Earthquake database for seismic hazard analysis

��

��

��

��

��

��

� ��

��

��

��

��

Earthquake Earthquake database database for seismic for seismic hazard analysishazard analysis

MultiscaleMultiscale analysis of earthquake catalogsanalysis of earthquake catalogs

A. A. PeresanPeresan,, G.F. G.F. Panza Panza V. V. KossobokovKossobokov, I. , I. RotwainRotwain, , A. GoA. Gorshkovrshkov

IAEA – ICTP

Workshop on the Conduct of Seismic Hazard Analyses for Critical FacilitiesTrieste - Italy, 15 - 19 May 2006

OutlineOutline

Heterogeneity of earthquake catalogsHeterogeneity of earthquake catalogsUncertainties and catalog errorsUncertainties and catalog errorsMultiscaleMultiscale approach to earthquake catalogs analysisapproach to earthquake catalogs analysis

Examples of catalogs analysisExamples of catalogs analysis-- Local scale: Mt. VesuviusLocal scale: Mt. Vesuvius-- Intermediate scale: Central Italy (CN region)Intermediate scale: Central Italy (CN region)-- Large scale / largest events: Iberian peninsulaLarge scale / largest events: Iberian peninsula

Random and longRandom and long--lasting systematic errors in reported lasting systematic errors in reported magnitudesmagnitudesIntegrating information from different data setsIntegrating information from different data setsPattern recognition of earthquake prone areasPattern recognition of earthquake prone areas

What is an earthquake catalog?What is an earthquake catalog?

An earthquake catalog is a collection of information An earthquake catalog is a collection of information about a set of seismic events, basically including:about a set of seismic events, basically including:

-- Origin timeOrigin time-- LocationLocation-- Size of the earthquakesSize of the earthquakes

Additional information can be provided, ranging from Additional information can be provided, ranging from related damage to seismic source parameters.related damage to seismic source parameters.

A catalog may include several magnitude estimations, A catalog may include several magnitude estimations, generally with a precision of one digit, even if values generally with a precision of one digit, even if values provided by different agencies may differ more than provided by different agencies may differ more than one unit.one unit.

What is an earthquake catalog?What is an earthquake catalog?

Catalogs are compiled for different purposes and by Catalogs are compiled for different purposes and by different agencies. Therefore they differ in:different agencies. Therefore they differ in:

-- geographical coveragegeographical coverage-- time span time span -- level of detectionlevel of detection-- criteria of compilationcriteria of compilation-- type and quality of earthquake datatype and quality of earthquake data

Consequence: no unique catalog for a given territoryConsequence: no unique catalog for a given territory……but usually an heterogeneous set of catalogs (historical, but usually an heterogeneous set of catalogs (historical, instrumental, local, global, etc.), not always comparable, instrumental, local, global, etc.), not always comparable, which may require different tools of analysis.which may require different tools of analysis.

A positive step forward: compilation of global catalogs A positive step forward: compilation of global catalogs (e.g. USGS(e.g. USGS--NEIC and ISC)NEIC and ISC)

Global Number of Earthquakes vs. TimeGlobal Number of Earthquakes vs. Time

0.1

1

10

100

1000

10000

1900

1910

1920

1930

1940

1950

1960

1970

1980

1990

4.04.55.05.56.06.57.07.58.08.5

The USGS/NEIC Global Hypocenter Data BaseThe USGS/NEIC Global Hypocenter Data Base

The ISC Global Data BaseThe ISC Global Data Base

Bulletin of the International Seismological Centre

Regional catalog of earthquakes

Felt and Damaging earthquakes

Measuring the size of an earthquakeMeasuring the size of an earthquake

The information about the size of historical earthquakes The information about the size of historical earthquakes is generally provided in terms of is generally provided in terms of earthquake intensityearthquake intensity, , i.e. a quantitative estimation based on the observed i.e. a quantitative estimation based on the observed damage.damage.

Intensity providesIntensity provides a qualitative a qualitative descriptiondescription of the of the earthquake sizeearthquake size, , basedbased on the on the observationobservation of the of the related related damagedamage. . HenceHence, , for a given earthquake, the intensity for a given earthquake, the intensity I I can be different in different placescan be different in different places. .

Intensity valuesIntensity values are discrete; are discrete; undue accuracyundue accuracy in in related related computationscomputations can can be misleadingbe misleading..

Intensity scalesIntensity scales

The existence of many The existence of many different scales is a different scales is a demonstration of the demonstration of the complexity of the problem complexity of the problem of describing earthquake of describing earthquake effects. The multiplicity of effects. The multiplicity of scales generates some scales generates some problems in practical problems in practical applications, that must applications, that must therefore rely upon very therefore rely upon very conservative assumptions.conservative assumptions.

Comparison of seismic intensity scales:MM – Modified MercalliRF – Rossi-ForelJMA – Japanese Meteorological AgencyMCS – Mercalli-Cancani-SiebergMSK – Medvedev-Sponheuer-Karnik

Magnitude scalesMagnitude scales

Richter's original magnitude scale (Richter's original magnitude scale (MMLL) held only for ) held only for California earthquakes occurring within 600 km of a California earthquakes occurring within 600 km of a particular type of seismograph (i.e., the particular type of seismograph (i.e., the WoodsWoods--AndersonAnderson torsion instrument). Then it was extended torsion instrument). Then it was extended to observations of earthquakes of any distance and to observations of earthquakes of any distance and of focal depths ranging between 0 and 700 km.of focal depths ranging between 0 and 700 km.Because earthquakes excite both body waves, which Because earthquakes excite both body waves, which travel into and through the Earth, and surface travel into and through the Earth, and surface waves, which are constrained to follow the Earth's waves, which are constrained to follow the Earth's uppermost layers, two other magnitude scales uppermost layers, two other magnitude scales evolved evolved -- the the mmbb and and MMSS

Magnitude scalesMagnitude scales

Quite recently the magnitude scale Quite recently the magnitude scale MMWW has been has been introduced, which is computed from seismic momentintroduced, which is computed from seismic moment

Several other magnitude estimatesSeveral other magnitude estimates are are possiblepossible, , based based on on different properties different properties of the of the recorded seismic signalrecorded seismic signal, , such as such as the the duration magnitude duration magnitude MMdd..

Making useMaking use of of specific empiricalspecific empirical relations, relations, it is also it is also possible to possible to derive a derive a magnitude from intensitymagnitude from intensity MMII. .

Uncertainties and errorsUncertainties and errors

Uncertainty in magnitude determinationsUncertainty in magnitude determinationsEpicenter distance vs. Station magnitude for Epicenter distance vs. Station magnitude for the 108 determinations for the 08 September the 108 determinations for the 08 September 2002 earthquake NEAR NORTH COAST OF 2002 earthquake NEAR NORTH COAST OF NEW GUINEANEW GUINEA

10.0

10.5

11.0

11.5

12.0

12.5

13.0

42.5 43.0 43.5 44.0 44.5 45.0 45.5

Uncertainty in preliminary Uncertainty in preliminary epicentralepicentraldeterminationsdeterminationsFast determinations of the epicenter for the Fast determinations of the epicenter for the 14 September 2003 earthquake in Northern 14 September 2003 earthquake in Northern Italy by different seismological agencies to Italy by different seismological agencies to EuropeanEuropean--Mediterranean Seismological Mediterranean Seismological Centre (EMSC)Centre (EMSC)

The distribution of the difference between average magnitudes inThe distribution of the difference between average magnitudes inepicenter and antipodal hemispheresepicenter and antipodal hemispheres

MCHEDR 1990MCHEDR 1990--2000, all events that have three or more station magnitudes in e2000, all events that have three or more station magnitudes in each hemisphereach hemisphere

MS (4560 differences Average = -0.147, s = 0.198)

mb (8175 differences Average = 0.074, s = 0.274)

Herak, M. and Herak. D. (1993). Bull. Seism. Soc. Am., 83, 6, 1881-1892.

What can we learn from a catalog of earthquakes?What can we learn from a catalog of earthquakes?

There are two extreme opinions on the subject:There are two extreme opinions on the subject:

PessimisticPessimistic: : “…“… in the case of seismic data, most in the case of seismic data, most of the observed variations are, in fact, related to of the observed variations are, in fact, related to changes in the system for detecting and changes in the system for detecting and reporting earthquakes and not to actual changes reporting earthquakes and not to actual changes in the Earth.in the Earth.””

