Historical earthquake parameters by geological and seismic site analysis: the 1908 Cerbón earthquake (Spain)

Previous to the appearance of recording instruments for seismic events (dated around 1920), it was only possible to undertake incomplete enumeration of seismic events, along with their surrounding circumstances in the particular area of study. Hence, in the nineteenth century, Palacios (1890), Sánchez Lozano et al. (1894) and some other authors (Egozcue y Cía 1875; Naranjo 1908) enumerate the sites where earthquakes occurred in the provinces of Soria and La Rioja (Spain). The identified earthquakes were: 1817 Préjano and Arnedo Martín-Escorza (2005); 1819, 1845, 1893 Munilla; 1831 Herce; 1846, 1848, 1854, 1857, 1860 Navajún; 1847 Valdelprado; 1875 Arnedillo; and 1884, 1885 Yanguas earthquakes (Figs. 1, 2).

The 1908 Cerbón earthquake and some other recent events at the same site: on the night between 15 and 16 January 1908, an earthquake took place in Cerbón, causing evident ground deformations and a big landslide, which is analysed hereinafter in this paper. As previously mentioned, this earthquake is not included in the IGN seismic catalogue. Some other smaller events occurred previously in the same area in 1905 and 1907 (Naranjo 1908). Some seismological stations already existed in Spain, but it seems they were not accurate enough to perceive and record this particular event, and in fact, it is not listed in the seismic data set (Fig. 3).

This earthquake was clearly felt throughout the whole region until at least Trébajo (10 km South from Cerbón, Fig. 4), as reported in Naranjo (1908). Neither surveys nor any kind of deep analysis were carried out, and thus, the intensity reached in the surrounding towns is unknown. The investigations done in the scope of the present research among elderly inhabitants of the surrounding villages have necessarily been very limited, as there are no survivors (the earthquake occurred more than a century ago). Moreover, the only source of information is through oral tradition, since no written documents on this event, except the aforementioned Naranjo’s report, are available.

Memories of having heard news of their ancestors during this earthquake still exist in Cerbón, as well as in Fuentes de Magaña, where furniture moved and many houses cracked while tiles fell down from the roofs. According to these descriptions, this earthquake could be classified in the range of intensities VI–VII. El Avisador Numantino (the local newspaper) described how people panicked. The concern of the population forced the Province Civil Governor to request a recognized specialist in seismicity to write a report, trying to find out the origin of both earthquake and landslide (Fig. 4), as well as the possible relationship between them, and if these events could potentially occur again. The earthquake was noticed by means of a powerful explosion, followed by some smaller other ones, accompanied by vibrations and underground noises coming from the SE, i.e from the southern area of the Miraflores gorge. Three years before, in 1905, a period of intense quakes took place, causing furniture displacements in the houses. In 1907, several ground shakings and underground noises were perceived. In all three cases, Naranjo describes how the population heard strange noises and undulatory movements at the soil surface. Concerning the environmental effects (Fig. 4), the author highlighted the following:The most important effect, however, was the landslide and ground dislocation that took place 600m south of Cerbón, which will be described in detail hereinafter. A local orchard access road was blocked, and a close cote was partially destroyed. Naranjo justified this particular effect as due to the SE–NW orientation of the seismic waves’ main direction.

Apart from the 1961 Aguilar del Río Alhama earthquake, several smaller quakes have been reported in this region. Those registered by the IGN are posterior to 1962, because the Logroño seismic observatory was installed that year. Nowadays, when the Spanish seismic network has been significantly improved, plenty of quakes at this site have been recorded. Thus, it is possible that, in the period of time ranging from 1908 to 1962, other small magnitude, not-registered events occurred, which were, however, noticed by the population of Magaña village every 2 or 3 years.

The most recent recorded quakes in Soria are summarised in Table 1. It is worth noting that the epicentres of some of the quakes were located at Fuentes de Magaña (less than 1 km away from Cerbón), and, in general, close to this region, demonstrating a significant seismic activity at this site. This fact is also assessed by the available information on earthquakes in Logroño, which not included in this paper, but were also checked.

A magnitude 5.1 earthquake occurred in 18 February 1929. Its epicentre was allocated in Turruncún, close to Muro de Aguas (La Rioja, Spain) (Rey Pastor 1931). It caused destruction in the surrounding area, but the damage was particularly serious in Turruncún, where the majority of dwellings collapsed, injuring several people, though without any casualties.

