SOME NEW OUTCOMES OF THE INTERMEDIATE TERM EARTHQUAKE PREDICTION IN VIỆT NAM

1CAO ĐÌNH TRIỀU, 2GIULIANO F. PANZA, 2A. PERESAN, 2F. VACCARI,

2F. ROMANELLI, 1NGUYỄN HỮU TUYÊN, 1PHẠM NAM HƯNG,
1LÊ VĂN DŨNG, 1MAI XUÂN BÁCH, 1THÁI ANH TUAN, 3CAO ĐÌNH TRỌNG

 1Institute of Geophysics, VAST, 18 Hoàng Quốc Việt, Cầu Giấy, Hà Nội;
 2Department of Earth Sciences, University of Trieste, Italy;
3Peoples Friendship University of Russia

Abstract: In this paper the authors present some new results in the use of CN, M8, and M8S algorithms for intermediate-term earthquake prediction in Việt Nam. The following conclusions can be made:

     1. Three from 4 events were recognized by CN algorithm with the catalogue of 1964-2000 and one TIP was made since 2001 with the duration of 5 years. This event (M6.2) occured in Oct. 3rd, 2003 (coordinates: 21°27 N; 101°51 E).

     2. One TIP with magnitude of more than 5.5 can be made since 2006 in Extended NW Việt Nam with the catalogue of 1964-2005, using CN algorithm.

     3. M8 algorithm can be effectively used for earthquake prediction with the M7.0+ in SE Asia: the great Sumatra earthquake M9.0 (26/12/2004) was predicted on the basis of earthquake catalogue to the end of 2003 (R= 667 km and R= 3000 km); Java earthquake M7.7 (17/7/2006) was predicted on the basis of earthquake events to the end of 2005 (R= 281 km); and South Taiwan M7.1 (26/12/2006) also has been predicted on the basis of events to the end of 2005.

     4. Extended M8, M8S algorithm is the one that can be effectively used for earthquake prediction in Việt Nam: the Tuần Giáo earthquake M6.7 (24/6/1983) was predicted on the basis of earthquake events to the end of 1980, with the duration of TIP's from 1981 to 1986.

     5. During the time from 2006 to 2010, the danger of having earthquake M6.5+ occurs in the regions of NW Lao and N Hà Giang province, using M8S algorithm.


I. INTRODUCTION

The United States National Research Council, Panel on Earthquake Prediction of the Committee on Seismology suggested the following definition of the earthquake prediction [5]: “An earthquake prediction must specify the expected magnitude range, the geographical area within which it will occur, and the time interval within which it will happen with sufficient precision so that the ultimate success or failure of the prediction can readily be judged. Only by careful recording and analysis of failures as well as successes can the eventual success of the total effort be evaluated and future directions charted. Moreover, scientists should also assign a confidence level to each prediction”.

The stages of earthquake prediction are divided into: term-less prediction of earthquake-prone areas; and prediction of time and location of an earthquake of certain magnitude (Table 1). The Gütenberg-Richter law suggests limiting magnitude range of prediction to about one unit. Otherwise, the statistics would be essentially related to dominating smallest earthquakes.

The most recent researches focused on the maximum earthquake occurrences in Việt Nam territory which was carried out in the publications of Cao Đình Triều and Nguyễn Đình Xuyên [1-4, 13]. The primary results of the International Project cooperation between Institute of Geophysics and Trieste University for 2002-2005 periods defined that the highly seismic risk with the frequencies and intensity may occur in the region with magnitude of about 6.0- 7.0. The lasted results of the 10.EE.1 Project are not only upgrading the new information in studying the seismic hazard, but also more fruitful and significant to launch a new approach in earthquake prediction.

Table 1. Classification of earthquake prediction

Temporal, in years

Spatial, in source zone size L (km)

Long-term 10

Long-range up to 100

Intermediate-term 1

Middle-range 5-10

Short-term 0,01 - 0,10

Narrow 2-3

Immediate 0,001

Exact 1

In this paper the authors describe some new results of the intermediate-term earthquake prediction in Việt Nam on the basis of the CN and M8-M8S algorithms.