OptimisticOptimistic: Among existing data seismic catalogs : Among existing data seismic catalogs remain the most reliable record on distribution remain the most reliable record on distribution of earthquakes in space and time.of earthquakes in space and time.

What can we learn from a catalog of earthquakes?What can we learn from a catalog of earthquakes?

All catalogs have errors, which may render invalid All catalogs have errors, which may render invalid conclusions derived in a study based on a catalog of conclusions derived in a study based on a catalog of earthquakes. earthquakes.

Ways to handle with catalog errors:Ways to handle with catalog errors:

Postpone the analysis until the data are revised;Postpone the analysis until the data are revised;Perform a comparative analysis among different catalogs, Perform a comparative analysis among different catalogs, whenever available;whenever available;Use robust methods of analysis, within the limits Use robust methods of analysis, within the limits

of their applicability;of their applicability;Test the robustness of the obtained results against Test the robustness of the obtained results against possible errors in the data set.possible errors in the data set.

The GutenbergThe Gutenberg--Richter lawRichter law

Averaged over a large territory Averaged over a large territory and time the number of and time the number of earthquakes equal or above earthquakes equal or above certain magnitude, N(M) scales certain magnitude, N(M) scales asas

loglog1010N(M) = aN(M) = a--bMbM

This general law of similarity This general law of similarity establishes the scaling of establishes the scaling of earthquake sizes in a given space earthquake sizes in a given space time volume but gives no time volume but gives no explanation to the question how explanation to the question how the number, N, changes when the number, N, changes when you zoom the analysis to a you zoom the analysis to a smaller size part of this volume. smaller size part of this volume.

0.1

1

10

100

1000

10000

4.0 5.0 6.0 7.0 8.0 9.0

The analysis of global The analysis of global seismictyseismicty shows that a single Gutenbergshows that a single Gutenberg--Richter (GR) law is not universally valid and that a Richter (GR) law is not universally valid and that a multiscalemultiscaleseismicityseismicity modelmodel ((MolchanMolchan, , KronrodKronrod & & PanzaPanza, BSSA, 1997, BSSA, 1997)) can reconcile can reconcile two apparently conflicting two apparently conflicting conceptsconcepts the Characteristic the Characteristic Earthquake (Earthquake (CECE) the Self) the Self--Organized Criticality (Organized Criticality (SOCSOC))..

The The multiscalemultiscale seismicityseismicity model, implies that only the set of model, implies that only the set of earthquakes with dimensions that are small with respect to the earthquakes with dimensions that are small with respect to the dimensions of the analysed region can be described adequately dimensions of the analysed region can be described adequately by the Gutenbergby the Gutenberg--Richter lawRichter law..

This This conditonconditon, fully satisfied in the study of global , fully satisfied in the study of global seismicityseismicitymade by Gutenberg and Richter, has been violated in many made by Gutenberg and Richter, has been violated in many subsequent investigations. subsequent investigations.

WhichWhich are the limits of are the limits of validityvalidity of the powerof the power--law law scaling expressed byscaling expressed by the the GutenbergGutenberg--RichterRichter lawlaw? ?

The counts in sets of cascading The counts in sets of cascading squares, squares, ““telescopestelescopes””, estimate the , estimate the natural scaling of the spatial natural scaling of the spatial distribution of earthquake epicenters distribution of earthquake epicenters and provide evidence for rewriting the and provide evidence for rewriting the GutenbergGutenberg--Richter recurrence law the Richter recurrence law the generalisedgeneralised form: form:

GeneralisedGeneralisedGutenbergGutenberg--Richter LawRichter Law

loglog1010N = A + BN = A + B··(5 (5 -- M) + CM) + C··loglog1010LL

where N = N(M, L) is the expected where N = N(M, L) is the expected annual number of earthquakes with annual number of earthquakes with magnitude M in an area of linear magnitude M in an area of linear dimension L.dimension L.

((KossobokovKossobokov and and MazhkenovMazhkenov, 1988, 1988))

The first resultsThe first results

The method was tested successfully on artificial catalogs and apThe method was tested successfully on artificial catalogs and applied in a plied in a dozen of selected seismic regions from the hemispheres of the Eadozen of selected seismic regions from the hemispheres of the Earth to a rth to a certain intersection of faults.certain intersection of faults.

((KossobokovKossobokov and and MazhkenovMazhkenov, 1988, 1988))

MultiscaleMultiscale seismicityseismicity modelmodel

UCI2001 catalog1900-2002

1.6 2 2.4 2.8 3.2

Log(L)

0.1

1

10

100

1000

10000

100000

Num

ber o

f eve

nts

Magnitude33.544.555.566.5

0 1 2 3 4 5 6 7 8

Magnitude

0.1

1

10

100

1000

10000

100000

Num

ber o

f eve

nts

AreaLL/2L/4L/8L/16

Seismic Seismic CatalogsCatalogs AAnalysisnalysisScales of investigationsScales of investigations

area(R~25-50 km)

Long-range scaleMiddle-range scale

Narrow scale

• Small magnitude events • Temporal evolution of seismicity over short time periods• Local seismicity patterns

• Largest earthquakes• Temporal long term evolution of seismicity• Seismic hazard

• Strong events•Temporal evolution of seismicity over months/years • Regional seimicitypatterns

Local scale good quality instrumental data• Example: Catalog of volcanic earthquakes at Mt. Vesuvius

area(R~ hundreds km)

area(R~ thousands km)

Large scale and long term data, mainly intensity informationExample: Spanish catalog over the Iberian peninsula

National scale instrumental data, sporadically integrated with intensity information• Example: Italian catalog within CN regions

Comparing low magnitude local cComparing low magnitude local catalogsatalogs::seismicityseismicity at Mt. at Mt. Vesuvius Vesuvius

Data used:Data used: catalog of volcanic earthquakes catalog of volcanic earthquakes recorded at the station OVO (recorded at the station OVO (OsservatorioOsservatorioVesuvianoVesuviano –– INGV, Naples) INGV, Naples)

Time periodTime period: 1972: 1972--20042004

The OVO The OVO earthquake catalogearthquake catalog

OVOOVO

Time sequence of the events Time sequence of the events M(t) and yearly number of M(t) and yearly number of earthquakes with Mearthquakes with M≥≥MMCC=1.8 =1.8 reported in the OVO reported in the OVO catalogcatalog

1975 1980 1985 1990 1995 2000 2005Year

2.0

2.5

3.0

3.5

4.0

Mag

nitu

de

0

50

100

150

YearlyN

umber of events

Magnitude grouping:Magnitude grouping: The analysis The analysis evidences whether there are evidences whether there are dominating values of dominating values of magnitudes. It permits to magnitudes. It permits to choose the appropriate choose the appropriate intervals of magnitude intervals of magnitude grouping grouping ∆∆MM to be considered to be considered for the frequencyfor the frequency--magnitude magnitude distribution.distribution.

Catalog completeness: Catalog completeness: The The completeness of the catalog is completeness of the catalog is determined from the determined from the frequencyfrequency--magnitude magnitude distribution distribution λλ(M), where (M), where λ λ is is the number of earthquakes the number of earthquakes within each magnitude within each magnitude grouping interval grouping interval ∆∆M, M, normalized to the spacenormalized to the space--timetime--magnitude volume unit magnitude volume unit V=[1000 km2 x 1 year x 1 M]. V=[1000 km2 x 1 year x 1 M]. MMcompletenesscompleteness=M=MCC=1.8=1.8

The OVO The OVO earthquake catalogearthquake catalog

TheThe time variations of the btime variations of the b--value in the Gutenbergvalue in the Gutenberg--Richter Richter (GR) law, (GR) law, are are analysed analysed and and show that it decreases show that it decreases progressively from 1.8, before progressively from 1.8, before 1986, to about 1.0 1986, to about 1.0 in in 1996. 1996.

