.3 79.5 1978-03-25 7.0
42.3 80.2 1990-11-12 6.3
41.5 78.0
118 39.0 76.6 4.53 0.93 7.3 25.21 76 6.6 1902-08-31 7.3
40.8 75.9 1953-07-10 6.2
41.5 78.0 1961-04-14 7.0
39.8 78.9 1970-07-29 6.1
1977-12-19 6.2 1986-04-26 5.6 1995-05-15 5.7
119 38.2 74.5 4.89 0.93 7.6 25.74 58 6.4 1902-08-22 7.6
40.0 74.0 1919-07-24 6.5
40.8 75.9 1944-09-28 6.8
39.0 76.6 1955-04-15 7.2
1969-09-15 5.9 1978-01-08 6.1 1985-08-23 7.4
120 38.0 73.2 4.53 0.93 7.2 25.15 44 6.2 1918-12-01 6.5
39.5 72.3 1950-07-06 5.3
40.0 74.0 1963-10-16 6.7
38.2 74.5 1974-08-11 7.3
38.0 74.3 1987-08-06 5.4
1994-10-03 5.5
121 42.8 81.7 4.10 0.93 7.5 24.90 36 6.0 1921- - 6.5
42.3 80.2 1951-10-31 5.4
43.3 79.5 1958-12-21 6.7
45.2 81.0 1967-01-14 5.0
44.3 83.0 1970-11-16 5.5
122 42.8 84.0 4.45 0.93 8.0 25.53 72 6.5 1812-03-08 8.0
42.8 81.7 1944-03-10 7.1
45.2 83.8 1955-04-24 6.6
44.7 86.1 1966-07-06 5.0
1973-06-03 6.0 1983-06-29 5.3 1990-10-25 5.8 1995-05-02 6.1
123 45.2 81.0 3.49 0.93 6.0 23.43 8 5.6 1932-09-11 6.0
47.0 83.0 1962-03-28 5.0
46.2 84.8 1975-04-23 5.0
44.3 83.0
124 46.2 84.8 3.61 0.86 6.8 24.43 2 5.8 1982-03-20 5.0
47.0 83.0 1990-06-14 6.8
48.7 85.4 47.7 86.8
125 46.2 89.5 3.99 0.86 7.7 25.39 60 6.4 1931-08-11 7.7
48.3 88.6 1970-09-19 5.2
48.7 90.0 1987-09-19 6.2
46.8 91.1
126 47.7 86.8
C. Qin et al. / Journal of Asian Earth Sciences 19 (2001) 97±128
Table 2 (continued)
Region Latitude Longitude a b Mmax logmo P10(%) Mf Year M
46.2 84.8
127 44.3 91.3 3.60 0.86 6.5 24.23 25 6.0 1933-02-13 6.5
46.2 89.5 1954-02-19 5.7
46.8 91.1 1980-12-16 5.8
44.8 93.0
128 44.3 91.3
43.2 89.2 45.1 88.2 46.2 89.5
129 42.8 92.7 0.86 7.5
44.3 91.3 44.8 93.0 43.3 94.3
130 42.3 90.3 3.84 0.86 7.2 24.92 27 6.0 1914-08-05 7.2
43.2 89.2 44.3 91.3 42.8 92.7
131 40.3 107.6 3.45 0.86 6.2 23.89 9 5.7 1934-01-21 6.2
41.5 106.5 1979-08-25 6.2
41.5 109.5 39.7 109.5
132 39.7 109.5 3.41 0.86 6.2 23.85 10 5.7 1929-01-14 6.0
41.5 109.5 1976-04-06 6.2
41.0 113.0 39.3 112.7
133 39.3 112.7 3.96 0.93 7.0 24.47 9 6.0 1981-08-13 5.5
41.0 113.0 1989-10-19 6.4
40.8 115.0 39.0 114.4
134 37.5 112.0 3.67 0.93 7.2 24.29 ± ± 1952-10-08 5.6
39.3 112.7 1991-01-29 5.5
39.0 114.4 37.2 113.6
135 34.5 110.0 3.82 0.93 8.0 24.90 ± ± 1959-08-11 5.4
37.8 110.0 1965-01-13 5.8
37.5 112.0 1979-06-19 5.2
37.2 113.6 1989-12-25 5.3
34.0 112.5
136 34.0 112.5 4.40 0.93 7.5 25.20 26 6.2 1937-08-01 7.1
37.2 113.6 1983-11-07 5.9
36.7 115.9 33.8 115.2
137 37.2 113.6 4.43 0.93 7.1 25.00 18 6.0 1956-08-19 5.2
39.0 114.4 1966-03-22 7.3
38.7 116.0 1981-11-09 5.8
36.7 115.9
138 39.0 114.4 3.32 0.93 6.7 23.66 ± ± 1956-01-01 5.0
40.8 115.0 1967-07-28 5.4
40.5 116.5 38.7 116.0
139 37.8 110.0 0.93 5.7
39.7 109.5 39.3 112.7 37.5 112.0
140 36.7 115.9 4.55 0.93 8.0 25.63 36 6.5 1945-09-23 6.2
38.7 116.0 1967-03-27 6.3
40.5 116.5 1976-07-28 7.8
40.5 120.0 1991-05-30 5.7
39.0 120.0
141 43.8 125.4 4.27 0.93 7.2 24.89 47 6.7 1922-09-29 6.5
40.0 126.0 1974-02-04 7.2
39.0 120.0 1991-08-17 6.3
The ®rst column of this Table gives the region's number, which corresponds to the number marked in the following Figs. The second and third columns give the vertices of each region. In the fourth and ®fth columns the coef®cients of the Gutenberg±Richter relation are written. The sixth column gives the magnitude of the largest earthquake ever known to occur in each seismogenic region and the seventh column the annual moment rate calculated by Eq. (6). The eighth and ninth columns give the model probability in the next decade and the magnitude of the expected mainshock. The tenth and eleventh columns give the occurrence time and the magnitude of the mainshocks in each region.