II. APPLICATION OF THE CN ALGORITHM FOR INTERMEDIATE-TERM EARTHQUAKE PREDICTION IN VIỆT NAM

It is nowadays generally accepted that the lithosphere behaves as a non-linear system, due to a large number of different processes, which take part in stress redistribution at different scales, and such a complex nature of the seismic phenomenon does not allow for deterministic predictions. The most promising approach seems currently represented by the intermediate-term middle-range earthquake prediction, based on certain variations of the background seismicity preceding large earthquakes. Several possible scenarios of precursory seismic activity have been proposed. Nevertheless, only a few formally defined algorithms allow for a systematic monitoring of seismicity, as well as for a widespread testing of their performances. Among them there is the algorithm CN (Keilis-Borok, 1996 and references herein), which belongs to a family of medium-range intermediate-term earthquake prediction algorithms structured according to a general pattern-recognition scheme and based on a quantitative analysis of the seismic flow [6-17]. The worldwide tests of the algorithm CN, which is applied in about twenty regions with different level of seismic activity, show that it has the capability to predict about 80% of strong earthquakes with alarm periods covering about 20% of the total time interval considered (Rotwain and Novikova, 1999). The area where a strong earthquake is expected has linear dimensions about 100-300 km. The significance level of the obtained results (larger than 95%) substantiates the predictive capability of such algorithm. The intermediate-term earthquake prediction can be used to reduce casualties, economic losses and other damage inflicted by earthquakes. The key to this reduction is the diversity of preparedness and safety measures combined in flexible scenarios of response to a prediction.

The CN algorithm is structured according to a pattern recognition scheme to allow the identification of the time of increased probability (TIP) of strong earthquakes with magnitude above a fixed threshold Mo. The quantification of the seismicity patterns is based on phenomena that are observed in many non-linear systems before collapse (the response to a perturbation increases, becomes more chaotic, and acts at large distances). The theoretical and numerical aspects of earthquake prediction research have been developed in the framework of an existing long-term and fruitful co-operation of an Italian team with Russian colleagues of the IIEPT (International Institute of Earthquake Prediction Theory and Mathematical Geophysics, RAS, Moscow).

 1. Regionalization for CN application

The region for CN application in Việt Nam is defined taking into account the seismotectonic zoning and including regions with high density of epicenters and earthquake of large magnitude:

1. The length dimension is not less than 5L-10L, in which L=L (M0) is the source linear dimension responsible for earthquake with the magnitude M0 (correspoding to the formula of Wells and Coppersmith, 1994).

2. The boundary of the areas coincides with the zones of low seismic activity.

3. The number of earthquake versus magnitude within the defined region, the threshold M0 for selection of events to be predicted is fixed to be M0 = 5.5 and is not less than 3.

2. Catalogue of earthquake for intermediate-term earthquake prediction by CN algorithm

The Việt Nam catalogue (VNCAT catalogue) of earthquakes was compiled based on three data sources:

- Catalogue of historical earthquakes up to 1900.

- Catalogues from macroseismic data, investigating in public documentation (1900-1976).

- Earthquakes recorded from seismic network of Việt Nam (since 1963).

The Gütenberg-Richter distribution shows that the VNCAT catalogue alone cannot used for CN application and it is necessary to make use of different data set to compile an updated catalogue suitable for seismicity monitoring. A cross comparision between VNCAT and the global catalogues from NEIC, NOAA and ISC is perform. The events not reported in VNCAT are incorporated into the Vietnamese national catalogue. The completeness of the resulting catalogue is analyzed. 

Catalogues compared: - NOAA: National Oceanic and Atmospheric Administration, USA (460-1995);

- NEIC: National Earthquake Information Center, USGS, USA (1900-2005);

- ISC: International Seismological Centre, UK, (1964-2001); and

- VNCAT: National Catalogue of Việt Nam, IGP-VAST, Việt Nam (1900-2005).

Region: Lat: 40 N - 240 N; Lon: 1000 E - 1180 E.

