Paleoseismology plays a critical role for understanding the behavior of major seismogenic faults (i.e., How constant in time are the rupture length and slip distribution? How stable are segment boundaries? How frequently earthquakes repeat on the same fault?). As pointed out in several seismic regions of the world, the instrumental and historical records of seismicity are far too short to be representative of several seismic cycles for individual faults; in most cases they cover only part of one cycle, that means that, on some faults they may not even contain evidence of one earthquake. By extending back in time several thousand years, paleoseismologic investigations can fill this gap and provide the record of several seismic cycles. Unlike the historical record, paleoseismology establish a direct link between the earthquake and its causative fault. However, there are intrinsic uncertainties that should be considered. In general, age of the paleoearthquakes can be constrained only within an interval of a few to several centuries; thus, it is arduous to establish if very long faults ruptured during one single event or as a cascade of events which occurred within that interval of time. Merging information from different sources may solve this problem and paleoseismology based modeling of seismic behavior of large surface rupturing faults remains crucial and may have a substantial impact on traditional seismic hazard assessment. Two examples are discussed here that show this type of impact: (1) the study and characterization of the San Andreas fault in central California; and (2) the results derived from several paleoseismologic studies in Italy. 1) Recent paleoseismologic studies along the 1906 San Francisco earthquake rupture are used to compare the 1906 slip distribution with slip rates that integrate repeated earthquake slip over several thousand years. A slip rate gradient along this part of the San Andreas is found and the "missing" slip is known to be accommodated by other nearby structures. This part of the San Andreas is interpreted as a large displacement rupture that repeats itself in both length and slip distribution and, on a long term basis, releases most of the seismic moment on the fault with no need for several very frequent moderate events. 2) Frequency of large earthquakes and rates of extension across the Apennines in central and southern Italy were traditionally derived from the historical record. This suggested an average recurrence time of a few centuries, implying an extension rate across the Apennines of several mm/yr. Paleoseismologic investigations over the past 10 year suggested a rather different behavior of the seismogenic faults with consistent average recurrence intervals of one to three millennia and extension rates not exceeding 0.6 mm/yr.
Paleoseismology is one of the most powerful tools to study earthquake faults in active zones. Geological investigations of past earthquakes need, however, new conceptual approaches of seismogenic zones and the integration of new techniques. Various techniques including trenching, geomorphic levelling and geophysical prospecting were recently applied to fault scarp studies in different tectonic domains. Field investigations show that fault exposures present similar structural features and that except for the rate of slip, active faults in the Mediterranean active zones or along the Rhine graben (intraplate Europe) do not present much differences; the rupture geometry and record of successive displacements, structural features, and earthquake-induced phenomena are comparable. We observe also that field investigation techniques must be adapted to each tectonic situation and related geomophological signature. Implications for the seismic hazard are tremendous and although the occurrence of significant earthquakes (moderate or large) in time and space is complex, paleoseismological investigations enable us to constrain the faulting behaviour. A comparison between different studies of active faults reveals that beside segmentation, the fragmentation model can be applied to some seismogenic zones. The recently developed fault fragment model (Meghraoui and Camelbeeck, 1996) address the problem of the geologic record of moderate earthquakes with respect to the large seismic events.