In the following, a detailed discussion of each seismic area is presented. Unless otherwise indicated, theb-value is the one suggested by Qin et al. (1999a), whereas thea -value is normalized to the annual -value obtained by making theb-value ®xed. The ®rst event with M$5.0 reported to occur over the past one thousand years is named the earliest known earthquake. We cannot rule out the possibility of an earlier record. For example, in North China, the seismic record can go back as far as 1831 BC (Shi et al., 1974). The motions (deformations or offsets) mentioned in the text are the seismic ones obtained by Qin et al. (2000). The events where M$6.0 and M$7.0 are considered to be ªstrongº and ªlargeº, respectively. The terms ªhigh prob-abilityº and ªvery high probprob-abilityº are used to imply that the model probabilities (the eighth column in Table 2) are
within (0.45, 0.65) and (0.65, 1.0), respectively. For the cases where probabilities are P10,0.45, the term ªlow
probabilityº is used.
3.1. Area A (Front Himalayan area)
The Qinghai±Tibet Plateau is the largest area of elevated topography and thickened continental crust on the Earth and is widely considered as the typical example of continental collision. The present distribution of shallow seismicity is due to underthrusting of the continental lithosphere of the Indian plate (Isacks et al., 1968), whereas the occurrence of intermediate-depth earthquakes suggest the existence of the remnants of the old oceanic lithosphere (McKenzie, 1969; Rastogi, 1974). A complex history of Tethyan oceanic subduction, terrain accretion and continental collision of the Indian and Eurasian plates has lead to the current con®g-uration of the plateau. Seismic structure of the lithosphere provides important constraints on the large-scale spatial and long-term temporal evolution of the Indo±Eurasian colli-sion (Rodgers and Schwartz, 1998).
This area, situated around the Himalayan Frontal Belt, is controlled by thrust faulting and is one of the most active places in China. The strong negative Bouguer anomaly suggests that the Himalayas are underlain by a 65±70 km thick crust (Gupta and Narain, 1967; Shi et al., 1973; Verma and Subrahmanyam, 1984) with low shear-wave velocities
Region Latitude Longitude a b Mmax logmo P10(%) Mf Year M
142 39.0 120.0 3.94 0.93 8.5 25.31 24 6.3 1969-07-18 7.1
35.0 121.5 1992-01-23 5.6
144 32.1 118.4 3.95 0.93 7.0 24.46 24 6.4 1921-12-01 6.8
34.3 118.0 1975-09-02 5.3
35.0 121.5 1987-02-17 5.5
32.6 122.5
145 32.6 122.5 4.01 0.93 6.7 24.35 29 6.4 1974-04-22 5.5
31.0 123.0 1979-07-09 6.0
30.0 119.0 1984-05-21 6.5
32.1 118.4
146 32.1 118.4 3.50 0.93 6.2 23.55 19 6.2 1917-01-24 6.2
30.0 119.0 1954-06-17 5.3
(Gupta and Bhatia, 1981; Valdiya, 1992). The Himalayas can be further divided into three fracture belts: MFT, the Main Frontal Thrust (or HFT, i.e., Himalayan Frontal Thrust [Valdiya, 1992; Yeats et al., 1992]); MBT, the Main Bound-ary Thrust; MCT, the Main Central Thrust, which looks like an arc and extends in about 1000 km (Zheng and Shun, 1992). The high seismic activity is mainly due to the motion of the Indian plate. Instrumental seismicity is very high in a zone of 50 km width across strike in the Lesser Himalayas (India and Nepal), with a concentration of earthquakes just south of the MCT (Seeber and Armbruster, 1981; Ni and Barazangi, 1984; Yeats et al., 1992). Focal mechanisms are mainly of thrust type withP-axes tending to be normal to the MCT and MBF (Baranowski et al., 1984). Twelve seismo-genic regions are de®ned in this area (Fig. 11).
