RENDICONTI LINCEI

, Volume 21, Issue 2, pp 131–162

Recognition of earthquake-prone areas (M ≥ 5.0) in the Iberian Peninsula

Authors

    • International Institute of Earthquake Prediction Theory and Mathematical GeophysicsRussian Academy of Sciences
    • The Abdus Salam International Center for Theoretical Physics, SAND
  • A. A. Soloviev
    • International Institute of Earthquake Prediction Theory and Mathematical GeophysicsRussian Academy of Sciences
    • The Abdus Salam International Center for Theoretical Physics, SAND
  • M. J. Jiménez
    • Department of Volcanology and GeophysicsSpanish Council for Scientific Research, Museum of Natural History
  • M. García-Fernández
    • Department of Volcanology and GeophysicsSpanish Council for Scientific Research, Museum of Natural History
  • G. F. Panza
    • The Abdus Salam International Center for Theoretical Physics, SAND
    • Department of Earth SciencesUniversity of Trieste
Article

DOI: 10.1007/s12210-010-0075-3

Cite this article as:
Gorshkov, A.I., Soloviev, A.A., Jiménez, M.J. et al. Rend. Fis. Acc. Lincei (2010) 21: 131. doi:10.1007/s12210-010-0075-3

Abstract

Seismogenic nodes capable of earthquakes with M ≥ 5.0 or I0 ≥ VII have been identified in the Iberian Peninsula using the pattern recognition approach. Recognition objects, morphostructural nodes, have been delineated with the morphostructural zoning method. Most of the recognized seismogenic nodes (D) are scattered at the periphery of the Peninsula, while in its internal part, apart from the northern part of the Iberian Chain, there is no indication for the existence of D nodes. The performed recognition pinpoints a number of D nodes where moderate events have not been recorded to date, specifically, in the Cantabrian Mts, Portuguese basin, westernmost termination of the Betics, and in the area around Valencia. Some of the recognized D nodes are potential sources of seismic risk for nuclear and water power plants and large metropolitan areas.

Keywords

IberiaMorphostructural zoningPattern recognitionSeismogenic nodes

1 Introduction

The identification of prone areas to M ≥ 5.0 earthquakes in the Iberian Peninsula has been carried out applying the morphostructural zoning (MZ) method (Alexeevskaya et al. 1977; Rantsman 1979; Gorshkov et al. 2003) to delineate morphostructural nodes, i.e. specific structures formed at the intersections of fault zones. Pattern recognition algorithms CORA-3 and CLUSTERS (Gelfand et al. 1976; Gorshkov et al. 2003) have been used to select the nodes with significant seismogenic potential.

The fact that earthquakes nucleate at nodes was first established for the Pamirs and Tien Shan regions (Gelfand et al. 1972). The roles of intersecting faults are observed in different tectonic settings. Talwani (1988) found that large intraplate earthquakes are related to intersections and demonstrated (Talwani 1999) that intersecting faults provide a location for stress accumulation. Hudnut et al. (1989) and Girdler and McConnell (1994) observed the relationship between earthquakes and intersections for plate boundaries and rift structures, respectively. According to King (1986), fault intersections provide locations for initiation and healing of ruptures. A model proposed by Gabrielov et al. (1996) implies that block interaction along intersecting faults leads to stress and strain accumulation and secondary faulting about the intersection. This causes generation of new faults of progressively smaller size, so that a hierarchical mosaic structure, essentially, a node, is formed about the intersection.

The methodology has been previously applied in several seismic regions of the world (Cisternas et al. 1985; Gelfand et al. 1972, 1976; Gorshkov et al. 2000, 2002, 2003, 2004, 2009; Gvishiani et al. 1987, 1988). Recent earthquakes in each of the studied regions ensure the reliability of the methodology. As Gorshkov et al. (2003) demonstrated that 90% of the post-publication events with relevant magnitudes occurred at the nodes, and 84% of the post-publication events occurred at the nodes recognized as prone to large earthquakes.

The northern part of the Iberian Peninsula, the Pyrenees, has been studied with pattern recognition for the identification of nodes prone to earthquakes with M ≥ 5.0 by Gvishiani et al. (1986, 1987). In this work, the methodology is implemented to the large area of the Iberian Peninsula with heterogeneous tectonic structure and topography. The morphostructural nodes have been uniformly obtained in the different topographic and tectonic environments of the Iberian Peninsula with the MZ method, and the pattern recognition algorithms allow selecting nodes prone to earthquakes with M ≥ 5.0. We defined this magnitude cutoff due to the moderate seismicity observed in the Iberian Peninsula (e.g. Buforn et al. 2004). The number of large events reported in the earthquake catalogues (Martínez-Solares and Mezcua 2002; IGN 2007) is insufficient to formulate the problem of recognition for M ≥ 6.0 by analogy with our previous works for other Mediterranean regions (Gorshkov et al. 2002, 2004).

Several studies have been carried out to delineate seismogenic zones for seismic hazard evaluation in the Iberian Peninsula (e.g. Martin 1984; Sanz de Galdeano and Lopez Casado 1988; Jiménez et al. 1999, 2001; Ojeda et al. 2001; Peláez and López-Casado 2002). Recently, Cotilla Rodriguez and Cordoba Barba (2004) employed several different morphotectonic and morphostructural approaches, including some principles of MZ (Rantsman 1979), for the determination of the most active morphostructural units.

2 Methodology

The methodology applied to identify earthquake-prone areas in the Iberian Peninsula includes two main steps. The first step is the determination of the morphostructural nodes to be regarded as recognition patterns, using the MZ method. The second step is the classification of all mapped nodes into nodes where earthquakes with magnitude exceeding a certain threshold are possible and nodes where only earthquakes with smaller magnitude may happen, using the pattern recognition algorithms CORA-3 and CLUSTERS with the learning stage. The methodology is described in supplementary material.

The target region is divided into a system of hierarchically ordered areas characterized by homogeneous present-day topography and tectonic structure. MZ distinguishes (1) areas of different rank, called blocks; (2) their boundary zones, called morphostructural lineaments; and (3) sites where lineaments intersect, called nodes.

The use of the pattern recognition approach suggests that nodes already marked by one or more strong earthquakes might have a similar portrayal that can be used to identify nodes, which did not yet explicitly show up as earthquake prone. The goal of the recognition is to classify all the nodes delineated within a region into two classes:
  1. (1)

    class D containing the nodes where earthquakes with magnitude M ≥ M0 may occur;

     
  2. (2)

    class N containing the nodes where only earthquakes with M < M0 may occur.

     

Nodes are characterized by a set of topographical, geologic, and geophysical parameters. A vector of values of these parameters represents each node. The set of these vectors is the input for the recognition algorithms.

3 MZ of the Iberian Peninsula

The morphostructural map of the Iberian Peninsula has been compiled using topographic maps, satellite images, geological and tectonic maps. MZ is performed deliberately ignoring seismicity data (earthquake catalogues) in the study region.

The recent morphostructure of the Iberian Peninsula is mainly the result of the tectonic deformations related to the opening of the Atlantic Ocean and to the collision between the African and Eurasian plates (e.g. Sartori et al. 1994; Cloetingh et al. 2002; Vera 2004). An ongoing convergence of these plates governs the recent geodynamics in the region (Cavazza and Wezel 2003). The evolution of major topography features shown in Fig. 1 has been strongly affected by neotectonics (Cloetingh et al. 2005).
https://static-content.springer.com/image/art%3A10.1007%2Fs12210-010-0075-3/MediaObjects/12210_2010_75_Fig1_HTML.gif
Fig. 1

Major topography features of the Iberian Peninsula

The application of the MZ methodology to the Iberian Peninsula starts by defining its boundaries. We treat a first rank lineament, traced along the steep escarpment of the continental slope, as the structural boundary of the Iberian Peninsula (Fig. 2).
https://static-content.springer.com/image/art%3A10.1007%2Fs12210-010-0075-3/MediaObjects/12210_2010_75_Fig2_HTML.gif
Fig. 2

Iberian Peninsula: first rank morphostructures. Thick lines indicate the lineaments of the first rank. Continuous lines depict the longitudinal lineaments, while the discontinuous ones represent the transverse lineaments The 12 first rank blocks are: P Pyrenees, B Betics, Cm Cantabrian Mountains, Ic Iberian Chain, Cs Central System, Cr Catalan Coastal Ranges, Eb Ebro basin, Db Duero basin, Tb Tagus basin, Gb Guadalquivir basin, Mn northern Meseta, Ms southern Meseta, MF Messejana Fault. The lineaments are plotted in Figs. 2, 3, 4, 5 as lines of different width. We use zones of 10, 5, and 2 km in width for the first, second, and third rank lineaments, respectively. Such estimation of the zone’s size is in agreement with our field observations in the Caucasus (Gvishiani et al. 1988) and with the MZ performed in this work using cartographic sources

The uniformity of each morphostructural unit in MZ is determined by a certain set of morphometric features and their quantitative index.