The 1961 Aguilar del Río Alhama earthquake had its epicentre in the study area and will be described in greater detail. It occurred on 4 September 1961 at 23:33 h and its epicentre was officially located in the municipality of Aguilar del Rio Alhama. Its magnitude (Mw) was 4.6 according to IGN or 5.8 according to the GSHAP catalogue (Global Seismic Hazard Assessment Program).

The area is part of the northern sector of the Aragonese arm of the Iberian Range. In general terms, the stratigraphy comprises a Hercynian basement and a Permo-Triassic tegument, both of which outcrop at the heart of the anticlinorium near Moncayo. Overlying these, there are various Jurassic marine carbonate formations and continental “Purbeck-Weald” facies (Figs. 1, 2, 5, 6). Deposits in these zones are more than 3000 m thick, though elsewhere in the so-called Cameros Basin, the total thickness of the Wealden facies can exceed 9000 m.

The Oncala Group outcrops in Cerbón Tischer et al. (1966) (Fig. 2). It comprises some 2000 m of clastic and carbonate sediments. In the eastern part of the Cameros Basin, this group is differentiated into a number of formations/alloformations, which form part of the sedimentary fill sequence III of the Basin (Mas et al. 2004, 2011). Given the lithological characteristics of the sediments (Clemente 2010; Quijada et al. 2013), in this sub-basin, we can differentiate three large formations: Fm. Huérteles, Fm. Aguilar and Fm. Valdeprado. The first of these is characterized by a predominance of clastic sediments, and the other two by a preponderance of carbonate sediments. They are considered to date from the Middle Berriasian (Mas et al. 2004; Clemente 2010).

Given that the stratigraphic succession is apparently quite similar throughout, a 30-m sample of the column from the study area was logged in detail (Fig. 6) in order to get a precise understanding of the materials involved and the incompetent levels, to be able to explain the geomechanical characteristics of the landslide. As can be seen from Fig. 6, the stratigraphy consists of alternating limolitic and carbonate horizons. The landslide affects a packet 20–30 m thick, which contains two horizons that have very different mechanical behaviours from each other. The upper one, around 15 m thick, is competent, while the clayey layer below, 5–6 m thick, is not.

The base that slipped consists of whitish and grey sandy limolites, in beds 10–20 cm thick that are grouped in horizons of 1–1.5 m. This is followed by a carbonate horizon of quite crumbly pale limestones and calcareous clays (with sulphates), which are weathered at the surface, giving white or yellowish soil. This soft and slippery layer outcrops along the eastern edge of the landslide in the Arroyo de Miraflores and in its headwaters, as can be seen in the geological section in Fig. 7. The clayey base measures 4.5–6 m. The remainder of the series comprises limonites and sanstones.

The study zone lies on the southern edge of the Cameros Basin, on its border with the rest of the Iberian Range, and close to the eastern border of the basin with the Tertiary Ebro Basin (Fig. 1). The most significant structure, formed in the Tertiary compressive stage, is the Cameros nappe, which runs NE–SE; it continues NE in the nappe of the Sierra de la Demanda, and SE in the Sierra del Moncayo (Gil Imaz 2001).

The structure of the epicentre zone is apparently simple, corresponding to the eastern flank of the Valdeprado syncline, where the stratigraphic series is composed of well-stratified fine beds trending NW–SE and inclined towards the NE at a constant angle with a 20°–40° dip. Some of the stratigraphy traces, also trending NW-SE, are illustrated in Fig. 8.

Aerial pictures highlight that there are few fractures of any significance. These are also represented in Fig. 8, where they can be seen to have an almost N–S direction; the intense faulting in the Miraflores ravine stands out.

The seismic reflection surveys done for oil prospecting in this region show that there are some notable faults, such as the one that runs N–S to the east of Magaña and Fuentes de Magaña. Its manifestation at the ground surface may correspond to those indicated in Fig. 8.

The absence of significant Tertiary and Quaternary deposits in the epicentral zone—which might have been deformed by tectonic activity—makes it difficult to identify recent tectonics. As a result, the hydrographic catchment of the river Alhama has undergone continuous and intense erosion since the Miocene, and so little evidence remains of Quaternary sediments, except the Holocene alluvial of the drainage network. However, if we consider the broader study zone, there is a Tertiary-Quaternary microcatchment in the river Añamaza, around 15 km SE of Cerbón, which contains clear evidence of recent tectonics. It lies on the southern edge of the Cameros Basin, and several authors consider that the Moncayo nappe must pass through it.

Recent deformations are considered as those that have occurred after the Middle Miocene, the period when the main overthrusts on the northern edge of Cameros-Demanda massif ceased to act, after which they have behaved as the passive edge over which the Neogene materials (post Middle-Miocene) of the Duero and Ebro Basins lap.