ªHigh probabilityº has been found for the occurrence of an earthquake withMf$6.0 in four regions, namely regions
4, 8, 9, and 10. Four regions (region 2, 6, 7, and 12) are expected to be struck by strong events with ªvery high prob-abilityº (0.76, 0.68, 0.76, and 0.68, respectively) (Table 2).
3.2. Area B (Himalayan area)
This area is located adjacent to the previous one and is a large normal faulting area running along the Indus Tsangpo Suture. Almost all the T-axes of the available fault plane solutions have an east±west direction except for a few cases which show a strike±slip component or a normal one but with north±south directedT-axes close to the sharp turn of the Suture towards the Red River fault. Most of the earth-quakes were located around the sharp turn of the Indus Tsangpo River and the Red River area. These normal faults are generally located at higher altitude (.5000 m) and gradually change to strike±slip faults with the decrease of
C. Qin et al. / Journal of Asian Earth Sciences 19 (2001) 97±128
Fig. 12. The same as Fig. 11 for Area B.
altitude northwards (Molnar and Lyon-Caen, 1989). The calculations presented by Qin et al. (2000) con®rmed the relative decrease of the crustal thickness in the same direc-tion. Several possible explanations have been presented about these normal faults. England and Houseman (1988) explained the formation of these normal faults because of thermal activity. Zhang and Xu (1995) claimed that the normal faults only exist in the upper part of crust, while they change into the strike±slip type at the bottom of the seismogenic layer. However, we cannot rule out that the gravitation which dominates the stress causes the normal faults, since the upper part of the crust does not directly suffer the compression of the Indian plate.
Twenty-one seismogenic regions are de®ned in this normal-faulting dominated area (Fig. 12). Six of them have been struck by strong earthquakes (M$7.0) (M8.0 in 20, M7.0 in 21, M8.0 in 25, M7.0 in 29, M7.5 in 30, and M7.5 in 33). ªVery high probabilityº (0.75) was found for region 25 for the occurrence of a strong event of Mf6.6, whereas high probabilities of 48 and 59% were
estimated for regions 20 and 33, respectively, with expected magnitudes ofMf6.4 for the next 10 years. The
probabil-ities in regions 19 and 27 are very small (4 and 5%). No
probabilities have been calculated for regions 15 and 22 due to the lack of previous mainshocks withM$6.0.
3.3. Area C (Xianshuihe River and Red River area)
Pushed by the two reverse-faulting dominated areas (Area A and F), this area is believed to be the only channel to accommodate the creeping motion, which goes along the large Xianshuihe fault. Zheng and Shun (1992) pointed out that the upper mantle, 90±110 km beneath the Xian-shuihe fault, is the low S-wave velocity belt (3.7±3.8 km/ s), which implies a large deposit of the ª¯owingº material. The ª¯owingº is considered to re¯ect the crustal motion pushed in by the northward motion of the Indian plate (Zheng et al., 1992). The crust and uppermost mantle of the Qianghai±Tibet Terrain reveals a much higher crustal Poisson's ratio (Owens and Zandt, 1997; Rodgers and Schwartz, 1998), low average crustal S-wave velocities, lowQvalue (Rodgers and Schwartz, 1998) and the presence of volcanism in the northern part (Turner et al., 1993, 1996; McNamara et al., 1995, 1997), which strongly suggest the presence of the melt substance in the crust (Rodgers and Schwartz, 1998).