A single mountain country is a territory of the same orogenesis and has a certain combination of large topographic forms that MZ defines as the appearance of relief. The most prominent large-scale morphostructures of Iberia, both the positive (mountain ranges) and negative ones (basins), have been assigned to the first rank units (Fig. 2). Six of them are mountain countries represented by high ranges: Pyrenees, Betics, Cantabrian Mountains, Iberian Chain, Central System, Catalan Coastal Range. These mountain countries differ in the elevation level, predominant orientation of ranges, tectonic structures and history. The Pyrenees, Cantabrian Mountains, and the Betics belong to the young Alpine orogenic belt, while the Iberian Chain and Catalonian Coastal Ranges have been formed during the Paleogene phases of the Pyrenean orogeny through inversion of long-lived Mesozoic rift systems (Vegas et al. 1990). The Central System consists of the uplifted Hercynian basement (Salas et al. 2001). The largest Ebro, Duero, Tagus, and Guadalquivir basins have also been assigned to the first rank areas, as macroblocks (Fig. 2). The Variscan Iberian Massif has been divided into two-first rank macroblocks, the northern Meseta (Mn) and the southern Meseta (Ms) because they differ in the dominant trend of the tectonic structure and in the main topographic features. These macroblocks are separated by the first rank lineament, MF in Fig. 2, that was traced along the Messejana fault (Carvalho et al. 1985; Ribeiro et al. 1990).

First rank lineaments (Fig. 2) were traced along the topographically contrasting junctions of mountain ranges and basins, except the lineament dividing Betics and the Guadalquivir basin, which was traced in accordance with the tectonic boundary, as it is shown on the tectonic map of Iberia (Julivert et al. 1974). In general, the first rank lineaments delineated with MZ correspond to the prominent regional faults (Julivert et al. 1974; Vera 2004).

The second rank morphostructural area, the megablock, is an area revealing gradual increasing or decreasing altitude of ranges in a certain direction and/or successive changes in orientation of ranges that compose the mountain country. A sharp and considerable change of at least one feature distinguishes the megablocks from each other. In plains, megablocks differ in the orientation of the tectonic structures and in the drainage pattern. The megablocks obtained are shown in Fig. 3. As an example, we explain the megablocks delineated in the Betics.
https://static-content.springer.com/image/art%3A10.1007%2Fs12210-010-0075-3/MediaObjects/12210_2010_75_Fig3_HTML.gif
Fig. 3

Iberian Peninsula: second rank morphostructures. Thick lines denote the lineaments of the first rank. Medium lines show lineaments of the second rank. Continuous lines depict the longitudinal lineaments, while dashed ones represent the transverse lineaments

The longitudinal segmentation of the Betics into megablocks is based on the traditional subdivision of the orogen into the Internal and External Zones composed by Paleozoic rocks and Mesozoic–Cenozoic sedimentary rocks, respectively (e.g. Lorengan and White 1997). These zones are separated by the extended Cádiz–Alicante fault system characterized with a complex structure (Julivert et al. 1974). The second rank transverse lineaments divide the Betics into seven megablocks (B1–B7) that differ in the mean elevation and in the orientation of ridges. Most of second rank transverse lineaments correspond to tectonic faults shown on tectonic maps (Julivert et al. 1974; Vera 2004). Megablocks B1, B3, B5, and B6 are outlined within the External Zone, while megablocks B2, B4, and B7 (Fig. 3) are delineated in the Internal Zone. Two groups of megablocks are separated by the longitudinal second rank lineament that corresponds to the Cádiz–Alicante fault, with the exception of the westernmost segment of the lineament.

Megablocks have been divided into smaller units, blocks, by third rank lineaments (Fig. 4). In mountains, the third rank lineaments control local sharp changes in the altitude of individual ridges composing mountain country; they have been traced along elongated scarps and/or rectilinear segments of river valleys that are usually fault-dominated in tectonically active environments (e.g. Korzhuev 1974; Linzer et al. 1995). In plains, the network of third rank lineaments has been defined from the analysis of the variations in the drainage pattern.
https://static-content.springer.com/image/art%3A10.1007%2Fs12210-010-0075-3/MediaObjects/12210_2010_75_Fig4_HTML.gif
Fig. 4

Morphostructural map of the Iberian Peninsula and earthquakes with M ≥ 5.0 or I0 ≥ VII. Thick lines indicate the lineaments of the first rank; medium lines show lineaments of the second rank; thin lines depict the lineaments of the third rank. Continuous lines depict the longitudinal lineaments, while the dashed ones represent the transverse lineaments. Dots denote the epicenters of earthquakes with M ≥ 5.0 or I0 ≥ VII reported by Table 1

In total, MZ outlined 405 intersections of lineaments (Fig. 5) and we consider each of them as a node. These 405 nodes form the set of recognition patterns.
https://static-content.springer.com/image/art%3A10.1007%2Fs12210-010-0075-3/MediaObjects/12210_2010_75_Fig5_HTML.gif
Fig. 5

Morphostructural map of the Iberian Peninsula and numbering of the identified nodes. Thick lines indicate the lineaments of the first rank; medium lines show lineaments of the second rank; thin lines depict the lineaments of the third rank. Continuous lines depict the longitudinal lineaments, while the dashed ones represent the transverse lineaments. Numerals denote the nodes identified with MZ

Due to heterogeneous topography and the complex tectonic structure of Iberia the nodes delineated are situated in very different morphological and tectonic environments. This leads to considerable variations in the values of the parameters describing the nodes. Because of this, we have assembled four groups of nodes (groups I–IV), each including nodes located within more or less similar morphological and tectonic environments (Table 3).

4 Nodes and earthquakes

Since the nodes have been outlined from the cartographic sources without field investigations, their natural boundaries (geometry of the node) have not been defined. Here, as in Gorshkov et al. (2004), we define the node as a circle of 25 km of radius, centered at the point of intersection of the lineaments. Using this formal node definition, each point of lineament intersection is a node but, in reality, two or three closely situated intersections may belong to the same node (Rantsman 1979). Recently, we have made an attempt to map the geometry of the nodes in the Alps-Dinarides junction zone using large-scale cartographic sources (Gorshkov et al. 2009). This approach avoids the overlapping of the circles in areas where intersections are located in a short distance to each other. Therefore, the recognizing areas of potential hazard may be reduced in size, but this procedure of node geometry mapping needs further formalization.

The largest observed magnitudes and macroseismic intensities in the Iberian Peninsula are mainly related to historical events (Martínez-Solares and Mezcua 2002; IGN 2007). In this work, the objective has been to recognize the nodes prone to produce shallow earthquakes with M ≥ 5.0. To select sample nodes for the learning step of the recognition, we use the information on the recorded shallow events with M ≥ 4.8 (allowing a conservative 0.2 units error in the magnitude calculation), or epicentral intensity I0 ≥ VII from the most recent available Spanish Earthquake Catalogue (Martínez-Solares and Mezcua 2002; IGN 2007). The selected events are summarized in Table 1 and plotted in Fig. 4.
Table 1

Earthquakes used to select sample nodes for recognition

Date

Lat N

Log

Depth

I0_EMS

mbLg

Distance to the nearest node

Number of the nearest node

01/01/1048

38.08

−0.92

 

VIII

 

21.2

b391

01/01/1169

38.00

−4.00

 

VIII–IX

 

12.4

d282

01/03/1258

38.83

−0.60

 

VIII

 

21.2

f265

01/01/1344

38.90

−8.80

 

VII–VIII

 

13.0

f304

02/03/1373

42.50

0.75

 

VIII–IX

 

31.7

a78

18/12/1396

39.08

−0.22

 

VIII–IX

 

8.2

f264

02/05/1404

42.00

2.80

 

VII

 

13.2

a95

01/01/1406

37.25

−1.87

 

VII–VIII

 

17.0

b385

13/03/1427

42.00

2.60

 

VII–VIII

 

2.5

a96

14/03/1427

42.00

2.60

 

VII–VIII

 

2.5

a96

15/03/1427

42.00

2.60

 

VII

 

2.5

a96

19/03/1427

42.00

2.60

 

VIII

 

2.5

a96

22/04/1427

42.00

2.60

 

VII–VIII

 

2.5

a96

15/05/1427

42.20

2.50

 

VIII–IX

 

16.8

a91

17/06/1427

41.80

2.80

 

VII–VIII

 

8.3

c97

02/02/1428

42.35

2.17

 

IX–X

 

15.2

a91

24/04/1431

37.13

−3.63

 

VIII–IX

 

6.2

b362

24/05/1448

41.60

2.22

 

VII–VIII

 

10.1

d104

16/09/1450

42.60

2.80

 

VII

 

8.9

a86

10/10/1482

38.08

−0.92

 

VIII

 

21.2

b391

01/11/1487

36.83

−2.47

 

VIII

 

15.2

b381

26/01/1494

36.58

−4.33

 

VIII

 

12.6

b357

05/04/1504

37.38

−5.47

 

VIII–IX

 

21.1

f340

09/11/1518

37.23

−1.87

 

VIII–IX

 

15.7

b384

22/09/1522

36.97

−2.67

 

VIII–IX

 

11.1

b380

04/07/1526

37.18

−3.57

 

VII

 

8.0

b362

26/01/1531

39.00

−8.92

 

IX

 

10.3

f304

30/09/1531

37.53

−2.73

 

VIII–IX

 

14.4

b376

29/08/1547

38.75

−0.43

 

VII–VIII

 