N=100 events Mmin=1.8

1972 1976 1980 1984 1988 1992 1996 2000 2004Year

0.75

1

1.25

1.5

1.75

2

b-va

lue

(Maximum likelihood estimation by Wiemer & Zuniga, ZMAP software)

Time Time changes changes in in seismic activityseismic activity: : analysisanalysis of the OVO of the OVO catalocatalogg

The seismic energy release is studied considering the quantity The seismic energy release is studied considering the quantity E*E*,, energy energy normalised to normalised to the minimum the minimum magnitude eventmagnitude event,, computed from magnitude computed from magnitude according to the formula:according to the formula:

d d = const= const)( min10* MMdE −=

Seismic energy releaseSeismic energy release

Yearly normalised energy release E*(t)

(time windows of one year shifted by one month)

0 200 400 600 800 1000E*

0

4

8

12

16

20

N %

0

20

40

60

80

100

Ncu

m %

L

Formal definitions of the periods of quiescence, q and of the periods of activity, a.

ta1 a2 a3

q1 q2 q3

0

200

400

600

800

1000

1972 1976 1980 1984 1988 1992 1996 2000 2004

L (30%)

a4

q4

3.3

3.0

3.03.13.3

3.2

3.23.1

3.3

3.0

3.0

Ey*=Σyear101.5(M-Mmin)

E y*(

t)

3.63.1

Periods of high seismic activity are characterised by energy rates with an increasing trend in time, while the time intervals between subsequent periods of intense seismicity seems to decrease. An high number of earthquakes is not necessarily associated with high magnitudes.

Estimation of the parameter b of the GR law for the formally identifiedperiods of activity and quiescence (Molchan et al., 1997).

Time Time changes changes of the of the bb--valuevalue

Comparison of the b-value for the formally identified periods of activity and quiescence, as well as for some groups of them (Molchan et al., 1997).

Time Time changes changes of the of the bb--valuevalueIntervals compared π % Conclusion of the comparison

a1, a2, a3, a4 >99.9 b-values are different

a1, a2 92.6% the difference in b-values is not significant

a3, a4 61.0% the difference in b-values is not significant

q1, q2, q3, q4, q5 >99.9 b-values are different

q1, q2 55.0% the difference in b-values is not significant

q3, q4, q5 58.0% the difference in b-values is not significant

Intervals

compared

N b bmin bmax Conclusion of the

comparison

a1+a2

a3+a4

497

190

1.37

0.96

1.26

0.81

1.49

1.12

b-values are different

q1+q2

q3+q4+q5

625

250

1.69

1.21

1.55

1.05

1.84

1.37


a1+a2+q1+q2

a3+a4+q3+q4+q5

1122

440

1.51

1.09

1.42

0.98

1.61

1.20


Composite intervals

Individual intervals

The The time variations of the btime variations of the b--value and seismic energy release, observed for value and seismic energy release, observed for the OVO the OVO catalogcatalog, are checked , are checked performing performing a a similar analysissimilar analysis withwith a a different different catalogcatalog of of Vesuvian earthquakesVesuvian earthquakes, , as compiled fromas compiled from the the recordsrecords at the BKE at the BKE station station duringduring the the periodperiod 19921992--2003.2003.

Time Time changes changes in in seismic activityseismic activity: : comparisoncomparison withwith the BKE the BKE catalogcatalog

t0

200

400

600

1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004

BKEOVO

E m*(

t)

3.1

Em*=Σmonth101.5(M-Mmin)

3.0(3.0)

3.53.1

(3.6)(3.1)

3.4(3.3)

3.4

3.1(3.1)

(3.2)

(2.7)

Magnitude grouping:Magnitude grouping: The analysis evidences The analysis evidences the presence of dominating values of the presence of dominating values of magnitudes. It permits to choose the magnitudes. It permits to choose the intervals of magnitude grouping as intervals of magnitude grouping as ∆∆M=0.3.M=0.3.

The The BKEBKE earthquake catalogearthquake catalog

% %

% %Mag Mag

MagMag

N=3459 N=1361

N=32N=243

Catalog completeness: Catalog completeness: The The completeness of the catalog is completeness of the catalog is determined from the distribution determined from the distribution λλ(M), is estimated as:(M), is estimated as:MMcompletenesscompleteness=M=MCC=1.=1.00

λ(M)

The The analysis performed analysis performed using the using the catalogcatalogcompiled for the BKE compiled for the BKE stations confirms the stations confirms the bb--value decrement value decrement and and the the identification of the identification of the periods of quiescence periods of quiescence and activity, since and activity, since 1994.1994.

t0

200

400

600

800

1000

1200

1400

1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004

BKEOVO

E y*(

t)

Ey*=Σyear101.5(M-Mmin)

L(30%)3.1

3.0(3.0)

(2.7)

3.53.1

(3.6)(3.1)

3.4

3.1(3.1)

(3.2)

3.4(3.3)

a1 a2 a3

Time Time changes changes in in seismic activityseismic activity: : comparisoncomparison withwith the the BKE BKE catalogcatalog

N=100 events Mmin=1.8

1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004Year

0.75

1

1.25

1.5

1.75

2

b-va

lue

OVOBKE

The differences observed during the period 1992-1994 are well explained by a certain overestimation of BKE durations during such period of time, as shown by the comparisonof OVO and BKE durations for the common events

0 20 40 60 80 100 120 140 160DBKE

0

20

40

60

80

100

120

140

160

DO

VO

1992-1993Fit Least squares

N=357Least squares: Y = 0.65 X + 0.67

0 20 40 60 80 100 120 140 160DBKE

0

20

40

60

80

100

120

140

160

DO

VO


N=546Least squares: Y = 0.94 X - 3.75

0 20 40 60 80 100 120 140 160DBKE

0

20

40

60

80

100

120

140

160

DO

VO


N=926Least squares: Y = 0.94 X - 5.19

Time Time changes changes in in seismic activityseismic activity: : comparisoncomparison withwith the BKE the BKE catalogcatalog

Problem: identification of the common events in catalogs characterised by low magnitude and highly clustered events

Identification of Identification of common eventscommon events

∆∆TT≤≤1 1 minmin(no (no information available about information available about epicentral coordinatesepicentral coordinates))

0 0 0 1 0 1 1 3 6 3 7 15 3461

168

339

475

842

599

207

105

331417 12 7 4 8 8 7 1 3 0 0 2

-340 -300 -260 -220 -180 -140 -100 -60 -20 20 60 100 140 180 220 260 300 340

[M OVO- M BKE ]*100

0

100

200

300

400

500

600

700

800

900

1000

N. osservazioni

2983 46.89 N Validi Dev.Std.

0 0 1 0 0 0 1 4 3 5 9 2852

147

286

410

726

515

154

75

18 7 2 3 0 1 2 1 1 0 1 0 0 1

-320-300

-280-260

-240-220

-200-180

-160-140

-120-100

-80-60

-40-20

020

4060

80100

120140

160180

200220

240260

280300

320

[M OVO- M BKE]*100

0

100

200

300

400

500

600

700

800

N. osservazioni

N Validi Dev.Std.2453 38,03

All common events

Common events without clusters

The presence of several clusters The presence of several clusters of earthquakes within very of earthquakes within very narrow time windows (especially narrow time windows (especially in BKE), may lead to in BKE), may lead to uncorrectuncorrectassociation of common events.association of common events.

Apparently: large magnitude Apparently: large magnitude differencesdifferences

The The VesuvianVesuvian seismicityseismicity, instrumentally recorded since , instrumentally recorded since 1972, involves earthquakes with maximum magnitude 1972, involves earthquakes with maximum magnitude MMdd= 3.6. = 3.6.

TThe time properties of he time properties of seismicityseismicity at Mt. Vesuvius hat Mt. Vesuvius have ave been been analysedanalysed by means of CN algorithm by means of CN algorithm ((KeilisKeilis--BorokBorok and and RotwainRotwain, 1990, PEPI, 61, 57, 1990, PEPI, 61, 57--7373), ), with the aim to with the aim to characterisecharacterise the the precursory seismic activation that might precede the precursory seismic activation that might precede the moderate size earthquakesmoderate size earthquakesIn fact, though earthquakes in this area are not In fact, though earthquakes in this area are not particularly strong, due to their shallow depths and to particularly strong, due to their shallow depths and to the high urbanization of the area, they can cause the high urbanization of the area, they can cause significant concern and damages. significant concern and damages.

Analysis of precursory Analysis of precursory seismicityseismicity patterns patterns for moderate size earthquakes at Mt. Vesuviusfor moderate size earthquakes at Mt. Vesuvius

CN algorithm (Gabrielov et al., 1986; Rotwain and Novikova, 1999 ) is based on a set of empirical functions of time to allow

for a quantitative analysis of the premonitory patternswhich can be detected in the seismic flow:

Variations in the seismic activity Seismic quiescenceSpace-time clustering of events

Algorithm CNAlgorithm CN

It allows to identify the TIPs(Times of Increased Probability)

for the occurrence of a strong earthquakewithin a delimited region

The magnitude threshold M0, selecting the target earthquakes is varied within the range: 3.0 - 3.3.