Rules for identification of common events: - Time difference: Δt 1.0 minute;

       - Epicentral distance: ΔLa = ΔLon 1.0 degree; and

       - No limitation imposed on magnitude or depth difference: ΔM 9.0; ΔH 999.0 km;

The frequency diagram of the VNCAT-NOAA-NEIC catalogue is analyzed, considering different time interval: 1900-2003, 1900-1975, and 1975-2003. The number of events in the integrated catalogue (ALL) appears generally larger than for the Vietnamese catalogue only, showing an improvement of data at all magnitude, since 1964. The completeness threshold for the catalogue VNCAT-NOAA-NEIC can be confidently fixed at MC = 4.2 since 1974 and MC = 4.4 since 1964. The integrated catalogue, however, appears rather complete also for magnitude M > 3.2 since 1964.

3. CN algorithm

The CN algorithm is based on the set of empirical functions of time to allow a quantitative analysis of the premonitory parttern which can be detected in the seismic flow:

- Variation in the seismic activity;

- Seismic quiescence; and

- Space-time clustering of events.

It allows to identify the TIPs (Time of Increased Probability) for the occurrence of a strong earthquake within a delimited region.

TIP (Time of Increased Probability) is the interval of time when the probability for the occurrence of a strong earthquake, within a delimited region, increase with respect to the normal condition.

Normalization of the functions of seismic flow is necessary to ensure adequate uniform application of the algorithm with the same set of adjustable parameters in regions of different seismic activity.

We use the normalization by minimal magnitude cutoff Mmin, defined by one of the two conditions: - Mmin = M0 - C,  C = Constant

- Mmin such as N (Mmin) = A,     A = Constant rate of activity.


 

 

 

 

 

 

 

 

 

 

 

N1

N3

K

G

Sigma

Smax

Zmax

q

Bmax

Mmin

m3

m2

m2

m2

m1

m1

m1

m2

-

Mmax

-

-

-

-

M0-1

M0-1

M0-1

-

-

Rates of activity: m1 (a = 3.0); m2 (b = 1.4); m3(c = 0.4)


Aftershocks identification: Windows method: An earthquake j is an aftershock of the earthquake i if: t (j) ≥ t (i); M (j) ≤ M (i):

- t(i) - t(j) ≤ T(M(i)) Time difference;

- r(i,j) ≤ R(M(i))      Distance betwwen epicenters; and

- h(i,j) ≤ H(M(i))      Depth difference.

T (M), R (M) and H (M) are numerical functions of the magnitude M of the main shock.


M

3.9

4.0 - 4.4

4.5 - 5.4

5.5 - 6.4

6.5

T(M)*

46

91

182

365

730

R(M)

50 km

(Keilis-Borok et al., 1980)


Area: 5L - 10L (L - the source linear dimension).

Magnitude:     M0 - 3 ≥ MC

Yearly everage number of events with magnitude above completeness >3: N (M ≥ MC) ≥3.

Choose the M0: Return period for events with M ≥ M0 not lower than 6-7 years.

Regionalization based on the seismotectonic model: It has been shown in the Italian territory that a regionalization strickly based on the seismotectonic zoning permit to reduce the spartual uncertainty of prediction. A general reduction of the percentage of total TIPs has been observed as well [14-17].

In the regionalization based on the seismotectonic model, precursors are detected incide seismogenically homogeneous areas associated to a dominating geodynamic process.

Characterization of seismic sources for Việt Nam and surrounding regions: The seismotectonic zoning of Việt Nam considered in this study has been defined by Cao Đình Triều et al. (2003). The area has been divided into 30 seismogenic zones, based on geological and seismotectonic study, gravity anomalies, magnetic anomalies, and seismic activity.

Regionalization for CN application in Việt Nam: The region for CN application in Việt Nam is defined taking into account the seismic zoning and including areas with high density of epicenters and earthquakes of large magnitude.

According to CN rules and taking into account the distribution of the events versus magnitude within the defined region, the threshold M0 for selection of events to be predicted is fixed to be M0 = 5.5.