10.1

b401

30/01/1579

37.68

−1.70

 

VII

 

3.3

b393

18/06/1581

36.72

−4.42

 

VII

 

10.9

b351

26/12/1598

38.92

−0.12

 

VII–VIII

 

20.5

f264

02/12/1620

38.70

−0.47

 

VII–VIII

 

15.8

b401

19/06/1644

38.80

−0.42

 

VIII

 

9.0

b401

31/12/1658

36.83

−2.47

 

VIII

 

15.2

b381

21/06/1660

42.97

0.07

 

VIII–IX

 

12.3

a67

15/01/1673

38.08

−0.92

 

VIII

 

21.2

b391

28/08/1674

37.68

−1.70

 

VIII

 

3.3

b393

09/10/1680

36.80

−4.60

 

VIII–IX

 

15.7

b353

09/03/1743

38.00

−1.13

 

VII

 

0.9

b391

23/03/1748

39.03

−0.63

 

IX

 

4.5

f265

02/04/1748

39.03

−0.63

 

VII–VIII

 

4.5

f265

09/05/1750

37.20

−7.00

 

VII

 

14.7

f319

24/05/1750

43.07

−0.03

 

VIII

 

12.6

a66

15/06/1750

43.05

0.00

 

VII

 

11.9

a67

07/06/1778

43.10

−0.17

 

VII

 

0.8

a66

25/08/1804

36.77

−2.83

 

VIII–IX

 

6.3

b367

25/08/1804

36.77

−2.83

 

VII

 

6.3

b367

29/08/1804

36.77

−2.83

 

VII

 

6.3

b367

27/10/1806

37.23

−3.73

 

VIII

 

6.7

b361

22/05/1814

43.13

−0.40

 

VII–VIII

 

4.7

a64

18/03/1817

42.30

−2.25

 

VII–VIII

 

9.8

c140

15/09/1828

38.00

−0.70

 

VII

 

13.9

f386

21/03/1829

38.08

−0.68

 

IX–X

 

8.0

f386

05/01/1840

43.00

0.15

 

VII

 

19.3

a67

17/11/1850

43.10

−0.17

 

VII

 

0.8

a66

20/07/1854

43.03

−0.05

 

VII–VIII

 

8.2

a67

05/12/1855

42.83

0.50

 

VII

 

18.0

a70

12/01/1856

36.75

−7.67

 

VII–VIII

 

26.4

g322

11/11/1858

38.30

−8.92

 

IX

 

18.5

g309

22/08/1862

37.05

−5.17

 

VII

 

26.9

b344

15/01/1870

42.87

0.55

 

VII

 

13.1

a70

19/05/1872

39.23

−0.52

 

VII

 

7.9

f69

26/11/1873

43.03

0.15

 

VII

 

20.3

a67

25/12/1884

37.00

−3.98

 

IX–X

 

17.5

b359

05/01/1885

37.00

−3.98

 

VII

 

17.5

b359

27/02/1885

37.00

−3.98

 

VII

 

17.5

b359

11/06/1894

37.12

−2.67

 

VII

 

5.2

b379

10/02/1901

36.75

−5.37

 

VII

 

15.2

b343

25/05/1901

36.70

−3.50

 

VII

 

9.2

b364

06/05/1902

43.00

−0.08

 

VII

 

4.5

a67

09/08/1903

38.30

−9.00

 

VII

 

11.6

g309

13/07/1904

42.70

0.03

 

VIII

 

21.2

a63

04/10/1906

38.70

−9.20

 

VII

 

5.2

f303

16/04/1907

37.80

−1.50

 

VII

 

20.8

b394

22/10/1907

42.40

−0.50

 

VII

 

12.4

a57

29/09/1908

38.10

−1.30

 

VII

 

3.3

b390

23/04/1909

38.95

−8.82

 

IX

 

12.1

f304

24/04/1909

41.70

−8.60

 

VII

 

20.8

d194

28/04/1909

38.90

−9.10

 

VII

 

13.0

f304

04/05/1909

38.60

−7.90

 

VII

 

29.6

f313

04/05/1909

38.90

−8.80

 

VII

 

13.0

f304

02/06/1909

38.90

−8.80

 

VII

 

13.0

f304

11/06/1909

39.90

−8.10

 

VII

 

18.0

d292

12/06/1909

38.90

−8.80

 

VII

 

13.0

f304

01/07/1909

38.00

−0.67

 

VII

 

12.2

f386

06/08/1909

38.90

−8.80

 

VII

 

13.0

f304

12/08/1909

38.90

−8.80

 

VII

 

13.0

f304

14/08/1909

38.90

−8.80

 

VII

 

13.0

f304

17/08/1909

38.90

−8.80

 

VII

 

13.0

f304

15/09/1909

38.90

−8.80

 

VII

 

13.0

f304

16/09/1909

38.90

−8.80

 

VII

 

13.0

f304

28/09/1909

38.90

−8.80

 

VII

 

13.0

f304

05/11/1909

38.90

−8.80

 

VII

 

13.0

f304

16/11/1909

38.90

−8.30

 

VIII

 

21.3

f295

18/11/1909

38.90

−8.80

 

VII

 

13.0

f304

11/01/1910

41.20

−7.80

 

VII

 

8.2

d203

12/01/1910

38.40

−8.00

 

VII

 

30.6

f313

16/06/1910

36.67

−3.37

 

VIII

 

2.9

b364

16/06/1910

36.67

−3.37

 

VII

 

2.9

b364

11/09/1910

38.80

−7.80

 

VII

 

27.4

f295

24/11/1910

43.53

−8.25

 

VII

 

31.8

d180

21/03/1911

38.02

−1.22

 

VIII

 

7.4

b391

03/04/1911

38.10

−1.20

 

VIII

 

9.4

b390

10/05/1911

38.10

−1.20

 

VII

 

9.4

b390

16/05/1911

38.10

−1.20

 

VII

 

9.4

b390

31/05/1911

37.20

−3.70

 

VIII

 

4.2

b362

04/06/1911

37.20

−3.70

 

VII

 

4.2

b362

24/07/1911

43.18

−0.23

 

VII

 

9.8

a66

15/04/1912

41.60

−1.60

 

VII

 

17.6

e132

22/04/1912

37.03

−2.95

 

VII

 

5.6

b368

11/08/1913

36.80

−3.20

 

VII

 

14.8

b365

25/11/1913

37.78

−2.53

 

VII

 

9.1

b375

07/08/1914

42.77

0.53

 

VII

 

19.1

a70

11/08/1914

42.75

0.50

 

VII

 

22.4

a70

23/09/1914

39.00

−8.82

 

VII

 

15.1

f304

25/09/1914

39.00

−8.82

 

VII

 

15.1

f304

28/03/1915

42.53

0.62

 

VII

 

24.7

a68

28/11/1916

38.57

−0.95

 

VII

 

18.0

b400

28/01/1917

38.03

−1.27

 

VII

 

11.4

b390

28/04/1918

37.22

−3.68

 

VII

 

4.8

b362

10/09/1919

38.08

−0.83

 

VIII

5.2

21.1

f368

10/09/1919

38.08

−0.83

 

VII

5.1

21.1

f368

26/11/1920

42.40

−8.60

 

VII

 

29.7

d185

27/07/1922

36.98

−3.57

 

VII

 

7.1

b363

23/09/1922

42.80

2.50

 

III

 

13.2

a84

10/07/1923

42.55

−0.95

 

III

 

16.5

a56

19/11/1923

42.68

0.83

 

III

 

24.3

a70

22/02/1924

43.03

−0.52

 

III

 

10.2

a64

28/02/1926

38.58

−7.90

 

VII

 

28.1

f313

12/03/1927

41.70

2.47

 

VII

 

14.7

d104

18/02/1929

42.13

−2.10

 

VII

5.1

13.0

c140

05/07/1930

37.62

−4.63

 

III

4.9

11.1

f347

03/09/1930

38.07

−1.23

 

VII

3.7

9.1

b390

05/03/1932

37.42

−2.45

 

VIII

4.8

15.6

b378

14/03/1935

37.38

−4.58

 

V

5.0

13.2

f346

28/05/1936

36.70

−5.33

 

VII

4.6

13.4

b343

16/10/1938

43.25

−3.62

 

VII

4.9

13.5

c35

05/03/1940

36.87

−5.33

 

VII

4.6

15.5

b342

03/10/1940

39.60

−9.10

 

VII

0.0

22.0

f296

26/12/1943

42.95

0.22

 

VII

4.5

24.5

a67

23/02/1944

38.17

−1.15

 

VII

3.8

13.9

b390

01/07/1945

38.80

−0.58

 

VII

4.8

22.7

b401

23/06/1948

38.14

−1.76

 

III

5.0

25.4

b387

12/08/1948

40.07

−8.58

 

VI

4.9

16.3

c210

18/11/1948

41.50

−8.50

 

V

4.8

10.3

d194

31/01/1950

43.12

0.20

 

VII

4.1

28.9

a67

10/03/1951

38.18

−3.82

 

VIII

4.8

19.2

d282

19/05/1951

37.58

−3.93

 

VIII

5.1

23.1

b349

07/02/1952

43.10

−0.70

 

VII

4.3

17.9

a59

05/04/1952

43.00

0.00

 

