23
The The basic basic results results and and prospects prospects of of MEE MEE algorithm algorithm for for the the medium-term medium-term forecast forecast of of earthquakes earthquakes Alexey Zavyalov Schmidt Institute of Physics of the Earth Russian Academy of Science Str. B.Gruzinskaya, 10, Moscow 123995, Russia, [email protected]

Alexey Zavyalov Schmidt Institute of Physics of the Earth Russian Academy of Science

  • Upload
    tilly

  • View
    36

  • Download
    0

Embed Size (px)

DESCRIPTION

The basic results and prospects of MEE algorithm for the medium-term forecast of earthquakes. Alexey Zavyalov Schmidt Institute of Physics of the Earth Russian Academy of Science Str. B.Gruzinskaya, 10, Moscow 123995 , Russia, zavyalov @ ifz . ru. INTRODUCTION. - PowerPoint PPT Presentation

Citation preview

Page 1: Alexey Zavyalov Schmidt Institute of Physics of the Earth Russian Academy of Science

The The basic basic results results and and prospects prospects of of MEE MEE algorithm algorithm

forfor the the medium-term medium-term forecastforecast ofof earthquakes earthquakes

Alexey Zavyalov

Schmidt Institute of Physics of the EarthRussian Academy of Science

Str. B.Gruzinskaya, 10, Moscow 123995, Russia, [email protected]

Page 2: Alexey Zavyalov Schmidt Institute of Physics of the Earth Russian Academy of Science

INTRODUCTIONThe International Geophysical Year 1957 has given a

powerful pulse to development of geophysical researches in the USSR. Under its aegis detailed seismological supervision in sesmoactive areas of Soviet Union: on the Far East (east coast of Kamchatka, Kuriles), Caucasus, the Baikal region have been started.

One of the important results of Geophysical Year for seismology was creation of a seismic stations network which became a basis of the future Uniform Network of Seismic Supervisions of USSR (UNSS), nowadays functioning within the framework of Geophysical Service of Russian Academy of Science. Earthquake catalogues of UNSS/GS RAS of different level of detail became the basic information base for revealing regularities of seismic process and development of various algorithms for strong earthquakes forecast.

Page 3: Alexey Zavyalov Schmidt Institute of Physics of the Earth Russian Academy of Science

The Map of Expected Earthquakes (MEE) algorithm was established in 1984 by G. Sobolev, T. Chelidze, A. Zavyalov and L. Slavina on the basis of ideas about failure process of rocks and geological medium as a self-similar and self-organizing system of blocks of different scales. Based on the kinetic conception of strength of solid materials authors have made the image of anomaly behavior of different seismological parameters before strong (M5.5) earthquakes.

Page 4: Alexey Zavyalov Schmidt Institute of Physics of the Earth Russian Academy of Science

MEE algorithm uses the principle of space-time scanning of the earthquake catalog within the limits of the studied seismoactive region.

Using the Bayesian approach maps of conditional probability distribution of strong earthquake occurrence P(D1|K) where calculated. These maps were named as Maps of Expected Earthquakes (MEE).

During last twenty years MEE algorithm have been tested on regional earthquake catalogs of Caucasus, Kamchatka, Kuril Arch, Kopet-Dag, Kirgizstan, Southern California, North-East and South-West China, Greece and Western Turkey.

Page 5: Alexey Zavyalov Schmidt Institute of Physics of the Earth Russian Academy of Science

ChoiceChoice of of precursorsprecursorsclear physical meaning of precursors;physically substantiated relation of each precursor to the earthquake preparation process; availability of observation data for each precursor; these data should be representative in time (long-terms series of prognostic parameter values) and in space (the possibility of their mapping);availability of a formal procedure for the identification of anomalies in prognostic parameters based on a model of their behavior during the earthquake preparation;possibility of obtaining estimates for retrospective statistical characteristics of each precursor: probability of successful prediction (probability of detection), probability of a false alarm, prognostic efficiency (informativeness), and so on.

