Vol. 18 No.4 (379~391)

ACTA SEISMOLOGICA SINICA

July, 2005

Article ID: 1000-9116(2005)04-0379-13

Seismicity research in the subregions of Chinese mainland using strain accumulating and releasing model based on G-R relation* MA Hong-sheng l' 2) ( ~ ~ ~ ) ZHANG X i a o - d o n g 2 ) ( ~ )

LIU Jie 2) ( ~ ~) ZHANG Guo-min 2) ( ~ [~) WANG Huil'2)(SE ~ ) WANG Xin-ling z) ( S E ~ - )

1) Institute of Geophysics, China Earthquake Administration, Beijing 100081, China 2) Institute of Earthquake Science, China Earthquake Administration, Beijing 100036, China Abstract According to the deficiency of the strain accumulating and releasing curves and the previous models, the strain-accumulatingrate of the strain accumulating and releasing model has been deduced based on the G-R relation and the empirical formula between energy release and earthquake magnitude, where the strain-accumulating rate is relative independent of the strain-releasing rate. Five typical areas in Chinese mainland are selected on the basis of the hypothesis on active tectonic block, and small earthquakes from 1970 are imported to calculate the annual strain-accumulating rates considering the completeness of historical seismic data. Having introduced the strain-accumulatingrates into the amended model, present strain phases are got. According to the present stages in their own cycles, the future earthquake tendency of each sub-region is discussed. Key words: G-R relation; annual average strain-accumulating rate; strain accumulating and releasing model; present strain phases CLC number: P315.5 Documentcode: A

Introduction Earthquake activity analysis in different sub-regions in Chinese mainland shows that the distribution of strong earthquakes exhibits the different spatial patterns in different periods. Seismic activity in the same earthquake area or seismic belt is self-similar to some extent, and the characteristic in time is some "periodicity-like". It shows that the frequency and intensity are quite different in different periods (SHI et al, 1997). For a seismic region, strain is one of the physical elements, which is most closely to the tectonic stress and the seismic process, and often used in the partition of seismic stages (XIA, 1987). In order to characterize the strain accumulating and releasing in one region, the activity of the strong earthquakes can be divided into several periodic cycles. Each cycle includes active and inactive stages of strain releasing with relatively stable patterns. Based on the principle, many research* Received date: 2004-05-17; revised date: 2004-10-25; accepted date: 2004-12-06. Foundation item: State Key Basic Research Development and Programming Project of China (G19980407) and Social Commonweal Research Project of the Ministry of Science and Technology (2002DIAl0001). Contribution No. 05FE3015, Institute of Geophysics, China Earthquake Administration. E.mail of the first author: mhs @seis.ac.cn

380

ACTASEISMOLOGICASINICA

Vol.18

ers have studied the periods and cycles of strong earthquake activity in Chinese mainland as well as the different sub-regions (QIU and GAO, 1986; HUANG and CHEN, 1996; MA et al, 2002). However, the strain-accumulating rate was deduced by the historical earthquakes in the previous methods. It was the average rate of released strain in a certain time, and the present strain state was obtained by subtracting the practical released strain from the accumulated strain. Due to the incomplete historical earthquake records, the strain-accumulating rate is quite inaccurate. Besides, the original points and the terminal points of the strain accumulating and releasing curves are either at zero or at the same level. Therefore, all those methods have deficiency in correlation between the strain accumulating rate and the releasing rate. In this paper, the strain-accumulating rate is firstly deduced from G-R relation between annual frequency and earthquake magnitude as well as the empirical formula between energy release and earthquake magnitude. So the strain-accumulating rate is independent of the strain-releasing rate. In the calculation, small earthquakes recorded from 1970 were used to calculate the strain-accumulating rate. Finally, based on the hypothesis of the active tectonic blocks and the division of the active blocks (ZHANG and ZHAN~ 2000; ZHANG et al, 2003), several typical areas in Chinese mainland are selected to check the reliability of the modified model. According to the present strain stages, the future earthquake tendency of each sub-region is discussed.