VII

4.5

7.9

a67

28/09/1953

41.13

−1.58

 

VII

4.7

14.4

d153

13/10/1953

43.00

0.20

 

VII

4.5

23.2

a67

10/11/1953

36.70

−7.10

 

V

4.9

29.4

g329

08/01/1954

36.93

−3.88

 

VIII

4.2

6.3

b359

04.06.1955

37.13

−3.65

5

VII

5.1

5.6

b362

19.04.1956

37.19

−3.68

5

VIII

5.0

2.1

b362

16.08.1956

36.91

−8.61

5

VI

5.0

10.8

b323

08.08.1958

41.18

2.60

5

V

4.9

21.2

g102

25/11/1958

42.86

0.10

5

VII

4.6

18.4

a67

06/09/1960

37.10

−5.09

5

 

4.9

22.9

b344

10/02/1961

41.73

−6.20

 

VI

5.2

13.4

d199

03/09/1961

41.93

−2.08

 

VIII

4.6

9.4

c144

11/02/1962

37.23

−2.11

5

IV

4.8

26.2

b377

02/11/1962

42.23

2.28

 

V

5.1

5.5

a91

09/06/1964

37.74

−2.57

5

VIII

4.8

4.8

b375

09/09/1964

37.09

−3.62

5

VII

4.3

10.6

b362

07/01/1965

43.20

−4.10

 

VII

4.1

14.8

c30

13/08/1967

43.30

−0.68

5

VIII

5.3

15.5

f65

05/04/1970

42.48

1.69

 

V

4.8

21.2

a77

16/03/1972

37.42

−2.24

5

VII

4.8

33.4

b378

14/06/1972

36.66

−8.56

28

IV

4.8

24.0

g323

13/12/1973

43.13

−0.25

 

VII

4.2

6.6

a66

07/08/1975

36.42

−4.59

28

IV

5.2

3.8

b356

07/04/1978

38.34

−8.67

5

III

5.1

9.0

f306

14/08/1978

36.37

−6.98

31

IV

5.0

19.9

g329

29/02/1980

43.19

−0.36

5

VII

4.9

11.9

a64

05/03/1981

38.49

0.22

20

V

4.9

18.6

f392

06/01/1982

43.27

−0.98

24

VI

4.8

5.8

a58

25/02/1984

43.23

−1.14

10

IV

4.8

8.0

a58

24/06/1984

36.84

−3.74

5

V

5.0

10.9

b359

13/09/1984

36.98

−2.34

9

V

5.0

8.6

b377

13.09.1984

37.05

−2.40

5

 

4.8

8.9

b377

26.05.1985

37.79

−4.64

5

V

5.1

7.8

f347

20.12.1989

37.23

−7.39

23

VI

5.0

4.2

f320

23.12.1993

36.78

−2.94

8

VII

5.0

6.3

b367

04/01/1994

36.57

−2.82

2

VII

4.9

10.1

g366

15/05/1995

40.87

1.61

2

IV

4.9

23.5

g110

18/02/1996

42.82

2.60

12

V

5.0

11.3

a84

21/05/1997

42.78

−7.26

13

VI

5.1

9.5

c181

22/05/1997

42.85

−7.29

17

 

4.9

1.4

c181

27/10/1998

42.86

−2.01

10

V

5.2

18.0

a52

06/08/2002

37.89

−1.84

1

V

4.8

7.7

b387

04/02/2002

37.09

−2.54

1

V

5.1

7.2

b380

17/11/2006

43.01

0.00

 

V

4.8

8.6

a67

The catalogue available for the region (Martínez-Solares and Mezcua 2002; IGN 2007) gives for some earthquakes half-integer epicentral intensity values (Table 1) that formally do not belong to the EMS scale. Nevertheless, this does not affect our results since we consider the intensity threshold to be VII.

The distribution of the epicenters by their distances to the nearest intersection of lineaments is given in Table 2. For 181 epicenters the distance between epicenters and the nearest intersections does not exceed 25 km and most characteristic distances are in the interval from 0 to 20 km. The distance is greater than 25 km for 15 events out of 196 main shocks reported by Table 1. This discordance, with respect to our node definition, is acceptable in view of the accuracy of epicenter locations. In spite of these exceptions, we can conclude that the earthquakes considered nucleate at some of the nodes. Therefore, it is possible to apply pattern recognition to pinpoint other nodes where earthquakes with M ≥ 5.0 are possible.
Table 2

Distribution of the main shocks (Table 1) according to their distances from the intersections of lineaments

Mountain countries and macroblocks

Number of epicenters in different intervals of the distance

0–10 km

10.1–20 km

20.1–25 km

25.1–30 km

>30.1 km

(a) Pyrenees (47)

18

20

7

1

1

(b) Betics (72)

36

27

5

3

1

(c) Cantabrian Mountains, Iberian Chains, Central System, Estrela Range (9)

5

4

   

(d) Catalan Coastal Ranges, Galicia, Meseta and Toledo Mountains (12)

1

8

1

1

1

(e) Ebro, Duero, and Tagus basins (1)

 

1

   

(f) Portuguese, Low Tagus and Guadalquivir basins (45)

9

25

7

3

1

(g) Continental slope (10)

 

4

3

2

1

Number of main shocks in Table 1 (196)

69, 35%

89, 45%

23, 12%

10, 5%

5, 3%

Numerals in the brackets depict the number of main shocks from Table 1 within a given unit(s)

5 Recognition of seismogenic nodes in the Iberian Peninsula

5.1 Selection of the training sets

The recognition procedure starts with the selection of the sample nodes. The selection of sample nodes for class D has been done in a slightly different way for groups I–III and for group IV.

5.1.1 Sample nodes for groups I–III

As it can be seen in Fig. 4 in many cases one epicenter can be attributed to several nodes. Therefore, we used the CLUSTERS algorithm that permits to assemble D0 from a number of subsets \( {\mathbf{D}}_{{\mathbf{0}}}^{{\mathbf{1}}} , \, {\mathbf{D}}_{{\mathbf{0}}}^{{\mathbf{2}}} , \ldots , \, {\mathbf{D}}_{{\mathbf{0}}}^{{\mathbf{K}}} \), each of which has to include at least one node hosting a recorded earthquake with target size and may have some nodes from class N (see supplementary material). To avoid a possible uncertainty in location and size of historical events, we have assigned to D0 only the subclasses related to earthquakes after 1900. Nodes marked with (*) in Table 3 compose subclasses D0 for groups I–III.
Table 3

Grouping of the nodes for recognition

Group I (93 nodes a–b)

Group II (155 nodes c–d)

Group III (98 nodes e–f)

Group IV (59 nodes g)

a37, a44, a45, a50*, a51*, a52*, a55*, a56*, a57*, a58*, a59*, a60*, a62, a63*, a64*, a66*, a67*, a68*, a70*, a71*, a72, a73, a74, a75, a76, a77*, a78, a79, a80, a81, a82*, a83*, a84*, a85, a86*, a90*, a91*, a92, a95*, a96*, b335, b336, b337, b342*, b343*, b344*, b345, b349*, b350, b351*, b352*, b353*, b354, b355, b356*, b357*, b358*, b359*, b360, b361*, b362*, b363*, b364*, b365*, b367*, b368*, b369*, b370, b371, b372, b373, b374, b375*, b376*, b377*, b378*, b379*, b380*, b381*, b382, b383, b384*, b385*, b387*, b389, b390* b391*, b393*, b394*, b399*, b400*, b401*, b403

d3, d4, d5, d6, c7, d8, c10, c11, c12, c13, c14, c15, c16, c17, c18, c19, c26, c27, c28, c29, c30*, c31*, c35*, c36, c38, c39, c40, d93, c97*, c98*, d103*, d104*, d105*, d106, d107*, d108, d112, d113, d114, 115, d116, d117, d118, c121, d122, d135*, c136*, c139*, c140*, c141, c142, c143, c144*, c145, c146, d147, d148, c149, d150, d151, d152, d153*, d154, c155*, c156, c157, c172, c173, c174, d175, d178, d179, d180, c181*, c182, c183, d184, d185, d187, d191*, d193, d194*, d195, d196, d197, c198, d199*, d200, d201, d202, d203*, c204, c205, c210*, c211, d212, −d213, c214, d215, d216, c217, c218, c219, c220, c221, c222, c223, d224, c225, c226, c227, c228, c229, d230, c233, c234, d235, d236, d237, d238, d239, d241, d242, c244, c245, c246, d255, d257, d258, d259, d260, d266, b267, d270, b271, d275, d276, d277, d280, d281, d282*, d284, d285, d286, d287, d288, d290, d291, d292*, d315, d316, d317, d318, d388, d405

e41, e42, e43, f48, f49, e53, e54, e61, f65*, f69*, f89, f94, f111, f119, e123, e124, e125, e126, e127, e128, e129, e130, e131*, e132*, e133, e134, e137*, e138*, e158, e159, e160, e161, e162, e163, e164, e165, e166, e167, e168, e169, e170, e171, f186, f190, f206, f207, e243, e247, e248, e249, 240, e250, e251, e252, e253, e254, e256, f261, f264*, f265*, f268, f269, f272, e273, e274, e278, e279, f283, f289, f293, f294, f295*, f296*, f302, f303*, f304*, f305*, f306*, f311, f312, f313, f314, f319*, f320*, f321, f324, f325, f330, f331, f338, f339, f340*, f341*, f346*, f347*, f348*, f386*, f396

g1, g2, g9, g20, g21, g22, g23, g24, g25, g32, g33, g34, g46, g47, g87, g88, g99, g100, g101, g102* g109, g110*, g120, g176, g177, g188, g189, g192, g208, g209, g231, g232, g262, g263, g297, g298, g299, g300, g301, g307, g308, g309*, g310, g322, g323*, g326, g327, g328, g329*, g332, g333, g334, g366*, g392*, g395, g397, g398, g402, g404

Subclasses related to the events which occurred before 1900 have been included in the set X that is not employed at the learning stage of the recognition; the nodes forming set X are classified at the recognition stage. X also includes nodes hosting events smaller than the target ones. Here, in the set X, we include the nodes located at the distance of 26–50 km from the earthquakes listed in Table 1. The remaining nodes in each group have been assigned to the set N0. The distances from the nodes included in the sets N0 from the epicenters listed in Table 1 exceeds 50 km.