Table 1. Main shocks of "Extended NW Việt Nam" with the M 5.5 (1964-2000).

Date

Latitude

Longitude

Depth

Magnitude

18 Dec.,1966

22.6

102.3

0

5.6

23 Jan., 1972

23.3

102.3

33

5.6

24 Jun., 1983

21.77

103.4

23

6.7

16 Jun., 1989

20.67

102.36

15

5.9

 

Table 2. Experiment with CN application for NW Việt Nam on the basis of data set to 2000

Time interval 1964-2000; Threshold setting period: 1968-2000

Time

Learning period

Mc

n/N

% TIPs

(N-n)/N %

(h +t) %

1964-2004

1968-2000

3.0

2/4

24.1

50

74.1

1964-2004

1968-2000

3.2

2/4

21.4

50

71.4

1964-2004

1968-2000

3.4

3/4

26.1

25

51.1

1964-2004

1968-2000

3.6

2/4

36.7

50

86.7

Time interval 1974-2000; Threshold setting period: 1978-2000

1974-2000

1978-2000

3.0

1/2

211

50

71.1

1974-2000

1978-2000

3.2

1/2

24.5

50

74.5

1974-2000

1978-2000

3.4

2/2

22.5

0

22.5

1974-2000

1978-2000

3.6

2/2

30.9

0

30.9


First variant for experiment with CN application to predict the eathquake in Extended NW Việt Nam: Using the new compiled earthquake catalogue VNCAT + NEIC + NOAA, a set of experiments is performed considering M0 = 5.5 and:

- Two different time intervals: 1964-2000; 1974-2000; and

- Four different values of the completeness threshold: MC = 3.0, 3.2, 3.4, 3.6

The best is obtained for the period 1974 - 2000 and for MC = 3.4 (3 out of 4 events with M >  5.5 are predicted. with 34.3 of alarm time).

Relatively good results are obtained with MC = 3.4 also for the period 1964-2000.

Second variant for experiment with CN application to predicte the eathquake in Extended NW Việt Nam (interval time 1990-6/2006):

We can make the following conclusions:

+ Catalogue of Earthquake from 1964 to 2000:

- 4 earthquakes: in 1966 - M5.6; in 1972 - M5.6; in 1983 - M6.7; in 1989 - M5.9;

- The earthquake in 1966 is not recognized;

- In 1972, 1983 and 1989 earthquakes are recognized;

- To identify 1 TIP from 2001 to 2005 (MC = 3.2 and 3.4). In 2001, one earthquake with M5.3 occurred in Điện Biên area (coordinates: 21.27 N - 101.51 E) (Fig. 7).


Table 3. Main shocks of Extended NW Việt Nam with the M 5.5 (1990-6/2006).

 Date

Latitude

Longitude

Depth

Magnitude

2001.2.19

21.33

102.85

12

5.3

2003.10.3

21.27

101.51

33

6.2

 

Figure 1. TIPs diagram for the best prediction result of NW Việt Nam
(interval time 1964-2000) from 2001 to 2005 with the magnitude more than 5.5

Figure 2. TIPs diagram for the best prediction result of NW Việt Nam
(interval time 1974-2000) from 2001 to 2005 with the magnitude more than 5.5

Figure 3. TIPs diagram for the best prediction result of NW Việt Nam
 (interval time 1990 - 6/2006) from 2007 to 2012 with the magnitude more than 5.5


+ Catalogue of Earthquake from 1974 to 2000:

- 2 earthquakes: in 1983 - M6.7; in 1989 - M5.9

- In 1983 and 1989 they are recognized;

- To identify 1 TIP from 2001 to 2005 (MC = 3.2 and 3.4). In 2001 one earthquake with M5.3 occurred in Điện Biên area (coordinates: 21.27 N - 101.51 E) (Fig. 2).

+ Catalogue of Earthquake from 1990 to 2006:

1 earthquake: in 2001 - M5.3

- In 2001 it is recognized;

- To identify 1 TIP from 2007 to 2012 (Mc=3.2 and 3.4) (Fig. 3).