By retrospective analysis, when a time scaling by a factor ϑ : 2.5 – 3 is introduced, more than 90% of the events with M ≥ M0 are predicted, with TIPs occupying about 30% of the total time considered.

CN algorithm applied at MCN algorithm applied at Mt. Vesuviust. Vesuvius

M0 ϑ n/N η τ k/K κ η+τ

3.0 3 12/13 0.08 0.31 7/19 0.37 0.39

3.1 2.5 7/7 0.0 0.31 7/14 0.50 0.31

3.2 2.5 6/6 0.0 0.32 7/13 0.53 0.32

3.3 3 4/4 0.0 0.33 7/11 0.64 0.33

n =number of predicted earthquakes with M ≥ M0; N = total number of main shocks with M ≥ M0; η = n/N statistic of failures to predict; τ = τΣ /T statistic of alarm time; k = number of false alarms;K = total number of alarms; κ = k/K statistic of false alarms

Diagnosis of Diagnosis of TIPsTIPsat Mat Mt. Vesuviust. Vesuvius

N° Date Magnitude

1 29.01.1989 3.0

2 19.03.1989 3.3

3 21.10.1989 3.0

4 4.03.1990 3.0

5 8.07.1990 3.1

6 10.09.1990 3.3

7 2.08.1995 3.2

8 16.09.1995 3.2

9 25.04.1996 3.3

10 5.11.1997 3.0

11 9.10.1999 3.6

12 22. 01.2000 3.0

13 27.09.2000 3.0

1974 1978 1982 1986 1990 1994 1998 2002 years

01234M

1974 1978 1982 1986 1990 1994 1998 2002 years

01234M

1974 1978 1982 1986 1990 1994 1998 2002 years

01234M

1974 1978 1982 1986 1990 1994 1998 2002 years

01234M

M0 =3.0

M0 =3.1

M0 =3.2

M0 =3.3

- predicted strong earthquake; - failure to predict;

- TIP preceeded a strong earthquake; - false alarm.

Fig.6List of main shocks with Md≥3.0

Diagram Diagram of of TIPsTIPs

““Seismic HistorySeismic History””ExperimentExperiment

The “Seismic History”experiment is a simulation of the forward prediction of strong earthquakes. The time period (TSP: threshold setting period) used to adjust the algorithm parameters, is progressively reduced.

1974 1978 1982 1986 1990 1994 1998 2002 years

01234M

1974 1978 1982 1986 1990 1994 1998 2002 years

01234M

1974 1978 1982 1986 1990 1994 1998 2002 years

01234M

1974 1978 1982 1986 1990 1994 1998 2002 years

01234M

1974 1978 1982 1986 1990 1994 1998 2002 years

01234M

TSP

TSP

TSP

TSP

TSP

- predicted strong earthquake; - failure to predict;

- TIP preceeded a strong earthquake; - false alarm.

Fig.7

The quality of these predictions appears quite stable with respect to variations of M0 and is similar to the quality of CN application in regions of tectonic seismic activity.

The control experiment “Seismic History” demonstratedthe stability of the obtained results and indicates that the algorithm CN can be applied to monitor the preparation of impending earthquakes with M≥3.0 at Mt. Vesuvius.

CN algorithm applied at MCN algorithm applied at Mt. Vesuviust. Vesuvius

IntermediateIntermediate--terterm middlem middle--rangerangeeearthquakearthquake ccatalogatalogs s analysisanalysis: :

CentralCentral ItalyItaly

CN is essentially based on the information contained in the earthquake catalogs.

It analyses the seismic flow using: origin time, epicentralcoordinates and magnitudes of earthquakes.

All catalogs are unavoidably affected by errors. Magnitudes are characterised by the most significant errors.

Systematic Errors Random Errors

How random errors on reported magnitudesaffect TIPs diagnosis by CN algorithm?

DICR MMMM ∆+∆+=

Randomised Magnitude

Mc : operating magnitude∆MI : measurement error ∆MD : error of discretisation

Measurement random errors

P(∆MI) = FTR(∆MI) Truncated normal probability distribution with: σ=∆Mmax /3

FTR(∆MI) = F(∆MI)/[2F(∆Mmax)-1]∆Mmax = maximum assumed error on magnitudes

maxI MM ∆≤∆

Errors of magnitude discretisation

P(∆MD) = Uniform probability distributionInterval: [-d/2; d/2)

d=discretization step=0.1

•DST:The analysis of CN functions in Central

Italy allowed us to detect a relevant long lasting change in the reported magnitudes.

The comparison of individual magnitudes, and , reported by ING and NEIC catalogs indicates, since 1987, an average underestimation of about 0.5 in the provided by ING. The hypothesis of general local

magnitude underestimation in the Italian ING bulletins is substantiated by the cross-comparison performed between ING, LDG and NEIC catalogs.The presence of the evidenced

magnitude change prevents the use of ING bulletins for CN algorithm application and makes necessary to use global data like NEIC.

•DST:The analysis of CN functions in Central

Italy allowed us to detect a relevant long lasting change in the reported magnitudes.

The comparison of individual magnitudes, and , reported by ING and NEIC catalogs indicates, since 1987, an average underestimation of about 0.5 in the provided by ING. The hypothesis of general local

magnitude underestimation in the Italian ING bulletins is substantiated by the cross-comparison performed between ING, LDG and NEIC catalogs.The presence of the evidenced

magnitude change prevents the use of ING bulletins for CN algorithm application and makes necessary to use global data like NEIC.

CCN N aalgorithmlgorithm in Central Italy:in Central Italy:rresultsesults obtained with the originalobtained with the original catalogcatalog

TIPs obtained with the original catalog with a different length of the “thresholds setting

period” (Learning)

26.9.1997

23.11.198021.8.1962

9.9.1998

5.86.0 6.5

5.76.0

1955 1960 1965 1970 1975 1980 1985 1990 1995 2000

Year

5.7

a)

Long Learning (1954-1998)

5.86.0 6.5

5.76.0

1955 1960 1965 1970 1975 1980 1985 1990 1995 2000

Year

5.7

b)

Short Learning(1954-1986)

PredictionPrediction of the of the events with events with MM≥≥5.65.6

•DST:In order to perform the magnitude comparison, the events common to the different catalogues are identified according to the following rules: a) time difference 1 minute; b) epicentral distance: 1° for the comparison with the global catalogue (Storchak, Bird and Adams, 1998). No limitation is imposed to magnitude or depth differences.

•DST:In order to perform the magnitude comparison, the events common to the different catalogues are identified according to the following rules: a) time difference 1 minute; b) epicentral distance: 1° for the comparison with the global catalogue (Storchak, Bird and Adams, 1998). No limitation is imposed to magnitude or depth differences.

Results obtained Results obtained with the with the randomisedrandomisedcatalogcatalog

∆Mmax=0.1

0 20 40 60 80 100

0

20

40

60

80

100

0 20 40 60 80 100

0

20

40

60

80

100

0 20 40 60 80 100

0

20

40

60

80

100

∆Mmax=0.2 ∆Mmax=0.3

Central ItalyLearning: 1954-1998

RandomisedOriginal

ηη=n/N =n/N : : the rate of the rate of failuresfailures--toto--predictpredict

ττ=t/T=t/T : : the rate of time the rate of time of alarmsof alarms

•N is the number of strong earthquakes occurred during the time period T covered by predictions

•The alarms cover altogether the time tand they have missed n strong events

•N is the number of strong earthquakes occurred during the time period T covered by predictions

•The alarms cover altogether the time tand they have missed n strong events

Results obtained with theResults obtained with therandomised randomised catalogcatalog::

variation of variation of target eventstarget events

•DST:In order to perform the magnitude comparison, the events common to the different catalogs are identified according to the following rules: a) time difference 1 minute; b) epicentral distance: 1° for the comparison with the global catalog (Storchak, Bird and Adams, 1998). No limitation is imposed to magnitude or depth differences.