1 The strain accumulating and releasing model based on G-R relation During the process of crust movement, the strain of the active tectonic block, which is equivalent to a seismogenic system, is accumulated with time. When the strain reaches to the peak strength of the rock, the strain will be released in the form of earthquakes so as to arrive to the next balance. Generally, the total accumulated strain should be corresponding to the released one in one or several complete seismic cycles. The seismic inactive phase is characterized by that the accumulated strain is much greater than the released one; On the contrary, the active phase is that the released strain is much greater than the accumulated. However, since the recurrence period of strong earthquakes is quite long and many historical strong earthquakes were not recorded in the limited catalogs, it is difficult to give out one or several complete seismic cycles in most regions. The strain-accumulating rate obtained from historical earthquakes cannot reflect the real rate of the whole cycle. To work out this problem, G-R relation and the empirical formula between energy release and earthquake magnitude are used to deduce a strain-accumulating rate, which is only dependent on parameters a and b values. Due to the relative shorter recurrence period of small earthquakes, it may take place many times according to their catalogs. The present-day small earthquakes are introduced into calculating a and b. The G-R relation is widely used in seismicity researches. It means that there exists a certain relationship between the frequency of large and small earthquakes and the magnitude. If the frequency is displaced by the annual-frequency, the G-R relation is also tenable. Thus, the relatively independent and true strain-accumulating rate is available, and it can be used into the strain accumulating and releasing model. 1.1 The strain-accumulating rate deduced by G-R relation The released strain of one earthquake is based on the empirical formula between energy and magnitude (Gutenberg and Richter, 1956; Kanamori, 1977): lgE = d M s + c

(1)

where, E means the energy in unit of J, and Ms is the surface wave magnitude. The strain is the

No.4

MA Hong-shenget al: STRAINACCUMULATING-RELEASINGMODEL

381

square root of the energy. It is needed to point out that the released strain ~ represented by E m is an approximate expression (Benioff, 1951). Actually, E ~ / t e~/2, where/z is elastic coefficient. In our study,/t is taken as a constant for one block. It is different for the different blocks, showing the relative value of strain releasing of different-magnitude earthquakes in different blocks. Ei u2 means the released strain of the i-th earthquake. As the area is large enough, the frequency of different earthquake magnitude in this region agrees to G-R relation (Gutenberg and Richter, 1941): lg N = a - b M

(2)

where N is the number of earthquakes with magnitude between M and M+AM. Some small earthquakes were lost due to the uneven station distributions, while some strong earthquakes may be also lost because the recurrence period is too long to be recorded. So we adopt annual-frequency ?'(M, T, t) to avoid the trouble: z(M, T, t) -

N ( M , T, t)

(3)

T

where, T is the time-scale of the catalog, t is the initial time, and N ( M , T, t) is the earthquake number of magnitude between M and M+AM in scale T. The relationship between ?'(M, T, t) and M is similar to the G-R relation (LIU et al, 1997), namely: lg 2z(M, T, t) = a - b M

(4)

Then, the relation between annual-frequency ?'(M, T, t) and energy can be written as: (5)

lg~M,T,t)=a-b(lgE-c)

Taking differential to both sides, we have dy _ 2/

b dE d E

(6)

The more the frequency of the earthquakes, the smaller the magnitude of the earthquakes is. So, when M2<_M<_M1,the annual average strain of the corresponding magnitude interval is written as: 10 a-bM = Ier:(M=M:)~ r-l E d y = fe:(M=U:) b ~/ dE = - b f ~: 100.5(c+aM) 10c+dMdM J•(M=Mt) JEI(M=M') d ~ aM1

(7)

Its integral formula is as follows: ~N:(M=M:)~

.

b

= JuI(M=M,)~//za)" = (b - 0.5d)

l O(a+O,Sc)+(O.5d-b)MM=M:

M=M,

(8)

where, 74 and ?i denote the annual-frequencies of magnitude M1 and M2 respectively; value a and b can be gotten from formula (4), and values c and d were reported by the previous study, generally, c=4.8, d=l.5 (Department of Earthquake Disaster Prevention, State Seismological Bureau, 1992). It needs to be pointed out that the strain-accumulating rate is calculated by the annual-frequencies of different magnitude interval. 1.2 The values a and b in G-R relation In this paper, combining the historical earthquake catalogs with present small earthquake catalogs, the values a and b are calculated. To get a real strain-accumulating rate, it must confirm