The day after the Mw = 5.6 earthquake that stroke the Southern Apennines on September 9, 1998, we observed coseismic surface slip on a limestone normal fault plane at Galdo di Lauria. The reactivated slickensides trend N160° to N135° and dip 50 to 60° toward SW. They show open fractures at the bedrock-bedrock contact (Cretaceous limestone in the footwall, Miocene flysch in the hangingwall) over a length of ca. 200 m. A smaller section, ca. 5 m long, displays a free-face, up to 1 cm high, marked by a gray strip due to the mud still attached to the slickenside. The observed faulting at Galdo di Lauria fits very well the regional active tectonic setting. The earthquake occurs within the belt of capable normal faults affecting this sector of the Apennines since the early Quaternary, governed by four major, almost parallel fault segments (Vallo di Diano and Val d'Agri to the N and NE, Mercure and Pollino to the SE). The coseismic offset, location and kinematics of the limestone plane at Galdo di Lauria strongly suggest that the ruptured fault is at the NW-termination of the NW-trending Mercure fault segment, and has the same strike and style of faulting. Historical seismicity in the Lauria area, preliminary macroseismic observations and aftershocks distribution and the Harvard Quick CMT focal mechanism support this interpretation. In this framework, the coseismic rejuvenation of the Galdo di Lauria limestone scarp might help to calibrate paleoearthquake magnitudes inferred from trench studies along the Pollino fault. Seismic catalogues shows only subdue seismicity for the Mercure and Pollino segments. Therefore, for these faults seismic hazard assessment must rely heavily on paleoseismological data. To check if 1 cm of surface displacement is a reasonable value for the Mw 5.6 Lauria earthquake, we revise similar features for normal fault events in the Apennines. Maximum normal displacement on limestone slickensides during the 13.1.1915, Mw ca. 7.0 Fucino, 23.11.1980, Mw 6.9 Irpinia, and 26.9.1997, Mw 6.0 Colfiorito earthquakes, are 80, 50 and 8 cm, respectively. We complement these data with local Quaternary tectonic evolution, Holocene bedrock scarp morphology, historical seismicity and paleoseismic investigations at appropriate sites along the Celano - Gioia, Parasano, San Gregorio Magno and Pollino Faults. We show that characteristics of surface displacement on bedrock fault planes along capable normal faults are useful for evaluating the size of the causative earthquake, in this part of the Apennines and elsewhere.
Following the results of structural and morphotectonic researches previously carried out in Thessaly, which enabled the recognition of the major active structures affecting the area, we investigated two faults with a palaeoseismological approach. Both structures are 10-12 km long, dip-slip normal faults bordering the Late Pleistocene-Present Tyrnavos Basin. The Rodia Fault is ESE-WNW trending and south-dipping, while the Tyrnavos Fault is E-W trending and antithetic. According to a dedicated morphotectonic study and a geophysical investigation (georadar and seismic reflection) carried out across the more prominent scarps, sites for palaeoseismological trenching were selected.
One of the trenches crossing the Rodia Fault clearly shows a palaeoscarp, displaced alluvial deposits and colluvial wedges. From the geometry of the observed sedimentary bodies and associated structures, several seismic events can be recognised with a mean dislocation per event of 10-30 cm. Several samples have been collected for thermoluminescence analyses. Though dating should be further confirmed, the first results give ages ranging from 16.7 to 3.9 ka. Accordingly, at least two events occurred during the last 5000 years and possibly more than one between 7.5 and 16.7 ka. If we accept the available dating, the estimated slip-rate is about 0.1 mm/a and the return period is of few thousand years.
Along the Tyrnavos Fault, we excavated six parallel trenches; four of which within the alluvial deposits and two cutting the bedrock, here represented by Pliocene limestones. The two latter trenches were fully successful because they allowed the direct observation of the fault zone (viz. bedrock-loose deposits contact) as well as of several alluvial wedges accumulated along the bedrock scarp. According to microstratigraphic observations, it is possible to infer several seismic events, associated to vertical displacements of 20-50 cm. Also in this case, we obtained ages from thermoluminescence (TL) dating. Preliminary results range from 20.1 to 7.4 ka and the consequently estimated slip-rate is between 0.1 and 0.2 mm/a. This values are in agreement with those obtained from geological and morphological investigations (0.14-0.4 mm/a). Also the trenches excavated within alluvial deposits show some possible evidence of surficial perturbation occurred during past events. In particular, one of the trenches shows the local concentration of carbonate nodules, probably due to cracking and consequent enhanced permeability, in correspondence with three distinct zones of displacement. Another trench shows a 30-40 cm wide graben-like structure filled by coarse pebbles, thus suggesting the occurrence of past surficial extensional phenomena.