Event date M NS, % NP, %

23.11.1980 6.5 100.0 66.7

21.8.1962 6.0 100.0 93.3

26.9.1997 6.0 100.0 40.0

9.9.1998 5.7 83.3 56.0

19.9.1979 5.5 23.3 0.0

5.5.1990 5.5 12.0 0.0

7.5.1984 5.4 0.0 0.0

List of earthquakes with

Central Italy: 1950-19995.3M ≥

Results obtained with theResults obtained with therandomisedrandomised catalogcatalog

>∆< η



Average differences in prediction errors

OCOriginal catalog

RCsRandomised catalogs

StabilityStability of of TIPs diagnosisTIPs diagnosis

Θ(t): percentage of testsfor which the time t belongs to a TIP

1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000

time

0

20

40

60

80

100

Θ(t)

%

Central Italy(Long threshold setting period)

3.0=∆ maxM

Ψ=|Θ−50|

Ψ

No

of o

bs

0

200

400

600

800

1000

1200

1400

0 5 10 15 20 25 30 35 40 45 50

Ψ: percentage of tests for which the recognition of the time t does not change with

respect to its average value 0≤Ψ≤50



The results of prediction remain stable for ∆Mmax<0.3. The quality of predictions is mainly controlled by the percentage of failures to predict, which depends on the changes in the number of strong earthquakes. The identification of TIPs is very stable during most of the time and the randomisation does not introduce spurious alarming patterns associated with the occasionally strong events.

Stability of CN predictions with respect to Stability of CN predictions with respect to random errors in magnituderandom errors in magnitude

•DST:

The analysis of CN functions in Central Italy allowed us to detect a relevant long lasting change in the reported magnitudes.

The comparison of individual magnitudes, and , reported by ING and NEIC catalogs indicates, since 1987, an average underestimation of about 0.5 in the provided by ING. The hypothesis of general local magnitude underestimation in the Italian ING bulletins is substantiated by

the cross-comparison performed between ING, LDG and NEIC catalogs.The presence of the evidenced magnitude change prevents the use of ING bulletins for CN algorithm

application and makes necessary to use global data like NEIC.

•DST:


The comparison of individual magnitudes, and , reported by ING and NEIC catalogs indicates, since 1987, an average underestimation of about 0.5 in the provided by ING. The hypothesis of general local magnitude underestimation in the Italian ING bulletins is substantiated by

the cross-comparison performed between ING, LDG and NEIC catalogs.The presence of the evidenced magnitude change prevents the use of ING bulletins for CN algorithm

application and makes necessary to use global data like NEIC.

Revision Revision of the of the rregionalization foregionalization for CN CN applicationapplicationin Central Italyin Central Italy

Keilis-Borok, , Kutznetsov, Panza, Rotwain & Costa (1990)

Costa, Stanishkova, Panza& Rotwain (1996)

Peresan, Costa & Panza (1996)

1960 1970 1980 1990 2000

0

4

-100

100.00.51.0

01020

0480240

1020

04080

02040

N2

K

G

Sigma

Smax

Zmax

N3

q

BmaxTime diagrams of the standard CN functions obtained for the Central region(Peresan et al., 1999)

Functions Sigma, Smaxand Zmax and are evaluated for 4.2≤M≤4.6, functions K, G, N3, q for M≥4.5 and functions N2 for M ≥5.0

CN algorithm and long lasting changes CN algorithm and long lasting changes in reported magnitudesin reported magnitudes

Functions of the Functions of the seismic flowseismic flow

q t M,s( )= a M( ) ⋅s− N ti M,s( )[ ]i∑( )Ma : is the the average

yearly number of events

corresponds to the sum of the grey areas for the

time intervals (t-s,t)where the number of earthquakes is less than the average

( )s,Mtq

Quiescence

•DST:2) Quiescence:The sign + indicates that the sum includes only the positive terms; therefore only the time intervals (t-s,t) where the number of earthquakes is less than the average are considered.

The dotted horizontal line indicates the average number of events expected in the time interval of length s. The grey areas correspond to the periods of quiescence.

The larger is the value assumed by the function, the more marked and prolonged is the quiescence.

•DST:2) Quiescence:The sign + indicates that the sum includes only the positive terms; therefore only the time intervals (t-s,t) where the number of earthquakes is less than the average are considered.

The dotted horizontal line indicates the average number of events expected in the time interval of length s. The grey areas correspond to the periods of quiescence.

The larger is the value assumed by the function, the more marked and prolonged is the quiescence.

Functions of the Functions of the seismic flowseismic flow

i=1,...,n u=ns

It is a measure of the cumulative

changes in time of the earthquake numberV t M,s,u( )= i N ti M,s( )− N ti −1 M,s( )∑

Variation of seismic activity

V t M,s,u( )

DST:


Is the sum of the differences between the numbers of earthquakes in two consecutive time intervals.

It corresponds to the sum of the differences between the numbers of earthquakes in two consecutive time intervals, with fixed length s. The moments belong to the time interval (t−u,t), where u is multiple of s. Usually s is equal to one year.

DST:


Is the sum of the differences between the numbers of earthquakes in two consecutive time intervals.

It corresponds to the sum of the differences between the numbers of earthquakes in two consecutive time intervals, with fixed length s. The moments belong to the time interval (t−u,t), where u is multiple of s. Usually s is equal to one year.

CN algorithm: normalization of CN algorithm: normalization of ffunctionsunctions

--

--

BmaxBmax

--

m2m2

qq

MoMo--11

M1M1

ZmaxZmax

MoMo--11MoMo--11--------MmaxMmax

m1m1m1m1m2m2m2m2m2m2m3m3MminMmin

SmaxSmaxSigmaSigmaGGKKN3N3N1N1

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

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

We use the normalization by minimalWe use the normalization by minimal magnitude cutoffmagnitude cutoff MMminmin, defined , defined by one of the two conditions:by one of the two conditions:MMminmin= M= M00 –– C, C: constantC, C: constantMMminmin such as such as N(MN(Mminmin) = A, A: constant rate of activity) = A, A: constant rate of activity

catalogscatalogscomparisoncomparison

Common events:∆T≤1min

∆Lat, ∆Lon≤1°(Storchak, Bird and Adams, 1998)

ING

LDGML

PDE(NEIC)

ING

PDE(NEIC)

ML, Md

1000

1985

June1997

PFG

Paperbulletins

ING

Digital bulletins

ML, Md

ML, Md

1980

Revised

CCI1996



Magnitude comparison: Magnitude comparison: Central RegionCentral Region

Yearly Average differencesNEIC-ING

Duration Magnitude

Local Magnitude

Md(NEIC)-Md(ING)Md ≥3.0

(1983-1997)∆Md=0.30±0.04

Yearly Average

ML(NEIC)-ML(ING)ML ≥3.0

(1980-1986)∆ML=0.13±0.05

(1988-1997)∆ML=0.64±0.04

Yearly Average

1980 1982 1984 1986 1988 1990 1992 1994 1996 1998

Year

-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

∆ML

1980 1982 1984 1986 1988 1990 1992 1994 1996 1998

Year

-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

∆MdMd≥3.0

ML≥3.0



Time diagrams of the standard CN functions obtained for the Central region(Peresan et al., 1999) :

M=ML(ING)+0.5

CN algorithm and CN algorithm and long lasting changes long lasting changes in reported magnitudesin reported magnitudes

1960 1970 1980 1990 2000

0

4

-100

100.00.51.0

01020

0480240

1020

04080

02040

N2

K

G

Sigma

Smax

Zmax

N3

q

Bmax

Central Italy

ML(ING)+0.5 since 1987

Magnitude comparison: Magnitude comparison: Italian territoryItalian territory

Events used for Md analysis

Yearly number of common events used for the comparison between ING and NEIC catalogs

Events used for ML analysis

1980 1982 1984 1986 1988 1990 1992 1994 1996

Year

1

10

100

1000

Num

of o

bs

ML: Common events >4.0 >3.0All

1980 1982 1984 1986 1988 1990 1992 1994 1996

Year

0

1

10

100

1000

Num

of o

bs

Md: Common events >4.0 >3.0All

a)

b)

MLML

Md

Md




∆∆MMLL

∆∆MMdd




∆∆MMLL=0=0AllAll




Duration MagnitudeNEIC - ING

a)

b)1980 1982 1984 1986 1988 1990 1992 1994 1996 1998

Year

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

1980 1982 1984 1986 1988 1990 1992 1994 1996 1998

Year

-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

∆ML

∆Md

LocalMagnitudeNEIC - ING

MD(NEIC)-MD(ING)MD ≥3.0

(1983-1986)∆MD=0.30±0.02

YearlyAverage

YearlyAverage

ML(NEIC)-ML(ING)ML (NEIC)≥3.0

(1980-1986)∆ML=0.08±0.05

(1988-1993)∆ML=0.30±0.04

(1995-1997)∆ML=0.77±0.06




Local Magnitude:Yearly Average differences

ML(LDG)-ML(ING)ML ≥3.0

(1980-1986)∆ML=0.18±0.08

(1988-1996)∆ML=0.44±0.04

YearlyAverage

ML(NEIC)-ML(LDG)ML ≥3.0

(1980-1986)∆ML=0.03±0.06

(1988-1996)∆ML=0.08±0.03

Yearly Average

1980 1982 1984 1986 1988 1990 1992 1994 1996 1998

Year

-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

∆ML

1980 1982 1984 1986 1988 1990 1992 1994 1996 1998

Year

-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

∆ML

LDG - ING

NEIC - LDG

ML≥3.0

ML≥3.0




The comparison of individual magnitudes, reported by ING and NEIC, indicates, since 1987, an average underestimation of about 0.5 in the Local Magnitude provided by ING.