5.1.2 Sample nodes for group IV

To classify nodes of group IV, we used the CORA-3 algorithm because each earthquake related to the continental slope can be associated with one node. All 59 nodes of group IV were also a priori divided into three sets. The set D0 includes seven nodes from which the distance to the epicenters does not exceed 25 km. The set X has been assembled in the same way as for the CLUSTER algorithm and included 27 nodes. The remaining 25 nodes have been assigned to the set N0.

5.2 Parameters of the nodes used for recognition

A uniform parameterization of the nodes in the form of a common questionnaire is needed in order to apply the pattern recognition technique. In general, a questionnaire should reflect the characteristics of tectonic and geological environments responsible for the present-day seismic level. The requirement of uniform description leads to unavoidable loss of information on geological, geomorphic, and geophysical characteristics that might be important for the identification of earthquake-prone areas.

Accordingly with Gorshkov et al. (2004), we use the parameters listed in Table 4. The parameters describing the topographic altitudes and the area of soft sediments characterize indirectly the contrast and intensity of neotectonic movements, while those describing the geometry of the lineament network and gravity anomalies can be related to the degree of crust fracturing and deep heterogeneity. The values of the parameters have been measured within each node, i.e. inside a circle with 25 km of radius, from available topographic, geological, and gravity maps as well as from the MZ map (Fig. 4).
Table 4

Parameters used for pattern recognition

(A) Topographic parameters

 Maximum topographic altitude, m (Hmax)

 Minimum topographic altitude, m (Hmin)

 Relief energy, m (ΔH) (Hmax − Hmin)

 Distance between the points Hmax and Hmin, km (L)

 Slope (ΔH/L)

(B) Geological parameters

 The portion of the node area covered by soft (Quaternary) sediments, % (Q)

(C) Parameters of lineament-and-block geometry

 The highest rank of lineament in a node (HR)

 Number of node-forming lineaments (NL)

 Distance to the nearest 1st rank lineament, km (D1)

 Distance to the nearest 2nd rank lineament, km (D2)

 Distance to the nearest node, km (Dn)

(D) Gravity parameters

 Maximum value of Bouguer anomaly, mGal (Bmax)

 Minimum value of Bouguer anomaly, mGal (Bmin)

 Difference between Bmax and Bmin, mGal (ΔB)

Since the CORA-3 algorithm operates in a binary vector space, the factual values of the parameters were transformed into the binary vector space by discretization and coding. The range of the value of each parameter was divided into two or three parts (interval open to the left) by specifying one or two thresholds of discretization. The one-threshold discretization considers two intervals of the factual values, which are converted into one binary component with the value 1 (“small”) or 0 (“large”). Correspondingly, in the two-threshold discretization, the factual values of the parameters are converted into two binary components with the values 11 (“small”), 01 (“medium”) or 00 (“large”), and two thresholds are specified. The discretization has been done individually for each group of nodes with the a priori division of the nodes into subsets D0, N0 and X. The thresholds of discretization are given in Table 5.
Table 5

Values of discretization thresholds used to convert the factual values of the parameters into two binary components

Parameters (as in Table 4)

Group I

Group II

Group III

Group IV

Hmax (m)

1,418

 

2,030

1,098

 

1,500

653

 

1,094

−60

 

100

Hmin (m)

80

 

490

 

400

 

150

 

480

−2,050

 

−1,200

L (km)

23

 

35

30

   

28

 

44

 

47

Q

 

15

 

3

 

10

 

20

 

90

 

95

HR

1

 

2

1

 

2

 

1

  

1

 

NL

 

2

  

2

  

2

  

2

 

D1 (km)

0

 

43

0

 

48

 

0

  

0

 

D2 (km)

0

 

30

0

 

42

24

 

56

42

 

64

Dn (km)

23

 

29

23

 

30

 

32

 

24

 

37

Bmax (mGal)

 

−30

  

−30

  

−38

 

25

 

45

Bmin (mGal)

 

−90

 

−75

 

−50

−75

 

−50

 

−20

 

B (mGal)

45

 

75

 

30

  

30

 

45

 

60

H (m)

1,160

 

2,000

656

 

1,130

380

 

780

1,200

 

2,020

∆H/L

46.4

 

68.9

24.4

 

38.6

 

23

 

30

 

45.7

One value implies that the discretization of a given parameter has been done with one-threshold discretization, while two values depict that the factual values of a given parameter have been descritizated using the two-threshold discretization

5.3 Results of recognition

The recognition has been performed individually for each of the four groups. The recognized nodes, capable of earthquakes with M ≥ 5.0 or I0 ≥ VII, are plotted in Fig. 6 by circles with the radius of 25 km.
https://static-content.springer.com/image/art%3A10.1007%2Fs12210-010-0075-3/MediaObjects/12210_2010_75_Fig6_HTML.gif
Fig. 6

Nodes prone to earthquakes with M ≥ 5.0 or I0 ≥ VII. Circles denote nodes prone to earthquakes with M ≥ 5.0 or I0 ≥ VII. Dots are epicentres of the earthquakes with M ≥ 5.0 or I0 ≥ VII reported by Table 1. Black numerals denote some nodes referred in the text. Lines are the same as in Fig. 3

5.3.1 Group I

The 93 nodes located in mountain environments have been classified with CLUSTERS. With k1 = 23, \( \bar{k}_{1} = 1 \), k2 = 2, and \( \bar{k}_{2} = 0 \), the algorithm selected 11 D traits and 6 N traits (Table 6) when Δ = 1. The classification has been made with Δ = 1, i.e. a node is assigned to the D set, if the difference between the number of D and N traits, which a given node possess is greater or equal to 1 (see supplementary material). D nodes are shown by circles in Fig. 6 and marked by (+) in Table 7 that shows numbers of D and N traits for each node of group I. All nodes of group I hosting the target earthquakes have been recognized as D (Fig. 6). Furthermore, all nodes hosting historical events which occurred before 1900 are classified as D although they have not been included in D0, therefore they do not control the decision rule. The exception is the ancient 1373 event that is not related with the nodes delineated in the Pyrenees.
Table 6

Characteristics traits of D and N nodes from group I (objects a–b)

No.

Parameters

Hmin (m)

L (km)

Q (%)

NL

D1 (km)

Dn (km)

Bmin (mGal)

ΔB (mGal)

ΔH (m)

ΔH/L

D traits

 1

         

>68.9

 2

       

>45

 

>46.4

 3

      

≤−90

  

>46.4

 4

     

≤29

   

>46.4

 5

   

2

     

>46.4

 6

        

>2,000

 

 7

   

2

 

≤29

  

>1,160

 

 8

     

>23

 

>45

  

 9

   

2

   

>45

  

 10

  

≤15

  

≤29

    

 11

    

>0

     

N traits

 1

≤490

≤35

       

≤46.4

 2

   

2

0

   

≤2,000

 

 3

    

0

>23

 

≤45

  

 4

≤80

   

0

  

≤45

  

 5

 

≤23

  

0

 

>−90

   

 6

 

≤35

  