The Gulf of Corinth is a 130 km long seismic belt within the extensional Aegean domain. The seismic belt at present is undergoing a N-S extension. Most of this extension is accommodated by earthquakes hosted in WNW- and ENE- trending 10-40 km long normal faults. Records of the historical seismicity in the Gulf of Corinth show that two large historical earthquakes have occurred in one of these faults the Eliki fault. In this work we present paleoseismological data on this fault coming from three single slot excavations operated near the well known 1861 surface rupture (Schmidt 1879; Koukouvelas and Doutsos 1996). Logging in the trenches show two colluvial wedges and associated soils on the Eliki fault scarp. The older colluvial wedge rest on the buried fault free face and are affected by a series of small offsets on small scale normal faults. These small scale faults are secondary features to the main normal fault and increase the general backward tilting of the colluvial wedge. A soil horizon showing slumping overlies the older wedge. This wedge is affected by a number of newer coseismic ruptures as is indicated by a northwards inclined mixed layer including soil slumping, fragmented potteries and juckles. The mixed layer is buried by the newer wedge that includes coarse grained clastics and is characterized by a complex depositional history. Three small scale colluvial wedges construct this wedge with thickness ranging from 20 to 28 cm. The faulting history is interpreted using the colluvial wedge model (Wallace 1977). Three pre-1861 events are recognized which corresponds to the aforementioned wedges. The mean coseismic offset per event is 23 cm. The soil horizon associated with the newer wedge is recognized as the A-horizon. Fault's latest offset is denoted by the 0.94 m soil horizon throw that correspond to the coseismic displacement during the 1861 earthquake.
Koukouvelas I & Doutsos T, J. Stuct. Geol, 18, 1381-1388, (1996).
Schimdt J, Leipzig, 68-83, (1879).
Wallace RE, Geol. Soc. Am. Bul, 88, 1267-1281, (1977).
In the framework of the CEE-project PALEOSIS, investigations on active faulting and palaeoearthquakes have been undertaken in the Roer Graben. The Roer Graben, which crosses three countries (Belgium, The Netherlands and Germany) is bounded by two NNW-SSE trending Quaternary normal fault systems. The eastern boundary is defined by the Peel Boundary fault (Rurrand fault in Germany) and the western boundary is defined by the Feldbiss fault zone, on which evidence of Holocene and Late Pleistocene surface faulting have been already presented (Camelbeeck and Meghraoui, 1996, 1998).
The Holocene and Late Pleistocene tectonic activity of the Peel and Feldbiss faults is presented as a result of detailed paleoseismological investigations combining detailed geomorphic, geophysical and trenches analysis. This study confirms that for fault fragments of 10 km length, with a deformation rate of 0.1 mm/year, the return period for large earthquakes (M > 6.0) is of the order of 10,000 years.
Camelbeeck T & Meghraoui M, Eos, Transactions, American Geophysical Union, 77, 405-409, (1996).
Camelbeeck T & Meghraoui M, Geophys. J. Int., 132, 347-362, (1998).
Activity of seismicity is a common fact recorded by synsedimentary deformations during early deglaciation on large Palaeozoic shields as Scandinavia or Canada. Iceland, and especially the North Eastern Rift Zone is characterised by a thin, warm crust, active volcanism and active sismicity related to rifting. During glaciation, fissures swarm area of activity is wider because of the ice sheet weight (Bourgeois et al., 1998). During OIS6 (Saalian Glaciation), the thickness of the ice sheet was exceptionally thick, despite a warm base in the NRZ resulting in efficient subglacial erosion. OIS 6 is the main phase of landscape shaping in this region. Opposite, the last Glaciation (OIS4-2) has in this sector a cold base and is only preserved inland in form of a discrete ablation till. Eemian lacustrine, fluviatile and aeolian deposits are extensively preserved defined as the Sydra Formation (OIS5 s.l.) and allowed a fair record of paleo-seismicity in form of various co-seismic synsedimentary deformations (excavations and outcrops). Paleo-seismicity zone is for OIS 5 similar to the present-day published area. Because of its latitudinal position and the sea surface temperatures (SST), deglaciation seems to proceed rather fast as also the glacio-isostatic rebound (coastal record). A comparison between OIS 6/5 and Late Glacial-early Holocene deglaciations allowed evidencing the same pattern. Rather continuous, seismicaly induced, surficial micro-load casting activity is evident during ice sheet retreat period though sometimes enhanced by volcanic eruption (mostly effusive) (load cast level covered by ashes). After the meltout and drainage, few micro- or larger seismicaly induced load cast events, only connected to ash producing eruption occured, sometime associated with water escape figures or microfaulting. From synsedimentary and stratigraphical data, microseismicity stopped with the end of the fast rebound. It is deduced from these observations that the continuous microseismic signal could represent the glacio-isostatic rebound period, which seems too to be in Iceland also a period of enhanced fissure swarn and volcanic activities.