Such general local magnitude underestimation in the Italian ING bulletins is substantiated by the cross-comparison performed between ING, LDG and NEIC catalogs.

CN algorithm and long CN algorithm and long lasting changes lasting changes

in reported magnitudesin reported magnitudes

•DST:


The comparison of individual magnitudes, and , reported by ING and NEIC catalogsindicates, since 1987, an average underestimation of about 0.5 in the provided by ING. The hypothesis of general local magnitude underestimation in the Italian ING bulletins is

substantiated by the cross-comparison performed between ING, LDG and NEIC catalogs.The presence of the evidenced magnitude change prevents the use of ING bulletins for

CN algorithm application and makes necessary to use global data like NEIC.

•DST:



substantiated by the cross-comparison performed between ING, LDG and NEIC catalogs.The presence of the evidenced magnitude change prevents the use of ING bulletins for

CN algorithm application and makes necessary to use global data like NEIC.

(Peresan, Panza & Costa, GJI 2000)

Compilation of a homogeneous Compilation of a homogeneous updated updated catalogcatalog for CN for CN

monitoring in Italymonitoring in Italy

Databases available to us:

CCI1996: PFG revised+ING bulletins (Italian catalog, available up to July 1997)Priority: ML, Md, MI

NEIC: PDE Preliminary Determinations of Epicentersfrom NEIC (global catalog).Priority: to be defined (available M: mb, MS, M1, M2)

ALPOR: Catalogo delle Alpi Orientali (local catalog for eastern Alps)Priority: ML, MI

•DST:


The comparison of individual magnitudes, and , reported by ING and NEIC catalogs indicates, since 1987, an average underestimation of about 0.5 in the provided by ING. The hypothesis of general local magnitude underestimation in

the Italian ING bulletins is substantiated by the cross-comparison performed between ING, LDG and NEIC catalogs.The presence of the evidenced magnitude change prevents the

use of ING bulletins for CN algorithm application and makes necessary to use global data like NEIC.

•DST:


The comparison of individual magnitudes, and , reported by ING and NEIC catalogs indicates, since 1987, an average underestimation of about 0.5 in the provided by ING. The hypothesis of general local magnitude underestimation in

the Italian ING bulletins is substantiated by the cross-comparison performed between ING, LDG and NEIC catalogs.The presence of the evidenced magnitude change prevents the

use of ING bulletins for CN algorithm application and makes necessary to use global data like NEIC.

Compilation of a Compilation of a homogeneous updated homogeneous updated catalogcatalog

for CN monitoring in Italyfor CN monitoring in Italy

Procedure:•Study of the completeness of PDE catalog;

•Study of the relations between different kind of magnitudes reported in the CCI1996 and PDE catalogs;

•Formulation of a rule for the choice of magnitude priority in PDE, similar to the priority used for CCI1996;

•Construction of the Updated catalog, integrating CCI1996, ALPOR and NEIC data (compatibly with the completeness of NEIC).

•DST:


The comparison of individual magnitudes, and , reported by ING and NEIC catalogs indicates, since 1987, an average underestimation of about 0.5 in the provided by ING. The hypothesis of general local magnitude underestimation in the Italian

ING bulletins is substantiated by the cross-comparison performed between ING, LDG and NEIC catalogs.The presence of the evidenced magnitude change prevents the use of

ING bulletins for CN algorithm application and makes necessary to use global data like NEIC.

•DST:



ING bulletins is substantiated by the cross-comparison performed between ING, LDG and NEIC catalogs.The presence of the evidenced magnitude change prevents the use of

ING bulletins for CN algorithm application and makes necessary to use global data like NEIC.

Operating Operating magnitmagnitudeudeselectionselection

M(CCI1996) – mb(PDE) M(CCI1996) – M1(PDE)

0 1 2 3 4 5 6 7 8mb(pde)

0

1

2

3

4

5

6

7

8

ML(

cci)

0 1 2 3 4 5 6 7 8mb(pde)

0

1

2

3

4

5

6

7

8

Mi(c

ci)

0 1 2 3 4 5 6 7 8mb(pde)

0

1

2

3

4

5

6

7

8

Md(

cci)

b=0.81a=0.60N=64sd=0.27P=6.25%

b=2.72a=-7.94N=118sd=0.29P=3.39%

b=1.19a=-1.07N=313sd=0.31P=6.07%

0 1 2 3 4 5 6 7 8M1(pde)

0

1

2

3

4

5

6

7

8

ML(

cci)

0 1 2 3 4 5 6 7 8M1(pde)

0

1

2

3

4

5

6

7

8

Mi(c

ci)

0 1 2 3 4 5 6 7 8M1(pde)

0

1

2

3

4

5

6

7

8

Md(

cci)

b=0.82a=0.51N=223sd=0.22P=5.38%

b=1.12a=-0.62N=17sd=0.33P=5.88%

b=0.91a=0.18N=452sd=0.24P=5.75

•DST:


The comparison of individual magnitudes, and , reported by ING and NEIC catalogsindicates, since 1987, an average underestimation of about 0.5 in the provided by ING. The hypothesis of general local magnitude underestimation in the Italian ING

bulletins is substantiated by the cross-comparison performed between ING, LDG and NEIC catalogs.The presence of the evidenced magnitude change prevents the use of ING bulletins

for CN algorithm application and makes necessary to use global data like NEIC.

•DST:


The comparison of individual magnitudes, and , reported by ING and NEIC catalogsindicates, since 1987, an average underestimation of about 0.5 in the provided by ING. The hypothesis of general local magnitude underestimation in the Italian ING




0 1 2 3 4 5 6 7 8MS(pde)

0

1

2

3

4

5

6

7

8

Ml(c

ci)

0 1 2 3 4 5 6 7 8MS(pde)

0

1

2

3

4

5

6

7

8

Mi(c

ci)

0 1 2 3 4 5 6 7 8MS(pde)

0

1

2

3

4

5

6

7

8

Md(

cci)

b=0.63a=1.77N=3sd=0.06P=0%

b=0.74a=1.59N=2sd=0.002P=0%

b=0.64a=1.72N=9sd=0.25P=0%

0 1 2 3 4 5 6 7 8M2(pde)

0

1

2

3

4

5

6

7

8

ML(

cci)

0 1 2 3 4 5 6 7 8M2(pde)

0

1

2

3

4

5

6

7

8

Mi(c

ci)

0 1 2 3 4 5 6 7 8M2(pde)

0

1

2

3

4

5

6

7

8

Md(

cci)

b=0.86a=0.49N=91sd=0.20P=4.40%

b=1.07a=-0.45N=31sd=0.34P=3.23%

b=0.98a=0.05N=122sd=0.26P=4.10%

M(CCI1996) – MS(PDE) M(CCI1996) – M2(PDE)

•DST:



substantiated by the cross-comparison performed between ING, LDG and NEIC catalogs.The presence of the evidenced magnitude change prevents the use of ING bulletins for CN

algorithm application and makes necessary to use global data like NEIC.

•DST:



substantiated by the cross-comparison performed between ING, LDG and NEIC catalogs.The presence of the evidenced magnitude change prevents the use of ING bulletins for CN

algorithm application and makes necessary to use global data like NEIC.

Operating Operating magnitmagnitudeude selectionselection

(PD E)M S

(PDE)1M

(PDE)2M

(P D E )m b

(CCI)Md

(CCI)M L

Poor statisticNot representative of any of CCI magnitudes

)M,M,(MM IdLCCI )M,1M,2(MMPDE S

•DST:The analysis of CN functions in Central Italy allowed us to detect a relevant

long lasting change in the reported magnitudes.


ING bulletins is substantiated by the cross-comparison performed between ING, LDG and NEIC catalogs.The presence of the evidenced magnitude change prevents the use of ING

bulletins for CN algorithm application and makes necessary to use global data like NEIC.

•DST:The analysis of CN functions in Central Italy allowed us to detect a relevant

long lasting change in the reported magnitudes.