382

ACTA SEISMOLOGICAStNICA

Vol.18

that values a and b should be accurate, i.e., the accurate data, reasonable calculation method and enough sample numbers are needed (CHEN et al, 2003). Generally, these three conditions cannot be guaranteed, so it is unavoidable to lead to the quite large error or even affect the final results of the model. Generally, the historical earthquakes, namely the strong earthquakes, are used to calculate the parameters a and b in a large area and a long time scale. Surface wave magnitude Ms is often used in our study. When Ms is between 6 and 8, the error will be very small. When Ms is larger than 8, there would be large error in the results because of magnitude being saturated and the fitting curve of G-R relation being non-linear. When Ms is smaller than 6, there would also be large error in the results because of station being absent and surface wave dispersing deficiently. For the smaller earthquakes, the lg 7~M curve will be more non-linear owing to the limited data. While the small earthquakes or even micro-earthquakes are used to calculate values a and b in a small area in a short time, ME is usually used. When ML is between 2.5 and 4.5, the error wilt be quite small. When ME is larger than 4.5, the error would be large because of near-station record exceeding its record range. When ME is smaller than 2.5, the data will be half-baked and the curve of lg 7:M will be non-linear any more (CHEN et al, 2003). In order to make fully use of these two kinds of data, annual-frequency ~M, T, t) is imported (select annual-frequency of different earthquake interval with enough numbers) so that the two kinds of data are used together in calculating values a and b. The least squares method or maximum likelihood method is applied. For least squares method, all the data are participated in calculation with equal weight. But, for the smaller earthquakes being half-baked, annual-frequency will be smaller. The lower limit of maximum likelihood method is uncertain (CHEN et at, 2003). In this paper, we select least squares method. As the Ms and ML are different magnitude grade, the values a and b from Ms and ME cannot be directly compared with each other. In this study, the formula of Ms= 1.18ML-1.08 (Department of Earthquake Disaster Prevention, State Seismological Bureau, 1992) is applied to transform ME into Ms; the lower limit of magnitude is determined by the monitoring capability of earthquake station-network. When the lgT~M curve is plotted, the minimal magnitude is selected at the point where the curve turns flat for historical earthquakes or small earthquakes. The data obviously deviating from the lg),M curve are abandoned. The upper limit of magnitude is selected at the point where the frequency is more than 5 at least, thus it can ensure that there are enough number in determining the recurrence period of earthquake (LIU, 1998). Compared with the previous models, the strain-accumulating rate is more accurate because value a and b are obtained by both historical earthquakes and small earthquakes, and the rate is independent of the strain-releasing rate calculated from historical earthquakes. 1.3 The model based on G-R relation When the strain-accumulating rate is available, the strain of present stage is calculated on the basis of the historical earthquakes and its complete research information (HUANG et al, 1994). Present strain energy can be expressed by detracting the practical releasing strain E from accumulated strain ~t, namely: N(O i=1

fK

(9)

Thus, we got the ~0, which presents the strain level at time t. N(t) is the number of earthquakes

No.4

MA Hong-sheng et al: STRAINACCUMULATING-RELEASINGMODEL

383

occurred before time t. According to equation (9), the curve of ~t) with time t was plotted. Based on some basic data, it shows the periodic fluctuations of earthquake activities. It needs to be pointed out that the ~t) is a relative value. The inactive phase was chosen as the starting time of this model because there are fewer strong earthquakes in the inactive phase and the strain can be accumulated. When the strain is accumulated to some extent, earthquakes begin to occur and strain reaches to their summit. To some extent, the strain releasing in the active stage is related to what has accumulated in the inactive stage. So, it is quite reasonable to take the inactive phase as the beginning point of one seismic cycle (SHI et al, 1974). How to select the strain initial stage in the model is a big problem. Generally, the beginning stage of the strain is set to be zero. If there is still any strain residual at the beginning stage, it will make the strain of some time negative, affecting the estimation of the future earthquake activity. On the other hand, since the periodic fluctuations of strain accumulating and releasing with time t is unchanged, the shape of the curve cannot change with time t. The beginning stage of the strain can only influence the idiographic value at time t. There are only a constant difference between the selection stage and the real stage. To solve this problem, we choose the temunation point of the largest release, namely the neap of the curve, as zero point. In fact, the strain before zero point can also be inversely output by this method. On the condition of the data being relatively complete, it was proposed that the total accumulated strain be equal to the total released strain in an independent and complete system over one or several seismicity cycles. But the catalogs in many areas are half-baked, and the data are too few to create one or several complete seismic cycles. In our study, the accumulated strain is not just equal to the total released strain. It will be discussed in details in the following parts.