The motion direction in this area, deduced by the results of Qin et al. (2000), changes from NE (region 73) in the north, through the diamond block circled by the Xianshuihe in the northeast, the Anninghe, the Zemuhe, the Xiaojiang in the east, the Red River in the southwest and the Jinsajiang fault in the west (Xu and Deng, 1996), to the SEE in the south of the area (regions 36 and 38). The change of direction, to some extent, re¯ects the contrast of the ª¯uidº crust hardness with that of the ªshoreº crust. The great change is found around the north-west and southeast parts of the diamond block. From the region 77 down to region 37, the motion is dominated by clockwise rotation. From region 38 to region 42, this kind of motion is sharply changed to the counterclockwise rotation, which seems to be a ªwhirlwindº.
Area C is one of the most active areas in continental China, which forms the southeast boundary of the Qinghai±Tibet plateau. The northern part goes along the Xianshuihe fault, whereas the southern part goes mainly along the Red River fault. It is dominated by strike±slip faults, with many strong earthquakes having occurred there. Fifteen seismogenic regions are de®ned in this area (Fig. 13). Large earthquakes have struck 12 out of 15 regions in this area with M$7.0 (Table 2). Some of the regions have been ruined by such events more than once. Three regions (41, 74, and 77) were found to have ªvery high probabilityº for the occurrence of strong (M$6.0) earthquakes over the next 10 years. For six regions, (13, 35, 39, 73, 76, and 78) ªhigh probabilitiesº have been found for the occurrence of strong (M$6.0) earthquakes in the next 10 years.
3.4. Area D (Qinghai±Tibet area)
This area is believed to be the major part of the Qinghai±
Tibet plateau, which accommodates the intrusion of the northward movement of the Indian plate. The fault plane solutions are primarily of strike±slip type. The southern boundary of the area is supposed to have a distinguished strike±slip character compared to its neighbor area. Armijo et al. (1989) thought that the Karakorum, the Gyaring, the Beng Co, and the Jiali fault, which are right-lateral strike± slip faults, are the southernmost boundary which has clear eastward motion (Fig. 14). The Himalayan thrusting Arc (Fig. 2) and the Tianshan thrusting fault together with the Qilianshan fault (Fig. 5) constitute the harder rims of the Qinghai±Tibet plateau. The intruding part of the Indian slab can only ª¯owº eastward because there is almost no open-ing in the west of Qopen-inghai±Tibet plateau to pass through. The northward motion component is gradually changed to an eastward one, because of the force that is exerted from the north. The large left-lateral thrusting Altun fault seems to be the product of the northward subduction of the Indian plate and the impeding of the Tianshan thrusting fault. The Altun fault starts from the western rim of the Tarim basin and goes along the southwest/south edge of the basin until it reaches the Qilianshan fault in the northeastern part of Qinghai±Tibet plateau. This left-lateral strike±slip permits the eastward movement of the Qinghai±Tibet Plateau rela-tive to the Tarim basin (Tapponnier and Molnar, 1977; Meyer et al., 1996). The rare seismic activity in the Tarim basin partially implies its rigidity or homogeneity so that the impeding force of the Tianshan thrusting fault is transferred to the southern edge of the Tarim basin.
The seismic activity in this area is irregularly distributed. In its central part, the Qaidam basin forms a large aseismic region. However, many different geological time scale blocks also show the difference in the seismic activity (Ma, 1987). The left-lateral Kunlun fault runs from the
C. Qin et al. / Journal of Asian Earth Sciences 19 (2001) 97±128
east of the area to the western China where it meets the Karakorum fault. Its northern boundary, along the Altun fault, is adjacent to the Tarim Basin, which is a very low seismicity area, whereas the seismicity in the southern part is high. Generally, the seismic activity decreases from south to north. This area has been divided into 30 seismogenic regions (Fig. 15).
The seismic motion generally goes northeastward. The motion rate decreases from the south part of the area to the southern edge of the Tarim basin. The seismic deforma-tion rate ranges from 10 to 16 mm/year and the velocity around the Kunlun Fault region changes direction rapidly (Qin et al., 2000). The Tarim and the Qaidam basins may in¯uence it; the seismic motion changes its direction rapidly to the west border of the Qaidam basin.