In recently glaciated terrains, investigations of prehistorical earthquakes face difficulties which are distinctive from those studying palaeoseismicity in non-glaciated areas. The apparent tendency of large ice sheets both to remove or obscure signs of pre-glacial crustal movements and to inhibit seismotectonic activity throughout the period of glacial loading, means that palaeoseismic records are unlikely to extend back beyond the onset of permanently ice-free conditions around 10,000 years ago. Perhaps more importantly, glacial loading and unloading is likely to significantly modulate the crustal deformation cycle within an area, with a pulse of early postglacial faulting responding to the release of tectonic strains 'stored' during glacial loading and to rapid rebound of a glacially-depressed crust. In this context, to what extent are concepts such as 'earthquake recurrence intervals', 'characteristic earthquakes' and the 'earthquake cycle' applicable in deglaciated terrains? More fundamentally, since different driving forces and stress regimes mean that the loci, magnitude and frequency of earthquakes immediately following deglaciation may contrast with that found today, does palaeoseismicity itself present a potentially misleading guide to the likely incidence of future earthquake activity?
Practical challenges are provided by problems inherent in discriminating palaeoseismic phenomena from other types of glacio-tectonic deformation. Deglaciation is a period of dynamic environmental change, with widespread slope instability and draining of ice-dammed lakes producing landforms and stratigraphies (e.g. palaeoliquefaction horizons) that elsewhere are associated with seismic faulting. Local readvances of ice can result in tectonic deformation of postglacial landforms and sediments which may be construed as palaeoseismic. Prominent postglacial fault scarps, generally evidence for large Holocene earthquakes, may also result from shallow gravitational spreading of montane flanks (sackung) or differential glacioisostatic movements, while reconstructing the tectonic dislocation of the glacial sediments and landforms is difficult in the varied topography and stratigraphy of the postglacial environment. Similarly, periglacial processes may induce surficial deformation phenomena that closely mimic palaeoseismic disturbances.
In general, earthquake geologists in former glaciated terrains are aware of the affinity between glacial and tectonic deformation, and palaeoseismic studies are commonly undertaken by or with Quaternary scientists. However, a growing recognition that even comparatively small glacial loads may promote seismicity, particularly where faults are stressed close to failure, suggests that appreciating glacial influences on active tectonism ought to extend beyond the limits of continental ice sheets in north-west Europe. In tectonically active montane environments of southern and eastern Europe, for example, where active faults often occur in areas recently affected by valley glaciation, early Holocene fault movements may be anomalous perturbations of the tectonic behaviour of the fault by transient glacial stresses. Despite the methodological issues raised in the introduction, understanding the interaction between glacial unloading, crustal deformation and seismicity is important for assessing the present-day seismogenic behaviour of active faults across much of the European domain.
Present seismicity is covered by seismology. Past seismicity must be inferred from the traces left in nature; i.e. from paleoseismics and archeoseismics. A paleosesmic cover is absolutely necessary for any serious prediction of future seismic hazard. In the last decades, remarcable achievements have been obtained in the paleoseismic recording. This refers to the identification of past event, their structural characteristics and, not least, their dating. Furthermore, it refers to presently high-seismic areas as well as low-seismic areas. In formerly glaciated regions, we now appriciate that the deglaciation period was linked to high seismic activity. In Sweden, we have been abel to identify a large number of paleoseismic events, some of which, by varve chronology, are dated with an annual (in one case even seasonal) time resolution. We will discuss the autumn 10,430 BP event in the Mälaren area (with liquefactions over a wider area than that of the Alaskan 1964 event), five events with a 20-years spacing in the Stockholm area, and the extensive 9,663 BP event in the Hudiksvall area (with heavy bedrock deformation, liquefaction, tsunami effects and extesive turbidides).
Mörner, 1985, Tectonophysics, 117, 139-153, (1985).
Mörner, 1991, Tectonophysics, 188, 407-410, (1991).
Mörner, 1993, Zeitschrift für Geomorphology, N. F. , Supp-Bd, 94, 107-117, 159-166, (1993).
Mörner, 1995, Quaternary International, 27, 95-98, (1995).
Mörner, 1996, Quaternary Science Review, 15, 939-948, (1996).
Mörner, 1998, Ann. Geophys, 16, suppl. IV, C1164, C1183, (1998).