2 The research areas and treatment of the seismic data During the study of the strain accumulating and releasing of an area, the completeness of the seismic data must be considered. It involves two aspects: the relative completeness on time for the data that cover at least one complete seismic cycle, and the relative completeness on space for the data that match with the seismogenic tectonic block (seismic zone) on space. By analyzing the characteristics of earthquakes sequence in time, the seismicity has the characteristics of relative inactiveness and activeness by turns. So when we consider the completeness of the seismic data, this characteristic is known. Based on the hypothesis of the active tectonic blocks and the division of the active boundaries (ZHANG and ZHANG~ 2000; ZHANG et al, 2003), Five areas are selected in Chinese mainland as the studying areas, including the North China, Tianshan, Southeast Seashore, Northwest China and Sichuan-Yunnan (Chuan-Dian) respectively (Figure 1). Thereby, the range of North China is from longitude ll0°E to 124°E, and the latitude from 34°E to 42°E. The Southeast Seashore mainly consists of the active-tectonic block region in South China where most strong earthquakes had happened. The Chuan-Dian area includes Chuan-Dian rhombic block and its boundaries, where earthquakes take place frequently and the tectonic background is more complex. The range of Northwest China primarily includes the active-tectonic blocks of Qaidam and Qilian. The Tianshan area mainly consists of the active tectonic block of Tianshan and its two boundaries within Chinese mainland. The seismic data are chosen from Catalogue of Chinese Historical Strong Earthquakes (Department of Earthquake Disaster Prevention, State Seismological Bureau, 1995), Catalogue of

384

ACTA SEISMOLOGICA SINICA 70°E

80"

90°

loft

I lff

120"

Vol. 18 130°

140"

4fiN

30°

20"

10"

Ii0 °

Figure 1

120"

The active tectonic blocks and the studied areas ( Z H A N G et al, 2003)

Chinese Modern Earthquakes (Department of Earthquake Disaster Prevention, China Seismological Bureau, 1999), and Earthquake Short-Bulletin Catalogue of Chinese Digital Seismic Network Center ° . After getting these seismic data, aftershocks are deleted by removing aftershocks of some areas for some time after the main shock. If the two earthquakes with serial number i and j can satisfy the following formulas: tj - t i <_T(MI)

A~ <-R(Mi)

Mj <_Mi

(I0)

where thej-th earthquake is the aftershock of the i-th earthquake, AUis the distance between these two earthquakes, and T(Mi) and R(Mi) can be got by empirical relation (Keilis-Borok and Knopoff, 1980).

3 Results and analysis Both G-R relation and the process of the strain accumulating and releasing depend on the complete catalog. However since the seismogenic surrounding, earthquake historical record and present small earthquake information in different areas are not the same, the different starting time are adopted to calculate and analyze the values a, b. If the strain accumulating rate deduced by values a and b is precise and objective, the intensity and magnitude of the future earthquakes can be given half-quantitatively by analyzing the accumulated strain and its stage in the whole cycle, which does not need one or more earthquake cycles (MA et al, 2002). When calculating the annual strain-accumulating rate, Ms=8.5 is selected as the upper limit and Ms=6.0 as the lower limit. The ®

Chinese Digital Seismic Network Center. 2003. Earthquake Short-bulletin Catalogue of Chinese Digital Seismic Network Center.

No.4

MA Hong-shenget al: STRAINACCUMULATING-RELEASINGMODEL

385

selection of lower limit owes to the completeness of the historical earthquakes. In North China area, the catalog of Ms>6.0 has been relative complete since 1480 (HUANG et al, 1994). One Ms=8.0 earthquake occurred in AD 1303, and in AD 1556 another Ms=8.3 happened. When choosing the starting time, the strain-accumulating process and the influence of large releasing are considered. It is reasonable to regard the year AD 1400 as the starting time, and the lowest point of the curve as the year 1695. Figure 2 shows M-t curve of magnitude over 6.0 in North China area from AD 1400, earthquake strain accumulating and releasing curves of magnitude over 6.0, and magnitude-annual frequency plot (G-R relation). From the lgyversus M relation, the value of a is 3.50, and b is 0.74. Substituting values a, b into formula (8), the strain accumulating rate of North China area is ~ =4.00×106 jl/2/a. In Figure 2, the accumulating-releasing curve approximately reflects two inactive ~ active cycles, which last up to 600 years. In the first cycle, the strain had been fully released near the year 1695, and reached to its lowest. The sum covers 300 years from accumulating to releasing. In the second cycle, the strain has been almost completely released from 1696 to 1976. The trend of the curve shows us the recent strain is in an accumulating stage of a new cycle, and the accumulated strain is little. So we can assume that there would not occur strong earthquakes in recent years. 80[ (a)