III. APPLICATION OF THE M8-M8S ALGORITHMS FOR INTERMEDIATE-TERM EARTHQUAKE PREDICTION IN VIỆT NAM

The M8 algorithm is applied on the global scale for the prediction of strongest events:

Magnitude ≥ 8.0      Area: 667 km radius

Magnitude ≥ 7.5      Area: 427 km radius

The M8 algorithm uses traditional description of a dynamical system adding to a common phase space of rate (N) and rate differential (L) dimenssionless concentration (Z) and a characteristic measure of  clustering (B) to determine middle range predictions.

The MSc algorithm reduces the area of alarm outlining, such an area where the activity, from the beginning of seismic inverse cascade recognized by M8, is continuously high and infrequently drops for a short time. The phenomenon might reflect the narrow-range intermittence of the seismic premonitory rise near the incipient source of main shock [7-13].

The quality of prediction can be defined by using two prediction parameters:

- N0 = n/N is the rate of failure-to-predict; and

- T0 = t/T is the rate of alarm time (Molchan, 1997).

N is the number of strong earthquake occurred during the time period T covered by prediction. The alarm cover altogether the time t and they have missed n strong events.


Figure 4. Recognization result of the application M8S algorithm  for Tuần Giáo M6.7 earthquake (24 June, 1983).

Figure 5. The red colour of the circles is the areas having danger with the occurrence of the earthquakes M6.5+ from 2006 to 2010.


Seismic activity in Việt Nam is not activated and characterized by:

- The epicenter distribution is not widespread, mainly concentrated in the areas of Extended NW Việt Nam;

- Cumulative number of events is not great, smaller than 10, and mainly is from 3 to 6.

- The largest earthquakes occurred in this region have the magnitude not exceeding M7.0.

In this case, the M8S is effectively used for earthquake prediction (M6.5+, R = 192 km).

1. The Tuần Giáo M6.7 earthquake (14 June, 1983) is recognized by M8S with the radius R equal 192 km and the interval time of the catalogue is from 1964 to 1980. The grid board is 10 × 10, and magnitude threshold M6.5+ (Fig. 4).

2. Using M8S algorithm for prediction the areas to be danger with the magnitude M6.5+ in the North Việt Nam after 2005 year. Two areas are predicted to have danger with the occurrence of the earthquakes M6.5+ from 2006 to 2010 by M8S with the radius R equal to 192 km and the interval time of the catalogue is from 1974 to 2005. The grid board is 10 × 10, and M6.5+ (Fig. 5).

IV. CONCLUSIONS

The CN and M8S algorithms have been successfully applied in the Vietnamese territory for intermediate-term earthquake prediction. On the basis of the above obtained results, the following conclusions can be made:

1. Three from four events were recognized by CN algorithm with the catalogue from 1964 to 2000 and one TIP was made in 2001 with the duration 5 years. This event (M6.2) occurred on October 3rd, 2003 (coordinates: 21.27 N - 101.51 E).

2. One TIP with magnitude more than 5.5 can be made from 2006 in Extended NW Việt Nam with the catalogue from 1964 to 2005, using CN algorithm.

3. Extended M8-M8S algorithm is the one that can effectively be used for earthquake prediction in Việt Nam: Tuần Giáo M6.7 (24/6/1983) predicted on the basis of earthquake events to the end of 1980, with the duration of TIP's from 1981 to 1986.

4. During the time from 2006 to 2010, the danger of having earthquake M6.5+ occurs in the regions of NW Lao and N Hà Giang province, using M8S algorithm.

The authors would like to express special thanks to the Italian Ministry of Foreign Affairs, Italian Embassy in Hà Nội, SAND of ICTP, DST of University of Trieste, Việt Nam Ministry of Foreign Affairs, MOST, VAST and IGP of VAST for the help and facilities provided in the process of the project implementation. 

 REFERENCES

1. Cao Đình Triều, Nguyễn Thanh Xuân, 1996. Probability approach for long-term earthquake Ms>6.5 prediction in Tây Bắc region based on the regulation of epicentral distribution. J. of Geology, A/238 : 33-38. Hà Nội.