ING bulletins is substantiated by the cross-comparison performed between ING, LDG and NEIC catalogs.The presence of the evidenced magnitude change prevents the use of ING

bulletins for CN algorithm application and makes necessary to use global data like NEIC.


M(CCI1996) – M2(PDE)

0 1 2 3 4 5 6 7 8M

0

0

1

10

100

1000

10000

Ncu

m/y

ear

CCI Time: 1900-1985PDE Time: 1986-1997

CCI(ML,Md,Mi)

PDE(M2,M1,MS) recalculated

PDE(M2,M1,MS) not recalculated

Mc

0 1 2 3 4 5 6 7 8M

0

1

10

100

1000

Lat:37-42 Lon:12-18

NEIC: 1992-1993

1994-1995

1996-1997

0 1 2 3 4 5 6 7 8 9M

1

10

100

1000

Lat:37-42 Lon:12-18

1986-1987

1988-1989

1990-1991

NEIC:

NEIC:

NEIC:

NEIC:

NEIC:

a)

b)

Southern Italy

Central Italy

•DST:


The comparison of individual magnitudes, and , reported by ING and NEIC catalogs indicates, since 1987, an average underestimation of about 0.5 in the provided by ING. The hypothesis of general local magnitude underestimation in the Italian ING



•DST:


The comparison of individual magnitudes, and , reported by ING and NEIC catalogs indicates, since 1987, an average underestimation of about 0.5 in the provided by ING. The hypothesis of general local magnitude underestimation in the Italian ING



The The Updated Updated catalog catalog of of ItalyItaly

UCI2001UCI2001

1000

1980

1986

2002

CCI ALPOR NEIC

ML(PFG)MI (PFG/CFT)

MS(NEIC)mb(NEIC)

ML(contrib)

UCI

MI (PFG/CFT/Alpor)

ML(PFG/Alpor/NEIC)

ML(INGV)Md(INGV)

NEIC(PDE)

ML(Alpor)MI(Alpor)

MS(NEIC)mb(NEIC)

M1(contrib)M2(contrib)

ML(INGV/NEIC)Md(INGV)MS(NEIC)mb(NEIC)

MS(NEIC)mb(NEIC)

MS(NEIC)mb(NEIC)

M1(contrib)M2(contrib)

The The Updated Updated catalog catalog of of ItalyItaly

1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000Year

1

10

100

1000

10000

Ncu

m

Mmax>MAll1.02.03.04.05.06.07.0

UCI catalog

CN algorithm (Keilis-Borok et al., 1990; Peresan et al., 2004)

M8S algorithm (Kossobokov et al, 2002)

Main features:Fully formalized algorithms and computer codes available for independent testing;Use of published & routine catalogs of earthquakes;Worldwide tests ongoing for more than 10 years permitted to assess the significance of the issued predictions.

Italy:Stability tests with respect to several free parameters of the algorithms (e.g. Costa et al., 1995; Peresan et al., GJI, 2000; Peresan et al., PEPI, 130, 2002);CN predictions are regularly updated every two months since January 1998. M8s predictions are regularly updated every six months since January 2002.

Real time prediction experiment started in July 2003Real time prediction experiment started in July 2003

IntermediateIntermediate--term middleterm middle--range earthquake prediction range earthquake prediction experiment in Italyexperiment in Italy

Updated to MayUpdated to May 1 20061 2006

M≥6.5 M≥5.5M≥6.0

Monitored region

Alerted region

CNCNTimesTimes of of Increased Increased Probability for Probability for the the occurrence occurrence of of events events with with M>Mo M>Mo within within the the monitored regionsmonitored regions

M8sM8s

Northern Region, Mo=5.4

6.5 5.4 5.8 6.0

1965 1975 1985 1995 2005

5.55.65.5

Central Region, Mo=5.65.86.0 6.5

5.76.0

1955 1965 1975 1985 1995 2005

5.7

Southern Region, Mo=5.65.86.0 6.55.8

1955 1965 1975 1985 1995 2005

5.7

2003.5-2004 2004-2004.5 2004.5-2005

The experiment, launched starting on July 2003, is aimed at a real-time test of M8S and CN predictions in Italy. Updated predictions are regularly posted at:“http://www.ictp.trieste.it/www_users/sand/prediction/prediction.htm”

A complete archive of predictions is made accessible to a number of scientists, with the goal to accumulate a collection of correct and wrong predictions, that will permit to validate the considered methodology.Current predictions are protected by password. Although these predictions are intermediate-term and by no means imply a "red alert", there is a legitimate concern about maintaining necessary confidentiality.

IntermediateIntermediate--term middleterm middle--range earthquake prediction range earthquake prediction experiment in Italyexperiment in Italy

Algorithm CN predicted 12 out of the 13 strong earthquakesoccurred in the monitored zones of Italy, with 32% of the considered space-time volume occupied by alarms.(updated to January 1 2006)

SpaceSpace--time volume of alarm in CN application in Italytime volume of alarm in CN application in Italy

ExperimentSpace-time volume

of alarm (%)Confidence

level (%)

Retrospective*(1954 – 1963)

41 93

Retrospective(1964 – 1997)

27 >99

Forward(1998 – 2004)

47 90

All together(1954 – 2004)

32 >99

ExperimentSpace-time volume

of alarm (%) n/N

Retrospective*(1954 – 1963)

41 3/3

Retrospective(1964 – 1997)

27 5/5

Forward(1998 – 2006)

41 4/5

All together(1954 – 2006)

32 12/13

IntermediateIntermediate--term middleterm middle--range earthquakerange earthquake predictionprediction

* Central and Southern regions only

Experiment M6.5+ M6.0+ M5.5+

Space-time volume, %

n/N Space-time volume, %

n/N Space-time volume, %

n/N

Retrospective(1972-2001)

36 2/2 40 1/2 39 9/14

Forward(2002-2006)

49 0/0 43 0/0 25 4/8

All together(1972-2006)

37 2/2 40 1/2 38 13/22

SpaceSpace--time volume of alarm in M8s application in Italytime volume of alarm in M8s application in Italy

Algorithm M8s predicted 62% of the events occurred in the monitored zones in Italy, i.e. 16 out of 26 events occurred within the area alerted for the corresponding magnitude range. The confidence level of M5.5+ predicitions since 1972 has been estimated to be about 97%; no estimation is yet possible for other magnitude levels.(updated to January 1 2006)

A review of the application of the algorithms CN and M8 to the Italian territory, about the input data, as well as detailed information about their performances is provided in:

“Intermediate-term middle-range earthquake predictions in Italy: a review” (2005), by A. Peresan, V. Kossobokov, L. Romashkova and G.F. Panza. Earth Science Reviews (69, 97-132, 2005).

Historical and instrumentalHistorical and instrumental ccatalogsatalogs::the the exampleexample of of Spain Spain

TThe he catalogcatalog of of seismicityseismicity for the Iberian peninsulafor the Iberian peninsula (SSIS)(SSIS)

CatalogoCatalogo sismicosismico de la Peninsula de la Peninsula IbericaIberica (880 a.C.(880 a.C.--1900)1900)Martinez Solares and Mezcua Rodriguez (2002)

Cumulative number of events vs. time and intensity


Before 1400 the completeness threshold cannot be clearly identified. The catalog is not complete for I>8 up to 1360

Time interval: 1000-1400

High percentage of events without any intensity estimation in the SSIS catalog, with some sensible reduction only since 1500


The catalog appears complete for I>8 during the period 1400-1750



The catalog appears complete for I>6 during the whole period 1750-1900and for I>5 (and eventually for I>4) since 1986.



The frequency-intensity distributions, obtained for the SSIS catalog over three different time intervals, are comparable only in the intensity range 7.0-8.5.

The cumulative distribution for the whole time interval 1400 – 1900 appears preferable for I>7.

The frequency of the events with I>9 appears much larger during the period 1750-1900 than for 1400-1750.

The catalog seems to be quite complete and homogeneous for intensities I>6 after 1750; nevertheless the time interval 1750-1900 alone is not enoughto characterise the frequency of the large events.


The overall increase in the number of earthquakes seems not due to an increased “detection” level, that should not affect the largest intensities, but would rather imply a progressive increase of the number of small events.We can see that the intensities reported in the two catalogs SSIS(EMS-98) and IGN (MSK) are not homogeneous and, if used all together, do not provide a consistent picture of seismicity.