(c) <4

~ , ~ K l g 7 =(3, 50+_0. 16)-(0. 744-0. 03)Ms

1400

1500

1600

1700 Year

18(/0

1900

2000

C ~c

' (b)

1 0 Largeearthquakein 1400~2003

14t)0

J

1500

2.0

1600

! 700 Year

1800

[900

3.0

4.0

5.0 M

6,0

o ~,,._

7.0

8.1)

2000

Figure 2 The strain releasing curve (a), the M-t plot (b) and the G-R relation (c) in North China area In Southeast Seashore, since there have happened three strong earthquakes above Ms=7 from 1650 to 1700 and there had few earthquakes before 1800, 1800 is chosen as the starting time as shown in Figure 3. From the l g y M relation, we obtained a =2.96, and b= 0.73. Substituting a, b value into formula (8), the strain accumulating rate of Southeast Seashore is ~ =1.46x106 Jln/a. In Figure 3a, the accumtflating-releasing curve also approximately reflects two inactive---> active cycles, and the first cycle lasts from 1800 to 1920, and the second cycle is from 1921 to 1994. Near the time of 1994, its strain reached to its minimum, i.e., O. From the shape of the curve, we can also see that the recent strain is in the accumulating stage of a new cycle, and the accumulated strain is even fewer. So there will not occur strong earthquakes in recent years, too. In Northwest China, the historical seismic data before 1900 is seriously missed (HUANG et al, 1994). If the earthquake of Ms=8 in 1879 is regarded as the end of an active phase of strain

386

ACTA

SEISMOLOGICA

SINICA

Vol. 18

12 t (a) (e) (e) (2. 96+0. 17)-(0. 73_+0. 04)Ms

)

1. O. 1800

1850

1900

1950

2000

~

Year

0 -0, 5 -1.0

• Sma9 each uqak¢ in 1970~ 2002

oN,,~o

-1,5

~7

2.0

1400

15'00

Ill1600 , ,1700

1800

, h, 1900

3,0

5.0

4.0 M

6.0

2000

Year

Figure 3

The

strain releasing curve (a), the M-t plot (b) and the G-R relation (c) in Southeast Seashore 9 - (c)

12

(a) 8

10

~

~7

6

880

1900

1920

1940

1960

1980

2000

1880

Year 5

1900

1920

1940 Year

1960

1980

2000

2,0

(b)

1.6

4

L

~3

~94o 19~o 1960 ~970 19~0 19'90 20bo Year Figure 4

The

(d)~ lg}=(3.15~0. (17)-(0. 6t 4-_0.02)Ms 1.2 0.8 ~0.4 0 . o Largeearthquakein " ~ • 1400~2003 -0. 4 * Smallearthquakein ~ 1970~20ff2 , , "NNK ' -0. 8 3 4 5 6 7 M

strain releasing curve ( a , b ) , t h e M-t p l o t ( c ) and the G-R relation (d) in Northwest China

releasing, it experienced two cycles from 1880 to 1940 and from 1941 till now in the past 120 years or so. So, it is quite reasonable to put the year 1880 as the starting time of the strain curve, but the starting time of the lgTzM relation begins from 1900 in this area. Using the same way, we get the value of a is 3.16, and b is 0.61. Bringing a, b value into formula (8), the strain accumulating rate of Northwest China area is ~=l.4x107 JU2/a. In Figure 4a, besides the large releasing in 2001, there is still an active stage between 1920 and 1940, and the zero point of the strain curve just takes place in 1940 or so. For this, we chose the year 1940 as the starting time and draw another curve shown in Figure 4b. The strain accumulating-releasing curves of Figure 4a and 4b display that the strain of Northwest China area is confronted with large releasing since the earthquake of Ms=8.1 in 2001, and its remainder is about 1.6xl08 fin, which equals to the strain released energy of one strong earthquake of Ms=7.7 or three strong earthquakes of Ms=7 occur.