Structures in bedrock and sediments show that the Fennoscandian Shield was subjected to a higher seismicity at the end of the last glaciation than it is today. The annual varves of the glacial clays date this seismicity up to one year near. In the Stockholm area, four paleoseismic events have so far been recorded in the varved clay chronology. The most prominent level of deformation occures at varve -1073, corresponding to 10.429 varve years BP. This event is particulary well documented in the Albysjön locality southeast of Stockholm. Here liquefied varves are underlaid by microfaulted undisturbed varves and covered by a thick layer of redeposited suspended matter. The redeposited matter is in turn covered by undisturbed varves. Other effects upon the varved clay due to these events, various forms of deformation, erosion and deposition, will also be discussed.
Tröften P.E., Neotectonics and paleoseismicity in southern Sweden with emphasis on sedimentological criteria, Ph.D. Thesis, Paleogeophysics & Geodynamics, Stockholm university, 8, 124 pp., (1997).
Tröften P.E.& Mörner N.A., Journal of Geodynamics, 24, 249-258, (1997).
Mörner N.A., Quaternary Science Reviews, 15, 939-948, (1996).
Mörner N.A., Tectonophysics, 117, 139-153, (1985).
Mörner N.A. & Tröften P.E., Z. Geomorph. N.F. Suppl.-Bd., 94, 107-117, (1993).
The Stuoragurra postglacial fault can be followed, in several discontinuous sections, for 80 km, in a NE-SW direction. Since 1983, the Stuoragurra Fault has been the subject of numerous field mapping, seismological and geophysical studies, as well as both percussion and core drilling. The fault is a reverse fault, with up to 10 metres of displacement.
This summer, two trenches were made across the Stuoragurra Fault, between Kautokeino and Masi. For the first time, the fault was directly observed in the bedrock. The hanging wall was seen to be thrust upwards over the footwall, with 7 metres vertical displacement evident from displaced glacial contacts (ablation material, including glaciofluvial sediments, overlying lodgement till). The fault did not penetrate the overlying glacial materials, but rather folded them, forming a blind thrust. Large liquefaction and other deformation structures were found in the glaciofluvial sediments in both trenches. Veins of angular and subangular pebbles from the local bedrock (Masi Quartzite) penetrate more than 10 metres laterally from the thrust plane and into the sediments in the footwall. It is thought that these veins were possibly injected during the fault activity. The overall deformation may have taken place in one or more phases, but the major deformation of the sediments has a décollement plane that continues laterally in the E/B horizon contact of the modern soil on top of the footwall. This may indicate that an initial pedogenesis had taken place before the fault activity occurred, however no macro plant fossils to support this were found in the possible buried soil.
Multidisciplinarity between tectonic, historical and archaeological researches is necessary for the correct understanding of the seismic behaviour of active faults and for evaluating the seismic hazard of an area. This work illustrates an example of the possibility of identifying and dating past events of surface faulting not only by paleoseismological trenching or by analysis of direct historical seismic descriptions, but also studying those sources derived from the "oral tradition" which survive nowadays as "legends".
Here such methodology is applied to Monte Sant'Angelo, a small town on the promontory of Gargano (Italy). It lies just close to the trace of one of the three main active faults which form the Mattinata system, the surface expression of a major crustal structure cutting the apenninic foreland. No strong earthquake is historically documented at Monte Sant'Angelo in the last 15 centuries, but witness of a strong shock is reported in a "sacred legend" traditionally dated to the 490-493 AD. The legend, one of the most important throughout the Middle Ages, refers to the descent on Earth of the mighty Archangel Michael. A Sanctuary was erected in the 5th century to preserve the place where the "footprints" of the Archangel were discovered; it remained in use until the present day and represents therefore a useful source for archaeoseismological evidence. Because of the scarce macroseismic data and the strong hagiographic orientation, the earthquake quoted in the legend has been excluded from the modern Italian seismic catalogue of strong earthquakes, and the area is therefore presently classified as having a maximum felt intensity of VII-VIII MCS.
This study confirms the occurrence of a "recent" strong earthquake, probably in the 5th-6th century. Structural-geological and quantitative morphotectonic analysis of the active fault (Piccardi, 1998) allowed the author to estimate values of the normal (0.7 ± 0.2 mm/yr) and right-lateral (1 ± 0.2 mm/yr) slip rates. As the Holocene main fault scarp shows clear evidence of cumulated coseismic slip increments, it was possible to estimate the average recurrence time (1860 ± 460 yr) for the maximum expected seismic event, with surface faulting up to 1-1.5 m. Indications from historical and archaeological data allow us to identify the last of these episodes of slip as the result of that "legendary" earthquake. Using the scale relations indicated by Sholtz (1990), such a maximum expected event would have a magnitude between 6 and 7, implying an Imax up to XI° MCS for area of the Monte Sant'Angelo.