2. Cao Đình Triều, 1999. Probable approach for long-term earthquakes prediction in Việt Nam based on regulation of epicentral distribution. J. of Geology, A/251 : 14-21. Hà Nội.

3. Cao Đình Triều, G.F. Panza, A. Peresan, F. Vaccari, F. Romanelli, Nguyễn Hữu Tuyên, Phạm Nam Hưng, Lê Văn Dũng, Mai Xuân Bách, 2004. An overview of seismic activity of Việt Nam territory and surrounding areas. Intern. Conf. on Nat. Evir., Sust. Prot. and Cons.: Italy - Việt Nam Coop. Persp..

4. Đặng Thanh Hải, Nguyễn Đức Vinh, Cao Đình Triều, 2002. Long-term earthquake prediction in Lai Châu - Điện Biên area on the basis of time-magnitude model. Vietnamese J. of Sci. and Techn., 40/4 : 45-53. Hà Nội.

5. Gabrielov A.M., Dmitrieva O.E., Keilis-Borok V.I., Kossobokov V.G., Kouznetsov I.V., Levshina T.A., Mirzoev K.M., Molchan G.M., Negmatullaev S.Kh., Pisarenko V.F., Prozoroff A.G., Renehart W., Rotwain I.M., Shebalin P.N., Shnirman M.G., Schreider S.Yu., 1986. Algorithms of long-term earthquake prediction. CERESIS, Lima, Peru, pp.61.

6. Keilis-Borok V.I., Kossobokov V.G., 1988. Premonitory activation of seismic flow: Algorithm M8. Notes of the Worksh. on Glob. Geoph. Inf. with Appl. to Res. in Earthquake Pred. and Red. of Seismic Risk, ICTP, Trieste, pp.17.

7. Keilis-Borok V.I., Kossobokov V.G., 1990. Premonitory activation of seismic flow: Algorithm M8. Phys. Earth and Planet. Inter., 61 : 73-83.

8. Keilis-Borok V.I., Kossobokov V.G., 1990. Times of Increased Probability of Strong Earthquakes (M o 7.5) Diagnosed by Algorithm M8 in Japan and Adjacent Territories, J. Geophys.Res., 95, No. B8, 12413-12422.

9. Keilis-Borok V.I., Soloviev A.A. (Ed.), 2003. Nonlinear dynamics of the lithosphere and earthquake prediction. Springer, Heidelberg.

10. Kossobokov V.G., Healy J.H., Keilis-Borok V.I., Lee W.H.K., 1997. Algorithm for earthquake statistics and prediction. IASPEI Softw. Libr., 6. Seism. Soc. Am., El Cerrito, CA.

11. Kossobokov V.G., Healy J.H., Dewey J.W., 1997. Testing an earthquake prediction algorithm. Pure Appl. Geophys., 149 : 219-232.

12. Kossobokov V.G., Romashkova L.L., Panza G.F., Perecan A., 2002. Stabilizing intermediate-term medium-range earthquake prediction. J. Seism. Earthqu. Eng., 4/2-3 : 11-19.

13. Nguyễn Đình Xuyên, Phạm Đình Nguyên, Phạm Quang Hưng, Nguyễn Ánh Dương, 2003. Long-term earthquake prediction on the basis of the foreshocks rule. J. of Sci. of the Earth, 25/3 : 193-200. Hà Nội.

14. Peresan A., Rotwain I.M., 1998. Analysis and definition of magnitude selection criteria for NEIC (PDE) data, oriented to the compilation of a homogenous up dated  for CN monitoring in Italy. The Abdus Salam Intern. Centre for Theor. Phys., ICTP. Internal report. Trieste. Italy. IC/98/97.

15. A. Peresan, G.F.Panza, G. Costa, 1999. Seimototectonic model and CN earthquake prediction in Italy. Pure and Applied Geophysics, 281-306.

16. A. Peresan, G.F.Panza, G. Costa, 2000. CN algorithm and long-lasting changes in reported magnitudes: The case of Italy. Geophysics J. Int., pp. 425-437.