No.4

MA Hong-shenget al: STRAINACCUMULATING-RELEASINGMODEL

387

In Tianshan area, the historical seismic data after 1900 is complete (HUANG et al, 1994). For the sake of the influence of the two strong earthquakes of Ms=8 in 1812 and 1902, the year 1860 was chosen as the starting time of the strain curve. From the lgTzM relation, the value of a is 3.50, and b is 0.67. Bringing a, b values into formula (8), the strain accumulating rate of Tianshan area is ~=l.18x107 J1/2/a. In Figure 5a, at the time of 1920, the strain reached to the lowest and then the strain has been accumulated for quite a long time. It is thought that the strain of Tianshan area confronts with large releasing in future, and its total is about 3. lxl08 j1/2. According to the former cycle, it may be presumed there is one strong earthquake of Ms=8.1 or seven strong earthquakes of Ms=7 in the future 15 years. (a)

7 6 % ~z

4 3 2 1 0 1860

8~

18'8/) 1900 19211 1940 Year

1960

(c)

,

i,

184~-1~6o l~8O 1ooo

198(I 2000

, , l ll,i,itl,.l.

1920 1940 1960 198(/ 2000 Year

(b) 1,s j~2

c"

("'

.. 996

~ l g r = ( 3 . 50+_0. lO) - (0, 67 +_.o,02)Ms

:" e~ 0.8 o Large earthquake in ~ 1900~2003 Small earthquake in 1970~2002

I.

1920

1940

1960 Year

1980

2000

-(I, 8

2

3

4

M

NK. 5

6

7

Figure 5 The strain releasing curve (a, b), the M-t plot (c) and the G-R relation (d) of Tianshan area In area of Chuan-Dian rhombic block, the historical seismic data after 1900 is complete (HUANG et al, 1994). From the l g y - M relation, the value of a is 2.90 and b is 0.56. Using the same method, it can be calculated the strain accumulating rate of Chuan-Dian area is ~ =1.86x107 JV2/a. Figure 6a and 6c show that the strain had completed a releasing phase by the time of 1940, and the seismic data with magnitude 5-7 were obviously missed before 1940. So, we chose 1940 as the starting time and draw another curve shown in Figure 6b. At the time of 1975, the strain reached to the lowest and then the strain has been accumulating till now. It is believed that the strain of Chuan-Dian area also faces large releasing in the future, and its total is about 3.0><108 j1/2, which equals to one strong earthquake of Ms=8 or seven strong earthquakes with Ms=7 occur. The strain accumulating rates of these five areas demonstrate that the whole level of strong earthquake activity in Chinese mainland is higher in the west than that in the east. The recurrence period of strong earthquake activity is shorter in the west than that in the east. These phenomena are consistent with the present earth crustal strain rate in the east and in the west of China. Besides, the strain stages in west areas of China are in the high strain accumulation and releasing stage, of which the area of Northwest China has been in releasing stage marked by the Kunlun Mountain (Kunlunshan) earthquake with Ms=8.1 in 2001. In the future time period, there are the big strain releasing in Xinjiang area and the Chuan-Dian diamond block. The seismic cycles in the East China area should cover more than 300 years, of which the seismic cycles in Southeast Seashore

388

ACTA SEISMOLOGICA SINICA

Vol. 18

5 4

~3 ~2 t 900

1920

1940

1960

1980

2000

Year

:t

I, t

1860

1880

1900

1920 1940 Year

1960

1980

2000

(b) " ~ . ~g~.=(2, 90_+0. 25)-(0. ,56-L-_o. 05)Ms

0 04 • 1940

Figure 6

1950

1960

1970 Year

1980

1990

2000

" ° Large earthquake in ;900 -- 2 0 0 3 ~ * Small earthquake in 1970~2002 2

3

4

M

5

6

7

The strain releasing curve (a, b), the M-t plot (c) and the G-R relation (d) in Sichuan-Yunnan area

(MA et al, 2002) still contains the obvious secondary small seismic cycles.