P recu rso rs

Q u a si -sta tio n a ry D y n a m ic

N u m b er o f tec to n ic fau lts ;

N u m b er o f fau lt c ro s s in g s ;

G rad ien t o f sp eed o f m o d ern m o v e m en ts ;

A n o m alie s o f a g rav it a tio n a l f ie ld

an d o th e rs

D en sity o f se i sm o g en ic fau lts K sf ;

R ecu rren ce p lo t s lo p e (b -va lue );

N u m b er o f se ism ic ev en ts p e r u n it tim e N ;

R eleased se ism ic e n e rg y 3/2E ;

P aram e te r ( sp VV / )

an d o th e rs

Page 6: Alexey Zavyalov Schmidt Institute of Physics of the Earth Russian Academy of Science

What parameters have been used:What parameters have been used:

K

KnKNn

Nb

n

/

)(

1lgˆ

0min

1965 1970 1975 1980 1985 1990 1995 2000Y e ar

-4

-2

0

2

4

6

8

K= 1 3 .6 M = 7 .9

ba l

*

b

b

alarm level

Nbb /ˆ

b-value (maximum likelihood estimate)

Model of b-value behavior under preparation of earthquake

Page 7: Alexey Zavyalov Schmidt Institute of Physics of the Earth Russian Academy of Science

fractures concentration parameter Kf

avr/3

1

lK f

1960 1965 1970 1975 1980 1985 1990 1995 2000Y e ar

10

20

30

40

9

8

7

6

5

4

K= 1 3 .6 M = 7 .9

K f

K f

a lT ex p

alarm level

where μ - volumetric density (concentration) of ruptures, identified on happened earthquakes,

n

iilnl

1avr /1 - average size of faults in a cell,

n - the number of events in a cell.

31

avr

RThe quantity has the meaning of the average inter-rupture distance between the centers of the ruptures.

Model of Kf behavior under

preparation of earthquake

Page 8: Alexey Zavyalov Schmidt Institute of Physics of the Earth Russian Academy of Science

number of earthquakes N

max

min

)(1 K

KiKN

TN

• seismic activation;

• seismic quiescence

1965 1970 1975 1980 1985 1990 1995 2000Y e ar

-3

-2

-1

0

1

2

K= 1 3 .6

M = 7 .9

n

n q*

n a*

n aa l

n qa l

Nn

Model of parameter N behavior under preparation of earthquake

Page 9: Alexey Zavyalov Schmidt Institute of Physics of the Earth Russian Academy of Science

released seismic energy of weak events

• seismic activation;

• seismic quiescence

)(

1

323/2 1 oKKN

iiET

E

1965 1970 1975 1980 1985 1990 1995 2000Y e a r

-2

-1

0

1

2

3

K= 13 .6 M = 7 .9

e

ea*

eq*

eaa l

eqa l

Model of parameter behavior under preparation of earthquake

3/2E

Page 10: Alexey Zavyalov Schmidt Institute of Physics of the Earth Russian Academy of Science

Efficiency of the used precursorsEfficiency of the used precursors((in timein time))

obstotal

exp.totalpr

/

/

TN

TNtJ

0123456789

10

Cauca

sus

Kyrgi

zsta

n

Turkm

eniy

a

S.Cal

iforn

ia

NE Chin

a

SW C

hina

Kamch

atka

Greec

e

W.T

urkey

Kuril

Avera

ge (1

0 re

g.)

Ind

icat

or

effe

ctiv

enes

s in

tim

e

Ksf

b-value

Nq

Na

Eq

Ea

obstotal

exp.totalpr

/

/

TN

TNtJ

Page 11: Alexey Zavyalov Schmidt Institute of Physics of the Earth Russian Academy of Science

Efficiency of the used precursorsEfficiency of the used precursors((on squareon square))

obstotal

exp.totalpr

/

/

TN

TNtJ

0123456789

10

Cauca

sus

Kyrgi

zsta

n

Turkm

eniy

a

S.Cal

iforn

ia

NE Chin

a

SW C

hina

Kamch

atka

Greec

e

W.T

urkey

Kuril

Avera

ge (1

0 re

g.)