0

>29

    
Table 7

Voting and classification of the nodes of group I

Node

Numbers of D and N traits possessed by nodes

Nodes of set D0

Subclass 1

 b391

6: 0+

Subclass 2

 a95

2: 1+

 a96

3: 1+

Subclass 3

 b384

5: 1+

 b385

6: 0+

Subclass 4

 a96

3: 1+

Subclass 5

 a90

4: 1+

 a91

7: 0+

 a96

3: 1+

Subclass 6

 a82

8: 0+

 a83

11: 0+

 a91

7: 0+

Subclass 7

 b361

4: 0+

 b362

8: 0+

 b363

11: 0+

Subclass 8

 a84

6: 0+

 a86

2: 2

Subclass 9

 b380

9: 0+

 b381

1: 0+

Subclass 10

 b351

6: 0+

 b357

0: 0

Subclass 11

 b379

10: 0+

 b380

9: 0+

Subclass 12

 b361

4: 0+

 b362

8: 0+

Subclass 13

 b375

4: 0+

 b376

5: 0+

 b378

4: 0+

Subclass 14

 b401

3: 1+

Subclass 15

 b393

3: 0+

Subclass 16

 b351

6: 0+

 b352

5: 0+

 b357

0: 0

Subclass 17

 a66

4: 0+

 a67

9: 0+

Subclass 18

 b351

6: 0+

 b353

8: 0+

Subclass 19

 b390

3: 0+

 b391

6: 0+

 b394

7: 0+

Subclass 20

 a64

5: 0+

 a66

4: 0+

 a67

9: 0+

Subclass 21

 b365

5: 0+

 b367

8: 0+

Subclass 22

 a64

5: 0+

 a66

4: 0+

Subclass 23

 b399

3: 0+

Subclass 24

a67

9: 0+

Subclass 25

 a70

7: 0+

Subclass 26

 a70

7: 0+

 a71

3: 2+

Subclass 27

 b359

9: 0+

Subclass 28

 b342

9: 0+

 b343

3: 0+

Subclass 29

 b358

7: 0+

 b364

8: 0+

Subclass 30

 a63

7: 0+

Subclass 31

 b393

3: 0+

 b394

7: 0+

Subclass 32

 a57

4: 0+

 a60

4: 0+

Subclass 33

 b364

8: 0+

Subclass 34

 b368

9: 0+

 b369

8: 0+

Subclass 25

 b364

8: 0+

 b365

5: 0+

Subclass 36

 b375

4: 0+

Subclass 37

 a68

7: 0+

Subclass 38

 b400

2: 0+

Subclass 39

 b359

9: 0+

 b362

8: 0+

 b363

11: 0+

Subclass 40

 a84

6: 0+

Subclass 41

 a55

1: 0+

 a56

4: 0+

Subclass 42

 a64

5: 0+

Subclass 43

 b378

4: 0+

Subclass 44

 b343

3: 0+

Subclass 45

 b390

3: 0+

 b391

6: 0+

Subclass 46

 b349

1: 0+

Subclass 47

 a59

7: 0+

 a64

5: 0+

Subclass 48

 b344

5: 0+

Subclass 49

 a91

7: 0+

Subclass 50

 a77

6: 0+

Subclass 51

 b35

1: 0+

Subclass 52

 a58

3: 0+

 a59

7: 0+

Subclass 53

 a50

1: 0+

 a58

3: 0+

 a59

7: 0+

Subclass 54

 b358

7: 0+

 b359

9: 0+

 b363

11: 0+

Subclass 55

 b377

4: 0+

 b380

9: 0+

 b381

1: 0+

Subclass 56

 b377

4: 0+

 b380

9: 0+

Subclass 57

 b365

5: 0+

 b367

8: 0+

Subclass 58

 a51

2: 0+

 a52

0: 0

Subclass 59

 b387

6: 0+

Subclass 60

 b377

4: 0+

 b379

10: 0+

 b380

9: 0+

Nodes of set N0

 a44

1: 2

 a62

2: 1+

 a74

1: 3

 a79

0: 1

 a80

0: 2

 a92

0: 3

 b271

0: 2

 b335

2: 3

 b372

1: 2

 b373

0: 2

Nodes of set X

 a37

1: 0+

 a45

2: 0+

 a72

7: 1+

 a73

6: 0+

 a75

0: 4

 a76

8: 0+

 a78

5: 0+

 a81

1: 0+

 a85

0: 2

 b267

4: 0+

 b336

6: 0+

 b337

1: 0+

 b345

3: 1+

 b350

0: 2

 b354

9: 0+

 b355

3: 0+

 b360

3: 0+

 b370

6: 0+

 b371

7: 0+

 b374

6: 0+

 b382

1: 3

 b383

1: 4

 b389

1: 0+

 b403

5: 0+

+ nodes recognized to be potential for M ≥ 5.0 or I0 ≥ VII

5.3.2 Group II

155 nodes located in low elevated mountains and plateaus have been classified with CLUSTERS. With k1 = 2, \( \bar{k}_{1} = 1 \), k2 = 26, and \( \bar{k}_{2} = 0 \), the algorithm selected five D traits and seven N traits (Table 8) when Δ = 1. D nodes are shown by circles in Fig. 6 and marked by (+) in Table 9. With the exception of node 199, marked with the 10/02/1961 earthquake (M = 5.2), nodes from group II hosting target earthquakes have been recognized as D. The 24/11/1910 earthquake of I0 = VII is not associated with any node. The nearest node 180 located at the distance of 31.8 km from this event is recognized as N.
Table 8

Characteristics traits of D and N nodes from group II (objects c-d)

No.

Parameters

Hmax (m)

Hmin (m)

L (km)

Q (%)

HR

NL

D2 (km)

Dn (km)

Bmax (mGal)

Bmin (mGal)

ΔB (mGal)

ΔH (m)

ΔH/L

D traits

 1

      

>42

 

>−30

   

>38.6

 2

≤1,500

        

≤−75

 

>656

 

 3

      

0

  

>−50

>30

  

 4

>1,098

      

≤23

 

>−50

   

 5

>1,500

≤400

   

>2

       

N traits

 1

 

>400

     

≤30

    

≤38.6

 2

 

>400

    

≤42

     

≤38.6

 3

  

<30

       

<30

≤1,130

 

 4

   

≤3

     

≤−50

   

 5

   

≤3

    

<−30

    

 6

   

≤3

1st or 2nd

        

 7

 

>400

 

≤3

         
Table 9

Voting and classification of the nodes of group II

Node

Numbers of D and N traits possessed by nodes

Nodes of set D0

Subclass 1

 d282

0: 2

Subclass 2

 c97

1: 0+

Subclass 3

 c97

1: 0+

 c98

0: 0

Subclass 4

 d103

0: 0

 d104

1: 0+

 d105

2: 0+

 d107

1: 0+

Subclass 5

 c139

0: 1

 c140

1: 0+

Subclass 6

 d191

0: 0

 d194

1: 0+

Subclass 7

 d292

1: 0+

Subclass 8

 d203

1: 0+

Subclass 9

 c97

1: 0+

 d103

0: 0

 d104

1: 0+

 d105

2: 0+

Subclass 10

 c136

1: 0+

 c140

1: 0+

Subclass 11

 c35

2: 0+

Subclass 12

 c210

1: 0+

Subclass 13

 d194

1: 0+

Subclass 14

 d153

1: 0+

 c155

0: 0

Subclass 15

 d199

0: 0

Subclass 16

 d135

1: 0+

 c144

0: 0

Subclass 17

 c30

1: 0+

 c31

2: 0+

Subclass 18

 c181

1: 0+

Nodes of set D0

 d3

0: 0

 d4

1: 1

 d6

0: 0

 d8

0: 0

 c10

0: 0

 c11

0: 0

 c13

0: 4

 c14

0: 1

 c15

0: 3

 c16

0: 2

 c17

0: 2

 c18

0: 0

 c19

1: 0+

 c27

0: 2

 c28

0: 0

 c40

0: 0

 d93

0: 5

 d112

0: 2

 d113

0: 1

 d114

0: 0

 d115

0: 1

 d116

0: 0

 d117

1: 1

 d118

0: 0

 c121

0: 2

 d122

0: 0

 d147

0: 2

 d148

0: 1

 d150

0: 3

 d151

0: 6

 d152

0: 6

 c157

0: 6

 c172

0: 0

 c173

0: 1

 c174

0: 1

 d175

0: 0

 d178

0: 0

 d179

0: 1

 c183

0: 3

 d184

0: 2

 d195

0: 1

 d196

0: 7

 d197

0: 7

 c198

0: 4

 d201

0: 4

 c205

0: 0

 d212

0: 4

 d213

0: 3

 c214

0: 2

 d215

0: 4

 d216

0: 3

 c217

0: 3

 c218

0: 3

 c219

0: 4

 c220

0: 4

 c221

0: 5

 c222

0: 2

 c223

0: 7

 d224

0: 6

 c225

0: 6

 c226

0: 4

 c227

0: 2

 c228

0: 3

 c229

0: 0

 d230

0: 0

 c233

0: 4

 c234

0: 6

 d235

0: 2

 d236

0: 6

 d237

0: 1

 d238

0: 2

 d239

0: 2

 d241

0: 0

 d242

0: 2

 c244

0: 4

 c245

0: 3

 c246

0: 3

 d255

0: 2

 d259

1: 0+

 d260

0: 0

 d270

0: 2

 d275

0: 0

 d276

0: 0

 d277

0: 1

 d280

0: 4

 d285

0: 0

 d286

0: 0

 d287

0: 1

 d288

0: 1

 d290

0: 2

 d315

0: 1

 d316

0: 0

 d317

0: 1

 d318

0: 0

 d388

0: 1

 d405

0: 0

Nodes of set X

 d5

1: 1

 c7

0: 4

 c12

0: 0

 c26

1: 0+

 c29

0: 0

 c36

1: 0+

 c38

0: 2

 c39

0: 3

 d106

0: 2

 d108

0: 0

 c141

1: 0+

 c142

1: 0+

 c143

0: 2

 c145

0: 0

 c146

0: 1

 c149

0: 5

 d154

0: 7

 c156

0: 3

 d180

0: 0

 c182

0: 3

 d185

1: 0+

 d187

1: 0+

 d193

1: 0+

 d200

0: 2

 d202

0: 1

 c204

0: 0

 c211

0: 1

 d257

0: 1

 d258

0: 5

 d266

1: 0+

 d281

0: 4

 d284

0: 0

 d291

0: 0

+ nodes recognized to be potential for M ≥ 5.0 or I0 ≥ VII

5.3.3 Group III

98 nodes located in basins have been classified with CLUSTERS. With k1 = 8, \( \bar{k}_{1} = 4 \), k2 = 20, and \( \bar{k}_{2} = 0 \), the algorithm selected four D traits and four N traits (Table 10) when Δ = 1. D nodes are shown by circles in Fig. 6 and marked by (+) in Table 11. All nodes composing subset D0 for group III are recognized correctly. In the Lower Tagus basin three events (04/05/1909, 12/01/1910 and 11/09/1910) are located outside of the identified nodes.
Table 10

Characteristics traits of D and N nodes of group III (objects e-f)

No.