4 Discussion and conclusions According to the strain accumulating and releasing, the activity of strong earthquakes could be divided into several periodic cycles. The strain accumulating and releasing with time characterizes the periodic activity of strong earthquakes. In the previous methods, only when the preconditions are satisfied, for example, there must be one or several cycles in one integrated-tectonic area, the accumulated stain could be equal to the released strain. However, the catalogs of many areas are half-baked, and there hardly exists one or several complete seismic cycles. Besides, the original points and the terminal points of the strain accumulating and releasing curves are either at zero or at the same level that makes the trend of the strain curve apart from reality and influences the estimation of the future earthquakes. In our study, since the strain accumulating-rate is independent of the strain releasing-rate, the strain accumulating-rate can be less affected by the absence of the historical earthquakes. Moreover, so long as the small earthquake catalog of one area is complete with only one strong earthquake activity cycle, the intensity and magnitude of the future earthquakes can be half-quantitatively estimated. In the present study, based on the hypothesis of the active crustal blocks, we have divided Chinese mainland into several sub-regions in seismicity, thus the rationality of the division can be believed. In addition, annual-frequency z(M, T, t) is adopted to calculate values a and b. At the same time, small earthquakes with the completeness are used in calculating values a and b, which is numerousness in catalogs. This method ensures values a and b objective and exact, which is quite important for calculation of the strain-accumulating rate. When calculating the strain-accumulating rate (formula 8), the conception of magnitude interval is introduced. The upper limit is Ms=8.5 and the lower limit is Ms=6.0. So the strain-accumulating rate is not the rate of the whole area but the rate of the appointed magnitude interval. So, the rate is the annual accumulated rate of Ms>6 in this study. When considering the strain releasing, it is found that the released strain of each magnitude interval is closely related to the value b. When b--0.75, the released strain of each magnitude inter-

No.4

MA Hong-shenget al: STRAINACCUMULATING-RELEASINGMODEL

389

val is equivalent; When b>0.75, the released strain of small magnitude interval is larger; and When b<0.75, the released strain of large magnitude interval is larger. This relation indicates the strain-accumulating rate is correlative with the selection of the magnitude interval. Usually, there are some uncertainties in the calculated results. But in our study, the uncertainties are mainly resulted from errors of the values a and b in G-R relation and the selection of the starting time in the model. Taking North China area as an example, these questions are further discussed in the following section. 4.1 The influence of the starting time on the model In Figure 2a, the accumulating-releasing curve approximately reflects two complete earthquake periods. To study the effects of the starting time on the model, we choose the year 1700 and 1800 as the staring time respectively as well as keeping the strain-accumulating rate unchanged, the result was shown in Figure 7a, 7b. It is necessary to point out that the year 1700 is about the beginning time of the latter cycles and the year 1800 is the time when the strain had been accumulated for quite a long time. Comparing Figure 2a with Figures 7a, 7b, it can be found that the second cycle of Figure 7a is much similar to that of Figure 2a. The recent accumulated strain is also similar to each other, namely e=3.0xl08 j1/2 This example also demonstrates that the recent strain can be gotten by the model only using the last large strain released time as original time. But the recent strain of Figure 7b is quite different from Figure 2a because the accumulated strain at the beginning of the inactive phase has been missed that makes a systematic error between the calculated strain and the actual strain. The total released strain is larger than the accumulated, resulting in the future earthquakes of estimation being lower than the actual ones. So, when selecting the starting time, it is necessary to consider not only the data completeness, but also the strain accumulating time before the active stage of the earthquake.

17t0

1770

1830 1 8 9 0 1 9 5 0 2010 Year

12[ (c) lo'

1710

~-~

1800

..-4

1770

1830 1890 Year

0~ . . . . . . . . . 1710 1 7 7 0 1830 1890 Year

1710

1 9 6 0 2000

1770

1830 1890 Year

1 9 5 0 2010

f (t)

12

tJ¢~ 1 9 5 0 2010

1 8 8 0 t920 Year

~ ~ (d)

1 9 5 0 2010

5~(e)

t840

0

1710

~ 1 7 7 0 1 8 3 0 t890 Year

1 9 5 0 2010

Figure 7 Analysisof the uncertaintyin the amendedmodel for North China area (a, b) The influence of starting time; (c, d) The influence of a value; (e, f) The influence of b value

Vol. 18 No.4 (392~401)

ACTA SEISMOLOGICA SINICA

July, 2005

Article ID: 1000-9116(2005)04-0392-10

An overview on shallow strong earthquake activity and earthquake live losses of Chinese mainland in the centenary from 1901 to 2001" FU Zheng-xiang 1) ( ~ J ~ ) D I N G Xiang 1~( T

L I U Gui-ping 2) ( ~ I j ~ . ~ )