Cumulative number of events vs. time for the SSIS catalog (1750 –1900) and for the IGN catalog (1900 – 2000). For the events reported without any intensity in the IGN

catalog, the intensity recalculated from IGN magnitude mb (e.g. Lopez-Casado et al., 2000)

is considered.

1760 1780 1800 1820 1840 1860 1880 1900 1920 1940 1960 1980 2000

Year

1

10

100

1000

10000

Ncu

m

Intensity>IAll123456789

SSIS - IGN

TThe he IGN IGN catalogcatalog of of seismicityseismicity

Distribution of the magnitude M(IGN) versus the MSK intensity reported in the IGN catalog (1900-2002)

Poor correlation among Intensity (MSK) and Magnitude (mb).

”Seismological database for seismic hazard assessment needs to beuniform and to cover a long enough time interval to allow the

occurrence of rare, large-magnitude events—generally associated with long return periods—to be estimated, notably for critical structures.”

Is the information on observed Is the information on observed seismicityseismicitysufficient to identify the sites where large sufficient to identify the sites where large

earthquakes may occur?earthquakes may occur?

Pattern Recognition Pattern Recognition of Earthquake Prone areasof Earthquake Prone areas

Pattern recognition technique is used to identify, independently from seismicityinformation, the sites where strong earthquakes are likely to occur

Assumption: strong events nucleate at the nodes, specific structures that are formed around intersections of fault zones.


The nodes are defined by the Morphostructural Zonation Method

delineates a hierarchical block structure of the studied region, based on:

topography tectonic datageological data


The Earthquake Prone AreasEarthquake Prone Areas are identified evaluating, among the others, the following topographic characteristics :

Elevation and its variations in mountain belts and watershed areas;Orientation and density of linear topographic features;Type and density of drainage pattern.

These features indicate higher intensity in the neotectonic movements and increased fragmentation of the crust at the nodes.

Topographic parametersMaximum topographic altitude, (Hmax)Minimum topographic altitude, (Hmin)Relief energy, (DH) (Hmax - Hmin)Distance between the points Hmax and Hmin, (L)Slope, (DH/L)

Geological parametersThe portion of soft (quaternary) sediments,

Gravity parametersMaximum value of Bouguer anomaly,Minimum value of Bouguer anomaly,

Difference between Bmax and Bmin,Parameters from the morphostructural map

The highest rank of lineament in a node Number of lineaments forming a node,Distance to the nearest 1st rank LineamentDistance to the nearest 2nd rank lineament Distance to the nearest node

Morphological parameters

Topographic altitudes and the area of soft sediments characterize indirectly the contrast and intensity of the present-day tectonic movements, while those describing the density of lineaments and gravity anomalies can be related to the degree of crust fragmentation and heterogeneity.

Input dataInput data forfor thethe ppatternattern rrecognitionecognitionof of earthquake earthquake prone prone areasareas

(Gorshkov et al., 2004)

Pattern Recognition Pattern Recognition of Earthquake Prone of Earthquake Prone

areasareas

This approach has been applied to This approach has been applied to many regions of the world. The many regions of the world. The predictions made in the last 3 decades predictions made in the last 3 decades have been followed by many events have been followed by many events ((84% of the total84% of the total) that occurred in ) that occurred in some of the nodes previously some of the nodes previously recognized to be the potential sites for recognized to be the potential sites for the occurrence of strong events.the occurrence of strong events.

GELFAND I., Guberman Sh., Izvekova M., KEILIS-BOROK V F. & RANTSMAN E. (1972) - Criteria of high seismicity determined by pattern recognition. Tectonophysics, 13 (1/4), 415-422.

Gvishiani A., and Soloviev A. (1980). On the concentration of the major earthquakes around the intersections of morphostructural lineaments in South America. In V.I.Keilis-Borok and A.L.Levshuin (eds), Methods and Algorithms for Interpretation of Seismological Data. Nauka, Moscow, 46-50 (Computational Seismology, Iss.13, in Russian).

GELFAND I., Guberman Sh., Izvekova M., KEILIS-BOROK V F. & RANTSMAN E. (1972) - Criteria of high seismicity determined by pattern recognition. Tectonophysics, 13 (1/4), 415-422.

Gvishiani A., and Soloviev A. (1980). On the concentration of the major earthquakes around the intersections of morphostructural lineaments in South America. In V.I.Keilis-Borok and A.L.Levshuin (eds), Methods and Algorithms for Interpretation of Seismological Data. Nauka, Moscow, 46-50 (Computational Seismology, Iss.13, in Russian).

First rank First rank morphostructuralmorphostructural unitsunits

Second rank Second rank morphostructuralmorphostructural unitsunits

MZ map and earthquakes used to select MZ map and earthquakes used to select training sets for the CORAtraining sets for the CORA--3 algorithm3 algorithm

MorphostructuralMorphostructural ZonationZonation map map and observed earthquakesand observed earthquakes

(Gorshkov et al., 2004, 2002)

Recognition of nodes where strong earthquakes may Recognition of nodes where strong earthquakes may nucleate in the Mediterranean areanucleate in the Mediterranean area

Is the information on observed Is the information on observed seismicityseismicity sufficient to sufficient to identify the sites where large earthquakes may occur?identify the sites where large earthquakes may occur?

(Gorshkov et al., 2004, 2002)

Integrated Deterministic Integrated Deterministic Seismic Hazard ScenariosSeismic Hazard Scenarios

•• Scenarios associated to CN regions (+ time)Scenarios associated to CN regions (+ time)•• Scenarios associated to earthquake prone areas Scenarios associated to earthquake prone areas •• MicrozoningMicrozoning ((including lateral heterogeneitiesincluding lateral heterogeneities))

Predicted earthquakes and magnitude

Alarms TIP: 29.2% of total timeUpdated to: 1-7-2004

Deterministic hazard scenarios associated to CN regionsDeterministic hazard scenarios associated to CN regions

Northern RegionNorthern RegionPrediction of earthquakes with M≥5.4

6.5 5.4 5.8 6.0

1965 1975 1985 1995 2005

5.5 5.6

M≥6.5 M≥6.0

Ground motion scenario

associated to a possible

earthquake with magnitude

M=6.5

Integrated deterministic seismic hazard

Scenario associated to the node determining the Scenario associated to the node determining the maximum ground motion in the city of Triestemaximum ground motion in the city of Trieste

corresponding to an earthquake with M=6.5corresponding to an earthquake with M=6.5(compatible with seismic history and (compatible with seismic history and seismotectonicsseismotectonics))

Imax computed

PGD (cm)

PGV (cm/s)

DGA (g)

ING ISG

Imax observed

2.0 – 3.5 4.0 – 7.5 0.04-0.07 VIII VII VII

Peak Ground Displacement (PGD), Peak Ground Velocity (PGV), Design Ground Acceleration (DGA) and maximum computed intensity (Imax computed), estimated using the conversion tables proposed by Panza et al. (2001). The observed intensity in the city of Rome is the same in the ING and ISG data sets. Intensity is on the MCS scale.

Based on the morphostructuralzonation, threepossible seismic sources have beenconsidered forground motionmodelling in Trieste:

A seismic source in the Bovec zone (65 km from Trieste)

A seismic source East of Gorizia (30 km from Trieste)

The closest seismic source at 17 km fromTrieste

Profil 1 - Bedrock “B” - Dist. 17 km - M=6.0Accelerations and Amplifications (RSR 2D/1D)

ConclusionsConclusions

Historical information is qualitative and even instrumental data are unavoidably affected by errors. Therefore robust methods of analysis should be used, within the limits of their applicability. Special care should be paid to verify data quality and homogeneity and the significance and stability of the obtained results should be evaluated.

Within areas with size comparable to the source dimensions of the large earthquakes, low-moderate seismicity does not appear informative about the frequency of the strong earthquakes. In such regions, large events can hardly be modeled as a random process and statistic is usually not enough to characterize their recurrence.

ConclusionsConclusions

Statistical characterization of earthquake occurrence requiresa large amount of homogeneous and complete data, over a long enough time interval. Detailed studies on individualearthquakes are essential but may be not sufficient to characterise earthquake recurrence.

The deterministic approach does not rely upon the unavoidably incomplete statistical description of large earthquakes occurrence, but rather on the appropriate description of the physical properties of earthquake sources and of the medium.

Information from data different than the seismological ones (morphological, geological, etc.) can be very useful to integrate the data base for seismic hazard analysis, provided the information is collected as much systematically and homogeneously as possible.

Documents

Earthquake database for seismic hazard analysis