In the upper valley of the river Alhama and in a 10 km radius around Cerbón, small to medium magnitude seismic events occur with certain frequency. Permanent deformations of the rocky substrate have been produced, as well as fissuring and collapse of some dwellings, at least in the following locations: Magaña, Aguilar del Río Alhama, Cerbón, Fuentes de Magaña, Valdeprado and Castillejo de San Pedro. Since it has been happening even for events of magnitude ranging from 2 to 3 for over more than one century, it indicates the seismic risk that certainly exists in the area (and which would need to be taken into account if new dams are to be built, and so on). It is worth noting that most of the towns were abandoned decades ago, and these buildings, with no maintenance, are very vulnerable.

On numerous occasions, alteration of the water regime affects the hydrogen sulphide waters. This phenomenon is not unusual, since the presence of hydrogen sulphide springs and wells is quite frequent in the Wealden facies in this area, due to the oxidation of pyrites or the presence of gypsum with organic matter (Sanz and Yélamos 1994). There are hydrogen sulphide springs near Valdeprado, Ágreda, Cigudosa, Las Fuesas, Castillejo de San Pedro,Magaña, Navajún, Dévanos and Suellacabras.

The explanation of hydrogeochemical anomalies, such as the appearance of new hydrogen sulphide upwellings in Trébago following the Cerbón earthquake, seems to correspond to an increase in the fluid pressure in the compression zone of the fault that generated the earthquake. During this event, semithermal, hydrogen sulphide waters were released from depth. As the new upwellings emerged at a temperature of 28 °C (bearing in mind that the springs in this area usually flow at 12 or 13 °C), if we assume a normal geothermal gradient, this would mean that the water came from a depth of around 500 m. Undoubtedly, these springs did not exist previously, and it would seem that they did not persist for very long time. As described by Naranjo, in Trébago no perennial cold or thermal hydrogen sulphide springs have ever existed, except for a number of seeps in the Dehesa de Magaña, close to the municipality of Trébago.

The majority of seismic events recorded in the Iberian Peninsula have had magnitudes of between 4.8 and 5.0 Mw and peak intensities of VI–VII MSK. In most of the cases, the macroseismic areas cover 80–100 km², but the relevant geological effects hardly affect more than 10 km². Likewise, very often the geological effects are related to small local displacements (of unstable areas), block displacement and rock falls (Bullas 2002; La Paca 2005); cracks of millimetric width and decametric length on Quaternary terrains that are poorly compacted, and, most importantly, the previously mentioned hydrological changes in springs and wells.

The collation of data about this historical earthquake, with the revision and interpretation of its side effects based on historical documents, might be enough to assign and use the Environmental Seismic Intensity scale, ESI-2007), aiming to establish the local intensity around the earthquake's epicentre.

According to this scale, this earthquake showed characteristics of intensity IX, including the presence of a rocky landslide and the fissure zone described above, with in excess of 106 m³ of remobilized material. The dimensions of the largest blocks mobilized exceeded \(50 \times 50 \times 25\) m, while the displacement was less than 10 m, since once they had fallen to the base of the valley they could not go any further.

In terms of the spatial distribution of the instabilities with respect to the epicentre and according to the magnitude of the event, we can refer to the limiting curves proposed by Keefer (1984), or those constructed by Delgado et al. (2011) for Spanish earthquakes. This type of coherent seismo-induced instability does not correspond to low-magnitude seismic events. Rather, allowing that it lays at the very epicentre (or very close to it), it would require an earthquake of magnitude between 4 and almost 6. The greater part of the coherent-type seismo-induced instability is associated with events of at least VI–VII.

There are characteristics of intensity VIII if we consider the appearance of new springs with different physicochemical properties and abnormally high temperatures. On the other hand, the presence of dust in the epicentral area of the Miraflores ravine would develop from the intensity VIII event or greater.

Nevertheless, the destruction of fences together with the landslide (although unaffected by this) and the damage consisting of small cracks and tiles falling from dwellings in Fuentes de Magaña would indicate a Mercalli intensity of no more than VIII.