Parameters

Hmax (m)

Hmin (m)

Q (%)

D2 (km)

Dn (km)

ΔB (mGal)

ΔH (m)

D traits

1

  

>20

  

>30

>380

2

  

>20

≤56

  

>380

3

 

≤480

>20

   

>380

4

 

≤480

>20

 

>32

  

N traits

1

>653

    

≤30

≤780

2

  

≤20

   

≤780

3

  

≤20

>24

   

4

 

>480

 

>24

   
Table 11

Voting and classification of the nodes of group III

Node

Numbers of D and N traits possessed by nodes

Nodes of set D0

Subclass 1

 f348

1: 0+

Subclass 2

 f304

3: 0+

Subclass 3

 f264

3: 0+

Subclass 4

 f340

0: 0

 f341

3: 0+

Subclass 5

 f265

3: 0+

Subclass 6

 f319

1: 0+

Subclass 7

 e137

0: 2

 e138

0: 2

Subclass 8

 f386

2: 0+

Subclass 9

 f305

2: 0+

Subclass 10

 f69

2: 0+

 f26

3: 0+

Subclass 11

 f303

4: 0+

Subclass 12

 f303

4: 0+

 f304

3: 0+

Subclass 13

 f295

1: 0+

Subclass 14

 e131

0: 0

 e132

2: 0+

Subclass 15

 f346

4: 0+

 f347

4: 0+

Subclass 16

 f346

4: 0+

Subclass 17

 f296

1: 0+

Subclass 18

 f65

1: 0+

Subclass 19

 f305

2: 0+

 f306

0: 0

Subclass 20

 f347

4: 0+

Subclass 21

 f320

1: 0+

Nodes of set N0

 e41

0: 2

 e42

0: 2

 e43

0: 4

 e61

0: 2

 f119

2: 0+

 e123

0: 2

 e124

0: 1

 e125

0: 0

 e126

0: 1

 e127

0: 0

 e128

0: 2

 e129

0: 2

 e133

1: 0+

 e158

0: 2

 e159

0: 2

 e160

0: 2

 e161

0: 0

 e162

0: 0

 e165

0: 2

 e166

0: 2

 e167

0: 3

 e168

0: 3

 e169

0: 3

 e170

0: 4

 e171

0: 2

 f186

2: 0+

 f207

1: 0+

 e240

0: 2

 e243

0: 0

 e247

0: 1

 e248

0: 1

 e249

0: 2

 e250

0: 1

 e251

0: 2

 e252

0: 1

 e253

0: 1

 e254

0: 4

 e256

0: 2

 e268

0: 4

 e269

0: 4

 e272

0: 4

 e273

0: 4

 e274

1: 1

 e278

0: 2

 e279

0: 3

 f289

0: 0

 f311

1: 0+

 f325

0: 2

 f330

3: 0+

 f331

0: 0

 f338

1: 0+

Nodes of set X

 f48

0: 0

 f49

3: 0+

 e53

0: 0

 e54

0: 1

 f89

2: 0+

 f94

1: 0+

 f111

3: 0+

 e130

0: 3

 e134

3: 1+

 e163

0: 1

 e164

0: 2

 f190

3: 0+

 f206

3: 0+

 f261

3: 0+

 f283

0: 2

 f293

0: 1

 f294

1: 0+

 f302

3: 0+

 f312

1: 0+

 f313

0: 2

 f314

0: 2

 f321

1: 0+

 f324

2: 0+

 f339

0: 0

 a396

3: 0+

+ nodes recognized to be potential for M ≥ 5.0 or I0 ≥ VII

5.3.4 Group IV

58 nodes sitting on the continental slope have been classified with CORA-3. With k1 = 2, \( \bar{k}_{1} = 1 \), k2 = 15, and \( \bar{k}_{2} = 0 \), the algorithm selected six D traits and five N traits (Table 12) when Δ = 1. D nodes are shown by circles in Fig. 6 and marked by (+) in Table 13. All nodes from the subset D0 are recognized correctly.
Table 12

Characteristics traits of D and N nodes of group IV (objects g)

No.

Parameters

Hmax (m)

Hmin (m)

L

Q (%)

NL

D2 (km)

Dn (k)m

Bmax

Bmin

ΔB (mGal)

ΔH (m)

D traits

 1

    

2

    

≤45

≤2,020

 2

>−60

    

>64

     

 3

   

>95

 

≤64

    

≤1,200

 4

    

>2

   

≤−20

≤60

 

 5

 

≤−1,200

     

≤25

 

≤60

 

 6

  

≤44

   

>24

  

≤60

 

N traits

 1

         

>45

>1,200

 2

  

>44

      

>45

 

 3

 

≤−1,200

       

>45

 

 4

 

≤−1,200

     

>25

   

 5

≤100

≤−1,200

         
Table 13

Voting and classification of the nodes of group IV

Node

Numbers of D and N traits possessed by nodes

Nodes of set D0

 g102

3: 0+

 g110

2: 0+

 g309

3: 0+

 g323

2: 0+

 g329

3: 0+

 g366

1: 0+

 g392

2: 0+

Nodes of set N0

 g1

0: 5

 g2

0: 1

 g9

0: 4

 g20

0: 5

 g21

0: 4

 g22

0: 2

 g23

0: 5

 g24

1: 0+

 g46

0: 3

 g47

1: 0+

 g88

0: 1

 g177

1: 5

 g188

0: 5

 g189

0: 5

 g192

0: 5

 g208

0: 5

 g209

0: 5

 g231

0: 1

 g232

0: 4

 g297

0: 4

 g298

0: 4

 g299

0: 4

 g300

0: 4

 g333

0: 1

 g334

1: 0+

 g397

0: 3

 g404

1: 0+

Nodes of set X

 g25

0: 3

 g32

2: 4

 g33

0: 4

 g34

1: 0+

 g87

0: 1

 g99

3: 1+

 g100

0: 5

 g101

1: 4

 g109

3: 1+

 g120

0: 1

 g176

1: 1

 g262

2: 0+

 g263

2: 0+

 g301

0: 3

 g307

0: 5

 g308

1: 3

 g310

1: 0+

 g322

2: 0+

 g326

1: 1

 g327

0: 5

 g328

1: 0+

 g332

3: 0+

 g395

0: 3

 g398

2: 0+

 g402

0: 0

+ nodes recognized to be potential for M ≥ 5.0 or I0 ≥ VII

5.4 Control tests

The reliability of the recognition results can be indirectly evaluated by a set of control tests relevant to the determination of earthquake-prone areas (Gorshkov et al. 2003). We define as main variants the classifications presented in Sect. 4.3 and shown in Fig. 6. Four tests have been run to search the stability of the node classifications obtained for each group of the nodes. The results of the tests performed support the validity of the main variants. During the control tests 3–11% of nodes of each group change its classification with respect to the main variants. Such insignificant changes with respect to the main variants are in agreement with the empirical stability criteria developed by Gvishiani et al. (1988).