Piccardi L, Geografia Fisica e Dinamica Quaternaria, 21, (in press), (1998).
Scholz CH, The mechanics of earthquakes and faulting, Cambridge University Press, 439, (1990).
The initial results of investigations of the record of environmental events going back at least 30000 years in lacustrine sediments are presented. Over 20 m of continuous sediments has been recovered at each of the two sites; former Lake Seewen in the northern Jura Mountains of Switzerland and Lake Bergsee on the southern rim of the Black Forest in Germany. Both sites are believed to have been shaken strongly during of the A.D. 1356 earthquake which destroyed Basle and surrounding areas. Other earlier events are suggested in the record which, on the basis of ongoing dating studies, occur within the Holocene and probably as far back in time as the Denekamp interstadial of circa 28000 years BP.
Earthquake hazard assessment in stable continental regions, such as northern Europe, has traditionally been evaluated on the basis of the instrumentally and historically recorded seismicity, which indicates relatively low hazard levels. Reliability of such estimates is a matter of debate as the long-term potential of large earthquakes usually cannot be determined based on short observational periods generally less than a few hundred years. A significant improvement to this lack of knowledge can be achieved by extending the past observations into the geological time scale. Paleoseismic investigations can provide valuable information to bridge this gap, where the potential for large earthquakes can be quantified both in magnitude and recurrence period, based on the observation of prehistoric earthquakes (paleoearthquakes) in the geological record (particularly in the last 20,000 years). However, using these records in seismic hazard analysis requires systematic treatment of uncertainties. Usually uncertainties are inherent to the interpretation of geological record, which leads at the end, to the identification of paleoearthquakes. Field observations used in the analysis may satisfy several alternative interpretations. Such interpretations become useless when alternative solutions exist but not documented in detail, and especially when the relative reliability of the favored interpretation with respect to the alternative interpretations is not known. A simple method, based on qualitative description of the uncertainties related to the paleoseismic data and especially in its interpretation, is used in the recent investigations performed on the Bree fault scarp, along the Feldbiss Fault (Roer Graben, Belgium). The cumulative uncertainties associated with the different stages of the study are expressed as the Paleoseismic Quality Factor (PQF), which can directly be used in seismic hazard analysis.
Active seismogenic fault zones in the central Apennines are important components of the Quaternary fault network of peninsular Italy which accomodate most of the brittle strain accumulated in the litosphere during the latest stages of the long deformation history of these sectors of the mediterranean. There is growing evidence that these active fault zones cannot be modelled as single-segment ruptures since they consist of structural networks that often interact with other components of a fault system and do not behave as isolated elements (Cello et al., 1997). In this study, the results of our work in this part of the Apennines which includes the mesoseismal area of the 1997 Colfiorito earthquake sequence (Cello et al., 1998), will be illustrated. The main points which are relevant for seismotectonic analysis in peninsular Italy are the assessment of a new fault segmentation model for the axial sector of the central Apennines, and the evaluation of the seismogenic potential (Tondi et al., 1997) of the main active faults in the area.
Cello G, Mazzoli S, Tondi E & Turco, Tectonophysics, 272, 43-68, (1997).
Cello G, Deiana G, Mangano P, Mazzoli S, Tondi E, Ferreli L, Maschio L, Michetti AM, Serva Land Vittori E, Journal of Earthquake Eng, 2, 1-22, (1998).
Tondi E, Cello G & Mazzoli S, Il Quaternario, 10(2), 411-416, (1997).