For the reasons listed below, it is difficult to identify the seismogenetic sources in the study area:However, we can make a number of orientative general comments: in the first place, on a wide regional scale, the distribution of historical and recent seismic events in Sierra de Cameros shows a clear dispersion within a higher density concentrated near the NE borders, linked according to zones with tectonic structures, where the amount of deformation during the Cenozoic component was greater. The type of epicentral distribution and the shallowness of the hypocentres point to the main seismogenetic source in this zone being the Moncayo and Cameros nappe, as well as the faults associated with them. According to the distribution of the seismic events that we are studying, these must be related to a fault in the Alhama valley in the Magaña sector. The location of the aftershocks around the study zone of the Aguilar del Río Alhama earthquake, and the intensity that was experienced in Magaña, leads us to believe that the Cerbón landslide falls fully within the seismic activity of the zone, and we cannot reject the possibility that the landslip was caused by the same seismogenetic source. Of the faults recognized at a local level, the Miraflores ravine stands out as a zone that is intensely tectonized; it lies 2 km from the landslide and from a N–S trending fault that passes to the east of Magaña and Fuentes de Magaña, but no further data are available. In the surrounding area, where the recent sediment record is preserved 10–15 km to the south and on the southern edge of the Cameros Basin, there is evidence of Miocene-Quaternary tectonic activity. This supports the hypothesis of recent activity in the zone of the Moncayo nappe. In 2004, there were several small earthquakes measuring less than 3 on the Richter scale, which were felt in Ágreda, San Felices, Matalebreras, Muro de Ágreda, and to a lesser extent in other nearby villages, all lying in the so-called Tierra de Ágreda. As we can see in Fig. 1, three epicentres of the microseismic events were located in the Moncayo fault, mentioned above, which runs west from Vozmediano to the Sierra del Madero.

The deformations produced in the ground and in the epicentral area of this quake are all very characteristic and suggestive. Accordingly, considering the local geology and topography, the following mechanism of generation of the phenomenon can be proposed: the epicentre of the earthquake must have been very close and to the SE (perhaps associated to the tectonized zone 2 km away, in the Miraflores ravine). Accompanied by a cloud of dust, the waves propagated from SE to NO, according to contemporary oral testimonies about the direction the noise came from, the way that walls were broken in the Majada del Colorado, and the orthogonal orientation of the pressure and traction structures of the rocky hill in Cerbón. All the rock mass is divided in thin layers with the same direction and dip throughout the epicentral area, which gave rise to a strongly anisotropic environment, wherein the orientation of the stratification, also NW–SE, the supposed presence of a reverse fault with the same direction, which manifested discontinuously along the stream bed, could have contributed to the preferential propagation in this direction. This would explain why the earthquake had more impact in Fuentes de Magaña than in Cerbón, assuming that the epicentre had been very close to the rocky hill in Cerbón. Fuentes de Magaña lies downdip, while Cerbón lies along the strata and is protected from the northern slope of the Miraflores ravine, which is higher than the southern one and which would oppose the propagation of seismic waves. The local geological and topographical conditions around the rocky hill in Cerbón were very significant. In the first place, the route of the Miraflores ravine coincides with the direction of the regional stratification; it fits specifically and closely to particular horizons. The strata dip against the slope on its left bank and they are stable. On the right bank, they dip with the slope. In no instance have they been exposed by the incisive vertical erosion of the river at the foot of the slopes, so that under static conditions, they could not have been moved except with the intervention of a seismic event. There is no other slope so extensive elsewhere in this ravine or in others in the zone that resembles a structural rocky hill like that of Cerbón. This feature means that the effects of the seismic waves could be felt in such a particular way, provoking its destabilization due to the contribution of the clayey detachment layer. There are other slopes further downstream, where it is thought that the earthquake may have also created momentary situations of quasi lack of equilibrium and thrusts, which have left their mark in the form of vertical stratifications at various points in the bed of the ravine on its right bank, but which did not cause a generalized mobilization.

The shallow movement in the unstable rock mass give an explanation to the systematic folding into several ridges and fractures appearing in the rock pavement, in the landslide. The amount of shortening of the anticline folds is approximately compensated by the traction displacement.

As we can see, this movement can be classified as a rock landslide but, thanks to the modest displacement involved, the stratigraphy of the rock was not disrupted as the mass movement progressed. Within the mass, the structure of the seismic waves that generated it were preserved as evidence. Nevertheless, a large amount of energy is required to set in motion a coherent mass of rock, such as this one. The greater part of coherent seismo-induced instability (the coherent typology is less frequent), are associated with events that have an intensity of at least VI–VII. The majority of seismo-induced instabilities in Spain Alfaro et al. (2012) are of disrupted typology (Keefer 1984). These instabilities usually correspond to rockfalls and soilslips, whose structure becomes disorganized as the mass movement progresses. This is why small excitations of M-2.6 can trigger such movements on slopes in strict equilibrium and they are very frequent for low- moderate magnitude events (Mw less than 5.5).