Ind

icat

or

effe

ctiv

enes

s in

sq

uar

e

Ksf

b-value

Nq

Na

Eq

Ea

obstotal

exp.totalpr

/

/

SN

SNtJ

Page 12: Alexey Zavyalov Schmidt Institute of Physics of the Earth Russian Academy of Science

Bayesian approachBayesian approach for MEE for MEE calculationscalculations

)|()()|()(

)|()( = )|(

21

211

1

11

1

1

DKPDPDKPDP

DKPDPKDP

n

ii

n

ii

n

ii

where P(Ki|D1) – conditional probability of strong EQ occurrence use a prognostic indicator Ki; P(Ki|D2) – conditional probability of false alarm; P(D1) is unconditional probability of strong EQ occurrence in the spatial cell under consideration; P(D2)=1-P(D1) is unconditional probability of absence of a strong EQ in the spatial cell under consideration; Ki is the presence of anomaly of the i-th prognostic indicator in the spatial cell.

Conditional probability of strong EQs occurrence in elementary spatial cell is calculating as

Page 13: Alexey Zavyalov Schmidt Institute of Physics of the Earth Russian Academy of Science

MMap of ap of EExpected xpected EEarthquakes for arthquakes for CaucasusCaucasus

for the period for the period 1986.1986.0101.01 - 1990.12.31.01 - 1990.12.31((compiled in May, 1988 by G.Sobolev, L.Slavina, A.Zavyalov, T.Chelidze)compiled in May, 1988 by G.Sobolev, L.Slavina, A.Zavyalov, T.Chelidze)

1

4

2

5

3

6

%20)|( 1 KDP

%10)|( 1 KDP

%40)|( 1 KDP

1 27.01.1986, K=12.7;2 13.05.1986, K=13.8 (Paravan EQs);3 3.05.1988, K=12.6;4 7.12.1988, M=6.8 (Spitak EQs);5 3.08.1989, M=5.1;6 24.08.1989, K=13.0.

Page 14: Alexey Zavyalov Schmidt Institute of Physics of the Earth Russian Academy of Science

Map of Expected Earthquakes Map of Expected Earthquakes for Kurilfor Kuril

for the period for the period 1994.1994.1010.01 - .01 - 19919988.0.033.3.311

-1600 -1500 -1400 -1300 -1200 -1100 -1000 -900 -800 -700 -600 -500 -400

-200

-100

0

100

200

300

41

42

145

146

147

148

149

150

151

152

153

44

45

46

47

48

49

50

Itu ru p is .

K u n a sh ir is .

U ru p is .

P a ra m u sh ir is .S im u sh ir is .

50

70

90

Page 15: Alexey Zavyalov Schmidt Institute of Physics of the Earth Russian Academy of Science

Map of Expected Earthquakes Map of Expected Earthquakes for Kurilfor Kuril

for the period for the period 1994.1994.1010.01 - .01 - 19919988.0.033.3.311

-1600 -1500 -1400 -1300 -1200 -1100 -1000 -900 -800 -700 -600 -500 -400

-200

-100

0

100

200

300

41

42

145

146

147

148

149

150

151

152

153

44

45

46

47

48

49

50

1994.10

1995.04

1995.04

1995.11 1996.01

Itu ru p is .

K u n a sh ir is .

U ru p is .

P aram u sh ir is .S im u sh ir is .

50

70

90

Page 16: Alexey Zavyalov Schmidt Institute of Physics of the Earth Russian Academy of Science

Map of Expected Earthquakes Map of Expected Earthquakes for Kurilfor Kuril

for the period for the period 20062006..0707.01 - .01 - 20092009..1212.3.311

((compiled in August, 200compiled in August, 20077))

-1600 -1500 -1400 -1300 -1200 -1100 -1000 -900 -800 -700 -600 -500 -400

-200

-100

0

100

200

300

41

42

145

146

147

148

149

150

151

152

153

44

45

46

47

48

49

50

50

70

90

Page 17: Alexey Zavyalov Schmidt Institute of Physics of the Earth Russian Academy of Science

Map of Expected Earthquakes Map of Expected Earthquakes for Kurilfor Kuril

for the period for the period 20062006..0707.01 - .01 - 20092009..1212.3.311

((compiled in August, 200compiled in August, 20077))

-1600 -1500 -1400 -1300 -1200 -1100 -1000 -900 -800 -700 -600 -500 -400

-200

-100

0

100

200

300

41

42

145

146

147

148

149

150

151

152

153

44

45

46

47

48

49

50

2006.09

2006.11

2006.12

2007.01

50

70

90

Page 18: Alexey Zavyalov Schmidt Institute of Physics of the Earth Russian Academy of Science

obstotal

expprMEE /

~/

SN

SNJ

0123456789

Eff

icie

ncy

ofM

EE

alg

orith

m, J

mee

P(D1|K)=70% P(D1|K)=90%

EEfficiency of fficiency of MEEMEE algorithm on algorithm on regionsregions

0102030405060708090

100

Cauca

sus

Kyrgizs

tan

Turkm

eniy

a

S.Cal

iforn

ia

NE Chin

a

SW C

hina

Kamch

atka

Greec

e

W.T

urke

yKuril

Avera

ge (10

reg.