The central and eastern Betic Cordilleras, including their intramontane depressions, are the seismically most active zones in Spain. Available earthquake data indicate, that in the last 600 years different parts of the Granada basin were place of major earthquakes with MSK intensities VII - X. During a field campaign in 1998, we identified a wide range of geomorphological features of displacements which are diagnostic for coseismic surfaces ruptures in Late Pleistocene and Early Holocene deposits. However, fault scarps are often degraded due intense land-use in the north of Granada. In general, (1) very high angle, partly listric, NW-SE normal faults and NW-SE subvertical faults with oblique striae and displacements of several metres; and (2) NW-SE subvertical dextral strike-slip faults, with subhorizontal striae and average offsets ranging in the order < 1 m have been observed. An example for the latter group is exposed in the Cubillas area, directly to the North of Granada, which exhibited in careful calculations evidence for a paleoseismic event with MS 6-7. The preliminary age assessment of the fault studied is latest Pleistocene to early Holocene. The frequency/magnitude distribution based on seismicity data is used to estimate the potential for large earthquakes in the area. The extrapolation suggests recurrence rates for earthquakes with MS 6 in the order of 60-100 years, and, respectively, MS 7 in the order of 600-1000 years for distinct parts of the Granada Basin. Present-day stresses obtained from earthquake focal mechanisms indicate a similar (N)NW-(S)SW horizontal shortening direction as the young faults in the Cubillas area in the northern part of the Granada Basin.
In the framework of the European Program Paleosis («Evaluation of the potential for large earthquakes in regions of present day low seismic activity in Europe»), the Institut de Protection et de Sûreté Nucléaire is carrying out a regional palaeo-seismological study in the Upper Rhine Graben in order to identify major palaeo-earthquakes, to define their recurrence time in order to precise the regional seismic hazard.
The methodology developed consists in the characterization of active faults with morphostructural analysis completed by specific geophysical (resistivity, conductivity, ground penetrating radar) and geological investigations (trenches).
Several recent faulting evidences have been pointed out particularly in the Northern part of the Graben mainly located along the 100 km-long lineament Lauterbourg-Selestat.
In Achenheim, a normal fault, first described by Wernert (1936) and re-discovered in 1998 , shows a 6 m vertical offset involving a marker level dated between 220 and 170 ky (Saalian, isotopic stages 7 and 6). Moreover, a level dated 15.4 ky (Late Weichselian, isotopic stage 2) is affected by the fault and a geomorphic depression, which can be considered as a fault scarp, indicates a recent movement probably Holocene. The field observations and the microtectonic study show that the offsets have not a gravity origin but a tectonic origin. Moreover, the fault geometry and its vertical evolution in the trench suggest at least 3 steps of faulting.
In Hangenbieten, a normal fault has shifted the vosgian red sands (Upper Cromerian, 400 to 450 ky, isotopic stage 11) with a vertical offset of 16 m. This fault, observed in a trench, is probably connected to a deep fault, which appears in a reprocessed seismic profile (#85ST13), which could be the deep seismogenic source responsible of the superficial deformation.
The vertical deformation rate inferred is about 0.03-0.04 mm/y since de Middle Pleistocene. One can estimate that, for a 6.0-6.5 magnitude earthquake, with a metric elementary displacement, the recurrence period should be 25 000 years since the Middle Pleistocene. These data have to be compared to those in the Lower Rhine Graben where vertical deformation rate of 0.06 ± 0.04 mm/y for the Late Pleistocene-Holocene period were inferred by Camelbeeck and Meghraoui (1996).
Camelbeeck Th and Meghraoui M, EOS, AGU, 77, (1996).
Wernert P, Bull. Soc. Préhist. Fr, 11, (1936).
In order to answer to this question, a study is carried out by the Institut de Protection et de Sûreté Nucléaire in the epicentral area of the 1356 Basel earthquake (Io = VII-VIII, macroseismic magnitude = 6.2). The Dieboldslöchli cave, located in the greatest damaged area, 4 km south of Basel, in the Blauen anticline, consists in a large room the southern wall of which corresponds to a fault plane trending NE-SW and dipping 55°SE.
Several broken stalagmites associated to 2-4 cm shifted re-growths are observed on the fault plane. Five samples were taken on the fault plane and at the opposite side of the room on a stratification plane. Carbonate laminations taken on both sides of the breaks on each samples have been dated. The U/Th disequilibrum measurements by alpha spectrometry show recent ages but the activity is too weak to provide precise ages. 14C dating by AMS shows regrowths age between 1020 and 1400 AD. Our study suggests that it is possible to consider that the shock of a major seismic event is responsible of the break of several stalagmites followed by a changing in the hydraulic conditions in the cave leading to a displacement of the re-growths. These phenomena could be related either to the 1356 major earthquake or to the 1021 one which seems to be another large earthquake of the Basel area.