Compared with other areas in the Qinghai±Tibet plateau,
the seismicity in this area is lower. Due to the compli-cated tectonics and irregular spatial distribution of seis-micity, considerable change in seismic behavior is observed from one region to another. Generally, the seis-micity tends to decrease from south to north. The regions near the Tarim basin have the lowest seismicity, espe-cially around the western part of Qaidam basin. The a -value, considered the parameter to describe the seismic activity (Rundle, 1989; Turcotte, 1989, 1992; Pacheco et al., 1992), is also found to be smallest in this place (Table 2). Only in three (50, 52, and 64) of the 30 regions have very strong events (M$7.0) occurred, whereas 11 (43, 46, 56, 57, 61, 65, 67, 68, 69, 70, 71) out of the 30 regions have had no strong earthquake in the current century. For the next 10 years, ªhigh probabilityº has been found for only one region (63).
Fig. 15. The same as Fig. 11 for Area D.
3.5. Area E (Southeast China area)
This area is considered as the continuation of the large strike±slip belt. Since it is located in the southeast part of China where the seismicity is low, no information on a large earthquake occurrence has ever been reported. The seismic activity in this area can be more or less treated as the passive response of Indian plate subduction. It is this passive motion which partially balances the Indian intrusion. Molnar and Gipson (1996) thought that the eastward movement of south China is consistent with Indian penetration causing extru-sion of material in front of it, the negligible movement implies that penetration is absorbed entirely by crustal thickening.
This area is characterized by very low seismicity and has been divided into seven seismogenic regions (Fig. 16). The tectonic loading in regions 84 and 85 is high as expressed by high moment rates. Only four (81, 82, 83, and 84) out of seven regions are expected to be struck by strong earth-quakes withM$6.0 in the next 10 years, whereas no prob-abilities were calculated for other three regions due to lack of previous strong mainshocks (Table 2).
3.6. Area F (Qilianshan area)
This area is circumscribed by the thrust Qilianshan fault on the northeast, the left-lateral strike±slip/thrust Altun fault on the north, and the left-lateral/thrust Longmenshan fault on the east. It is primarily dominated by thrust faulting. The Qilianshan fault is the most active one in this present century. The northeastward crustal motion due to the Indian plate is forced to change eastward when it hits this area. Maybe this is the reason why the strike±slip faults in this
area bear the reversing characteristics. Most northward hori-zontal motion caused by the Indian plate is absorbed by this area in the form of crustal thickening (Qin et al., 2000). For example, the eastern left-lateral Altun fault and the north-eastern left-lateral Longmenshan fault are gradually domi-nated by thrust faulting when they approach this area. Qin et al. (2000) treated this region as a part of the northern ªbankº of a special ªriverº. The seismic kinematics results also display the steady decrement of the velocity from southwest to northeast in the area. Compared with the harder crust on its north and the ª¯uidº crust on its southwest, this area behaves like a transition zone, which dampens the north-ward motion.
Nineteen seismogenic regions were de®ned in this area (Fig. 17). Among them, two (region 95 and 99) have been found to exhibit a high probability for the occurrence of strong earthquakes (M$6.0) within the coming 10 years. Although very strong earthquakes (M$7.0) had struck regions 90, 96, 97, 100, and 101 in the past, low seismic activity is expected for these regions for the following 10 years. No estimation was made for the future seismic activ-ity for regions 86, 91, 98, and 103 due to the lack of previous strong earthquake (Table 2).
3.7. Area G (Ordos area)
This area is distributed around the Ordos Block which is located in northern China. The Ordos is a hard block in China within which the seismicity is very low (Zhang et al., 1998). This block appears to be aseismic with all of the seismicity located along its boundaries. In accordance to the seismicity, geologic data indicate that the Ordos Block has not experienced deformation since Mesozoic
C. Qin et al. / Journal of Asian Earth Sciences 19 (2001) 97±128
time (Tapponnier and Molnar, 1977; Wesnousky et al., 1984). However, the margins of the Ordos block form one of the most seismically active areas in China. The Hetao and Weihe graben systems mark the northern and southern boundaries of the Ordos Block and the Ningxia and Shanxi graben systems are located on the western and eastern rims of the Ordos. These grabens can be further grouped into two major systems: The Yinchuan± Hetao graben system along the northern and northwes-tern margins of the block and the Weihe±Shanxi graben system along the southern and eastern margins of the Ordos. The S-shaped Shanxi rift is composed of NNE-trending transtensional segment in the middle and NE± ENE-trending extensional domains on its north and south (Xu et al., 1992). Indirectly in¯uenced by the collision between the Indian and Eurasian plates, this area was uplifted during the Paleocene with basaltic eruption in the Fanshi area (Chen et al., 1985; Xu et al., 1991). The Ordos Block is found to rotate
counter-clockwise as a whole (Xu et al., 1994; Xu and Deng, 1996; Zhang et al., 1998).