6 Seismogenic nodes and elements at risk

Information on the potential seismic sources affecting elements at risk (i.e. population, buildings, lifelines and critical facilities) represents a main issue for seismic hazard evaluation. Seismogenic nodes prone to earthquakes with M ≥ 5.0 or I0 ≥ VII as obtained in this work may contribute to the definition of earthquake source zones in the Iberian Peninsula. Figure 7 displays the locations of main metropolitan areas (larger than 500,000 inhabitants) in the Iberia Peninsula and power plant facilities (both nuclear and water power stations) in Spain with respect to the recognized seismogenic nodes. Table 14 reports the distance from each metropolitan area and power plant to the nearest seismogenic node. As it is seen from Table 10 and Fig. 7, the Cofrentes NPP and the Iznajar WPP are located within recognized D nodes. All main metropolitan areas, apart from Madrid and Oviedo, are also situated at recognized seismogenic nodes. This information can contribute to the revision of seismic risk for objects located near recognized nodes prone to M ≥ 5.0 or I0 ≥ VII.
https://static-content.springer.com/image/art%3A10.1007%2Fs12210-010-0075-3/MediaObjects/12210_2010_75_Fig7_HTML.gif
Fig. 7

Seismogenic nodes and location of main metropolitan areas in the Iberia Peninsula and power nuclear and water plants in Spain. Circles depict the recognized seismogenic nodes. Smaller circles mark earthquakes from Table 1. Dots show water power plants and operating nuclear power plants

Table 14

Distances from main metropolitan areas in the Iberia Peninsula and power plants in Spain to recognized seismogenic nodes

Site

Number of the nearest seismogenic node

Distance to the nearest seismogenic node (km)

Almaraz (NPP)

292

220

Asco (NPP)

119

60

Cofrentes (NPP)

69

15

Trillo (NPP)

153

80

Santa Maria de Garoña (NPP)

142

60

Vandellos II (NPP)

110

40

Alcántara (WPP)

294

125

Almendra (WPP)

203

125

Buendía (WPP)

153

120

Mequinenza (WPP)

133

95

Cíjara (WPP)

348

190

Valdecañas (WPP)

292

120

Esla/Ricobayo (WPP)

203

155

Iznájar (WPP)

360

25

Gabriel y Galán (WPP)

203

175

Contreras (WPP)

259

60

Alicante

267

6

Barcelona

108

3

Bilbao

37

11

Lisboa

303

0

Madrid

153

185

Málaga

351

8

Murcia

391

0

Oviedo

19

54

Porto

206

5

Valencia

261

14

Zaragoza

133

8

NPP nuclear power plant, WPP water power plant

7 Discussion and conclusions

Figure 6 shows that the overwhelming majority of the recorded earthquakes are related with recognized D nodes. The exception is the 10/02/1961 earthquake associated with node 199 which is identified as N. The magnitude (5.2) of this 1961 earthquake in Zamora, NW Spain, could be overestimated because the maximum intensity, VI, correlates better with magnitudes below 5.0 using both standard and Iberian-specific intensity-magnitude empirical relationships (e.g. Karnik 1969; Ambraseys 1985; Samardjieva et al. 1999; Lopez-Casado et al. 2000). If the magnitude is not overestimated, then node 199 is misclassified in this work.

Six events are not related with the nodes delineated in the Iberian Peninsula when a node is defined as a circle with 25 km of radius. The distance from them to the nearest nodes exceeds 25 km by 2–7 km (see Table 1). In Fig. 6 these events are identified with their date of occurrence. One of them is the ancient 1373 earthquake in the Pyrenees. A recent work on the earthquakes in Catalonia, which occurred in the XIV and XV centuries (Olivera et al. 2006) relocates the epicenter of the 1373 event 16 km NW with respect to the location given in the Spanish Catalogue (Table 1), with an estimated uncertainty of 20–50 km. This epicenter can be associated with nodes 68 or 70, both of which are recognized as D. Three earthquakes, the 04/05/1909, 12/01/1910, and 11/09/1910, are located in the Lower Tagus basin between nodes 295 and 312, both of which are classified as D (Fig. 6). These three events are close in time and space to the aftershock series of the 23/04/1909 Benavente earthquake, with I0 = IX, that lasted for more than 1 year. Each aftershock appears as a double entry in the Portuguese catalogue (Martins and Mendes Víctor 2001), with the same origin time and different location, one of them corresponding to the Benavente earthquake; thus, suggesting that they could belong to its aftershock series, and thus correspond to node 304.

No D nodes were indentified in the vicinity of the 24/11/1910 earthquake located in Galicia (Fig. 6). This event produced the largest recorded intensity (VII) at the Galician coast and lower intensities inland. The isoseismal distribution (Mezcua 1982) shows clear indications of an offshore location, out of the delineated MZ.

The recognized D nodes are mainly concentrated at the periphery of the Iberian Peninsula. Practically all nodes in the Betics and Pyrenees are identified as D including those where there is no record of target earthquakes. The nodes located in the interior of the peninsula, apart from several D nodes in the Iberian Chain, have been recognized as non-capable of M ≥ 5.0 (Fig. 6). In topographically low areas, potential nodes have been identified in the Portuguese, Lower Tagus, and Ebro basins, as well as inland of the Gulf of Valencia and the Cape St. Vicent. Few potential nodes were also recognized in the Cantabrian Mountains and on the continental slope (Fig. 6).

Some nodes recognized as potential for M ≥ 5.0 are located in areas where such earthquake potential was not considered in the source models from most of the available regional probabilistic hazard studies (e.g. Martin 1984; Jiménez et al. 1999, 2001; Peláez and López-Casado 2002). Specifically, the pattern recognition results show a significant potential for M ≥ 5.0 (or I0 ≥ VII) in the Cantabrian Mountains, while most of the regional probabilistic hazard studies assume maximum values close to M = 5.0 (e.g. Mmax=5.3 in Jiménez et al. 1999 or Mmax = 4.9 ± 0.6 in Peláez and López-Casado 2002), with low activity rates. The additional information provided by the node recognition technique could help to redesign the seismogenic sources, and it shows the relevance of combining elements of both deterministic and probabilistic methods to increase the information content in seismic hazard evaluation. For a more detailed discussion on deterministic and probabilistic approaches for seismic hazard assessment (see Bolt et al. 2003; Panza et al. 2003; Jiménez et al. 2009).

As in other Mediterranean regions (Gorshkov et al. 2002, 2003, 2004), most of the D nodes are located on the higher—first and second rank—lineaments, i.e. on the boundaries of the larger blocks (see Fig. 6).

The characteristic traits reported in Tables 6, 8, 10 and 12 provide additional information to deduce criteria for discriminating D and N nodes.

The characteristic traits of the nodes in the mountain environments (Table 6) include alliances of the ten parameters used for the recognition. The characteristic traits of D nodes presented in Table 6 indicate the contrasting neotectonic movements, evidenced by large values of ∆H and ∆H/L. In addition, the traits of D nodes point to an intense fracturing of the crust, as suggested by the large values of ∆B (∆B > 45 mGal) and the small values of Dn (Dn ≤ 29 km). In comparison, characteristic traits of N nodes of group I indicate less contrasting movements and smaller fracturing of the crust.

For the nodes of group II, located in the transition zone between lowlands and mountains, the decision rule (Table 8) includes assemblages of all 13 parameters for the recognition (Table 4). The large number of parameters composing the decision rule indicates the heterogeneity of the training sets D0 and N0 for this group. In spite the a priori grouping of nodes by their similarity in morphological and structural position, we could not completely avoid the heterogeneity of the learning set. The characteristic traits (Table 8) indicate that N nodes of group II are not situated in the lowest areas (Hmin > 400 m and Q ≤ 3%) exhibiting low contrast vertical movements (∆H/L ≤ 38.6). According to their characteristic traits (Table 10), D nodes in the lowest areas (group III) are not characterized by large values of minimum topographic altitudes (Hmin ≤ 480 m) in combination with large values of Q (Q > 20%) and not by small values of ΔHH > 380 m). The combination of D traits suggests the persistent subsidence throughout time of the areas accommodating seismogenic nodes. Characteristic traits defined for nodes of group III are of a special importance because it is the first time that they have been identified for seismogenic nodes located in topographically flat basins showing moderate seismicity. The characteristic features of Table 10 can be tested to identify seismogenic nodes in other basins like the Po and Pannonian ones where moderate earthquakes are rare and their number is insufficient to perform the pattern recognition in corpore.

The characteristic traits for group IV (Table 12) indicate that the offset of the continental slope should not be very large in amplitude (∆H ≤ 2,020 or ≤1,200 m) and the same is valid for the gravity gradient (∆B ≤ 60 mGal). On the continental slope most D nodes are located along the Betics offshore and in the Mediterranean section of the slope near the Catalonian coast. Practically all nodes sitting on the slope near the Betics have been classified D.

The structural and morphological settings of this area are very similar to that in the W-Caucasus where a steep mountain flank directly turns into a continental slope. All nodes delineated with MZ in West Caucasus have been recognized D for M ≥ 5.5 (Gvishiani et al. 1988; Gorshkov et al. 2003). The subsidence is characteristic of some D nodes recognized in the Caucasus (Gvishiani et al. 1988) and in the Dinarides (Gorshkov et al. 2004). An intense fracturing is a characteristic feature of seismogenic nodes in other regions of study as well (Gorshkov et al. 2002, 2003, 2004; Gvishiani et al. 1988). From the modelling of the block structure dynamics and seismicity, the increased fragmentation of the media has been established by Keilis-Borok et al. (1997) as a necessary precondition for the occurrence of the strongest earthquakes.

This study demonstrates that the pattern recognition approach to identify earthquake-prone areas is applicable to a region, like the Iberian Peninsula, that is characterized by a complex and heterogeneous tectonic structure and topography. The recognition performed pinpoints a number of D nodes where moderate events have not been recorded up to now, specifically, in the Cantabrian Mountains, westernmost Betics, Portuguese basin, and in the area around Valencia. This new result can provide additional information to improve seismogenic source models. Some of the recognized D nodes give more knowledge about seismic risk affecting special sites like nuclear and water power plants and large metropolitan areas.

Acknowledgments

This work has been partially supported by project CGL2007-62454/BTE of the Spanish Ministerio de Ciencia e Innovación.

Supplementary material

12210_2010_75_MOESM1_ESM.doc (54 kb)
Supplementary figure S1 (DOC 55 kb)

Copyright information

© Springer-Verlag 2010