S H A O Hui-cheng 3'4) ( ~ [ ~ ) ~ )

~)

1) Institute of Earthquake Science, China Earthquake Administration, Beijing 100036, China 2) Department of Earthquake Monitoring and Prediction, China Earthquake Administration, Beijing 100036, China 3) Earthquake Administration of Shaanxi Province, Xi'an 710068, China 4) Department of Engineering Mechanics, Northwest Industrial University, Xi'an 710072, China Abstract

This paper reviews and analyses briefly the general characteristics of shallow strong earthquakes (Ms>_6.0, focal depth h<70 km) in space, time and magnitude and earthquake fatalities of Chinese mainland in the centenary from t901 to 2001. During the period from 1901 to 2001, there occurred about 420 strong shallow earthquakes with magnitude Ms_>6.0, 88% of them occurred in the western part of the Chinese mainland, which might be related to the strong deformation and motion of the active blocks in the western part. The average focal depth (25 kin) in the western part is deeper than that (16 kin) in the eastern part, which might be related to obviously thicker crust in the western part. The inhomogeneous distribution of focal depths with the depth profile is related to the variation of frictional and theologic characteristics with depth in the crust. The shallow strong earthquake activity of Chinese mainland shows a tempo-spatial clustering process. The relation between earthquake magnitude Ms and cumulative frequency Nc is lgNc=8.64-0.99Ms. About 600 000 people died in the earthquakes of Chinese mainland from 1901 to 2001, the most serious earthquake live losses occurred in Hebei Province (250 723 people died) and Ningxia Hui Autonomous Region (246 269 died). There is no a statistically linear relationship between earthquake live losses and magnitudes. The statistical relationship between the earthquake deaths D and cumulative frequency N~.is lgN¢=2.40-0.391gD, which shows a fractal distribution. Key words: Chinese mainland; shallow earthquake activity; live losses; 1901-2001 CLC number: P315.5, P315.9 Document code: A

Introduction It is well k n o w n that China locates in the southeastern part o f Eurasian Plate, which subjects to the action o f westward subducting o f Pacific Plate and Philippine Sea Plate, and northward collision o f Indian Hate. So China is a country with strong seismicity. There are two kinds o f earth-

* Receiveddate: 2004-02-03;reviseddate: 2004-04-15;accepteddate: 2005-04-15. Foundation item: Social CommonwealResearchProject of the Ministry of Science and Technique(2004DIA3J010)and Joint SeismologicalScienceFoundationof China ( 104016). E-mail of the first author: [email protected]

No.4

M A Hong-sheng et al: STRAIN A C C U M U L A T I N G - R E L E A S I N G M O D E L

391

LIU Jie, CHEN Yong, CHEN Ling et al. 1997. Global seismic hazard assessment [J]. Earthquake, 17(Suppl): 2-10 (in Chinese). MA Hong-sheng, LIU Jie, ZHANG Guo-min et al. 2002, The seismicity research in the subregions of Chinese mainland using strain accumulating and releasing model [J]. Acta Seismologica Sinica, 15(6): 595~604. QIU Jing-nan and GAO Xu. 1986. The discussion of the seismicity phases of the Chinese mainland [J]. Earthquake, (6): 41~47 (in Chinese)~ SHI Zhen-liang, HUAN Wen-lin, CAO Xin-ling et aL 1974. Some characters of the seismicity in Chinese mainland [J]. Chinese J Geophys, 17(1): 1-13 (in Chinese). SHI Zhen-liang, WANG Liang-mou, FU Zheng-xiang et aL 1997. Research of Strong Earthquake Hazard Long-term and Medium-term Prediction in Chinese Mainland [M]. Beijing: Ocean Press, 1-174 (in Chinese). XIA Hao-ming. 1987. The character and the meaning of the seismicity phases division with Ms_>4in North China subregion [J]. Chinese J Geophys, 30(3): 281-291 (in Chinese). ZHANG Guo-min and ZHANG Pei-zhen. 2000. Academic progress on the mechanism and forecast for continental strong earthquake in the first two years [J]. China Basic Science, (10): 4-10 (in Chinese). ZHANG Pei-zhen, DENG Qi-dong, ZHANG Guo-min et al. 2003. Active tectonic blocks and strong earthquakes in continent of China [J]. Science in China (Series D), 46(Suppl): 13-24.