Around the hardest Ordos block, strike±slip faulting mainly controls the area, along the Huanghe River. Fifteen regions were de®ned in this area (Fig. 18). From historical information it is known that three regions (107, 109, 135) have been struck by very large earthquakes (M$8.0), and the other six (105, 108, 133, 134, 136, 137) by large ones (M$7.0). However, only one region (109) is found to exhi-bit ªhigh probabilityº for the occurrence of a strong (Mf$6.0) earthquake during the next 10 years, whereas
in most regions the estimated probabilities are low.
3.8. Area H (Tianshan area)
As a continuation of the Qilianshan thrust fault, this area constitutes the northwestern part of the thrust zone, which is believed to accommodate the northward movement caused by the Indian plate. Due to the Pamir±Tianshan collision
Fig. 18. The same as Fig. 11 for Area G.
and the Indian push, this area is dominated by horizontal compression, manifested by imbricate, low-angle thrust faults that separate the upper crust into a series of tectonic sheets (Lukk et al., 1995). Tianshan, Altun and the western Kunlun mountains overthrust the Tarim and Junggar Basins. This results in a series of reverse faults and active folds formed on the mountain front area, while compression inter-mountain basins controlled by reverse faults, such as Turpan basin, formed in the hinterland of the Tianshan mountain. The most typical ones are several rows of active reverse faults and folds developed on both sides of Tianshan Moun-tains. The reverse faults on the northern piedmont dip south-ward, while the axial plane of several arrays of fault-propagation folds controlled by the reverse faults are also southward dipping, steeper on the northern ¯ank, and gentler on the southern ¯ank. The active reverse faults and the axial plane of multi arrays of active folds controlled by the faults are dipping northward at southern piedmont of Tianshan, while the surface faults converge downward on a detachment (Feng et al., 1991; Deng et al., 1991, Zhang et al., 1994; Deng, 1995a,b). The mixture of the thrust and strike±slip fault plane solutions in the western part of the area illustrates the complexity of faulting.
Along the northern edges of the Tarim Basin, 10 seismo-genic regions were de®ned (Fig. 19). Kinematics analysis shows that the regions are moving northeastward. The high-est velocity is found in the northeastern part of the area. The con®guration of the local structure suggests local control of the stress±strain ®eld (Lukk et al., 1995). Along the Tian-shan Mountain, thrusting with high seismicity mainly controls this area. It also constitutes a deterrent belt beyond which the horizontal motion will be greatly reduced. This area, together with Area F, constitutes the north barrier to absorb the deformation caused by the Indian penetration. Both areas are found to be uplifting (Qin et al., 2000).
Six out of 10 seismogenic regions (regions 114, 115, 117,
118, 119, 120) have exhibited strong earthquake activity (M$7.0) in the past. Two regions (115, 118) are expected to have strong earthquakes for the next 10 years with ªvery high probabilityº, whereas for other two regions (113, 119) ªhigh probabilitiesº are estimated. ªLow seismic activityº has been estimated for region 111 over the next decade. No probability estimation was made for region 112 due to the lack of strong seismic activity for the past two centuries (Table 2).
3.9. Area I (Northwest China area)
This area surrounds the Junggar Basin in the northwes-ternmost China (Fig. 20). In its eastern part is the right-lateral Keketuohai±Ertai fault (Xu and Deng, 1996). This fault extends for about 200 km with a general trend of 3428
(Ge et al., 1996). In its northernmost edge is the Erqisihe fault along the Altay Mountains. The western side of the Junggar Basin is the Daerbute left-lateral fault with consid-erable reversal motion. The northern Tianshan fault is located in its southern part. The Tarim and Junggar basins have been subsiding with nearly 400 m of Quaternary sedi-ments deposited near the southern border of the Junggar Basin (Xinjiang Institute of Geography, 1986; Xu and Deng, 1996). The Erqisi fault in the Altay Mountains has a Holocene dextral slip rate of 5 mm/year and Holocene vertical slip rate of 2 mm/year (Ding, 1991). The kinematics analysis indicates that the largest part of the area is moving northeastward, while a smaller part is found to move in an EES direction (region 129). One possible explanation is that the eastward motion goes along the southern and northern margins of the Tarim Basin. Since it is believed that the northern Eurasia and the central China have hard crust, the motion will concentrate on the northern margin of the Ordos Block. Therefore, the northeastward motion changes
C. Qin et al. / Journal of Asian Earth Sciences 19 (2001) 97±128