)

Nu

mb

er

of

str

on

ge

art

hq

ua

ke

s i

n %

P(D1|K)=70% P(D1|K)=90%

Number of strong EQs

0

10

20

30

40

50

60

70

Caucasu

s

Kyrgizs

tan

Turkm

eniy

a

S.Cal

iforn

ia

NE Chin

a

SW C

hina

Kamch

atka

Greece

W.T

urkey

Kuril

Averag

e (1

0 reg

.)

Ave

rag

e ar

ea o

fal

arm

zo

nes

in

%

P(D1|K)=70% P(D1|K)=90%

Average area of expectation

Page 19: Alexey Zavyalov Schmidt Institute of Physics of the Earth Russian Academy of Science

Map of Expected Earthquakes for Map of Expected Earthquakes for Kronotskoe EQ (Kamchatka, Dec. 5, Kronotskoe EQ (Kamchatka, Dec. 5,

1997) preparation zone1997) preparation zone ((Prognostic period begins from Jan. 1, 1997)Prognostic period begins from Jan. 1, 1997)

-100 0 100 200 300 400-200

-100

0

100

BKI

KR Y KBG

1997.12

2001.08

50

70

90

-100 0 100 200 300 400-200

-100

0

100

BKI

KR Y KBG

1997.12

2001.08

50

70

90

H = 0 - 50 km H = 25 - 75 km

Page 20: Alexey Zavyalov Schmidt Institute of Physics of the Earth Russian Academy of Science

MEE algorithm in real-time MEE algorithm in real-time predictionprediction

A T H

T H E

35

36

37

38

39

40

19 20 21 22 23 24 25 26 27 28 29

-3 0 0 -2 0 0 -1 0 0 0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0

-3 0 0

-2 0 0

-1 0 0

0

1 0 0

2 0 0

3 0 0

J u l.2 0 ,9 6

N o v .1 8 ,9 7

M a y .1 6 ,9 7

O ct.1 3 ,9 7

A p r .2 9 ,9 8

7 0

9 0

The Greece MEEfor the period 1996–2002

(compiled in May, 1997)

Page 21: Alexey Zavyalov Schmidt Institute of Physics of the Earth Russian Academy of Science

Lacks of MEE algorithmLacks of MEE algorithm

One of essential lacks of MEE algorithm will be, that it does not give the answer to a question, in which area of the increased probability there will be a next strong earthquake.

Page 22: Alexey Zavyalov Schmidt Institute of Physics of the Earth Russian Academy of Science

Prospects of the further Prospects of the further development of MEE algorithmdevelopment of MEE algorithm

localization of the seismic processtraveltime ratio of P and S waves (parameter ) parameter Ksf including the fractal correctionearthquake clustering parameter RTL parameter

MEE algorithm is open for inclusion in it new physically and statistically proved predictors satisfying requirements described above. In such approach the author sees one of ways of development and perfection of a technique.

Page 23: Alexey Zavyalov Schmidt Institute of Physics of the Earth Russian Academy of Science

C o n c l u s i o nC o n c l u s i o nThe analysis of all set of received Maps of Expected EQs for the studied seismoactive regions has shown, that the efficiency of MEE algorithm at the retrospective forecast of strong earthquakes J=3-4. Up to 70% of strong earthquakes occur in zones of the increased conditional probability. In addition the area of these zones does not exceed 30% from the total area of supervision.

Results of long-term testing allow to recommend developed MEE algorithm for strengthening of supervisions in the allocated zones with high (more than 70%) level of conditional probability over precursors of another geophysical nature having more short-term character in comparison with used, and for acceptance necessary preventive measures on reduction of probable economic and social damage from the future strong EQ.

It is possible to improve prognostic abilities of MEE algorithm by insertion of additional precursors.