These results show that it is possible to correlate broken speleothems and major historical earthquake. However, the study of broken speleothems will be usefull to seismic hasard assessment only if this kind of observations helps to characterize unknown past earthquakes. In order to achieve this goal, on-going studies are presently carried out by IPSN on the physical phenomenon of the break of speleothem by seismic waves.
The Tjörnes Fracture Zone has been active as right-lateral transform zone since about 9 Ma and includes three major seismic lineaments. The main one is the Husavik-Flatey fault, trending N 120°E, expressed along the northern coast of the Flateyjarskagi peninsula by a large zone of crushed rocks. Collection more than 1300 fault slip data in 20 sites and determination of paleostress tensors in Flateyjarskagi allowed us to identify eight major brittle tectonic regimes arbitrarily named herein S1, S2, S3 and S4 (strike-slip in type) and N1, N2, N3 and N4, (normal in type). Frequent contradictions in the relative chronology data suggest that these tectonic states alternated in a complex manner. These eight regimes can arranged two by two (normal and strike-slip regimes with a same direction of <sigma>3). They have not the same importance in terms of numbers of sites and data. The tectonic regimes S3, S4, N3 and N4 are widespread. The main couple, S3-N3, indicates a N95°E trending extension on average. The second couple, S4-N4, indicates a N35°E trending extension on average. The two other couples, less important, S1-N1 and S2-N2 are related to N130°E and N175°E trending extensions on average, respectively. For each couple, the relationships Si-Ni involves simple permutation <sigma>1/<sigma>2. Considering the geometrical relationships between the directions of extension and the direction of the transform fault, these four couples can be gathered in two major groups. Each group contains dominating S-type and N-type regimes, consistent with right-lateral transform motion, but also minor S-type and N-type compatible with left-lateral motion. One group is constituted by S3-N3 and S2-N2 , the other one by S4-N4 and S1-N1. The dominating couple S3-N3 show an angle of 25° between the trend of extension (<sigma>3) and of the Flatey segment of the transform zone. This behaviour of the transform zone reflects moderate mechanical coupling. In contrast, the couple S4-N4 shows an angle of 85° between extension and the transform direction This reveals dilation across the transform zone, supported by the presence of thick dykes trending nearly parallel to the transform direction. This behaviour of the transform zone indicates very low mechanical coupling. The reversed regimes (S1-N1 and S2-N2) have little expression except near the transform, where large deformation occurred. The drastic reversal <sigma>1/<sigma>3 relative to the dominating stress regimes S3 and S4 and the minor ones S2 and S1, respectively, probably results from elastic rebound, fault block accomodation and magmatic injection phenomena. Variations in mechanical coupling across the Tjörnes Fracture Zone are the major source of the variations in the nature and orientation of tectonic regimes. Evidences for intermediate situations are few, suggesting that changes in coupling were abrupt rather than progressive.
Tectonic deformation is limited in France, and consequently the strain rates on sismogenic faults are low, probably less than 1 mm/yr. The recurrence time of large earthquakes would be in the order of several thousand years. Two observations can be deduced: (1) the interval of time, of about 10 centuries, covered by historical and instrumental seismicity is too short to evaluate correctly the seismic hazard in France; (2) the erosion rate (from climatic and human activity) is comparable to the tectonic activity. Thus, the present topography is mainly controlled by surface processes more than by tectonic activity or even higher. Therefore, the study of regions with these characteristics, such as the southeastern France, requires a multidisciplinary approach; in fact, the satellite images exhibit large geological structures but, because of the important surface processes (erosion and human activity), we need to combine several data sources, to have a reliable location of active fault traces. These data are: satellite images (SPOT and LANDSAT), Digital Elevation Model (DEM, from the French IGN, 50 meters per pixel), aerial photos, subsurface geophysics and field surveys. This study was performed on the Moyenne Durance Fault and in the Lower Rhone Valley, in collaboration with the "Institut de Protection et de Sureté Nucléaire" in France and in the framework of the Géofrance3D program. The data we collected allow us to analyze evidence of seismic activity and even to quantify the long-term displacement of the fault. The Moyenne Durance Fault appears to be a left-lateral structure, with an inverse component, with a slip-rate between 0.17 and 0.6 mm/yr. These estimates are in relative agreement with paleosismological observations within a single trench that suggested the possibility of a M~7 seismic event between 26 and 9 kyr. BP and thus confirming long recurrence intervals for this fault (Sébrier et al., 1997).
Presently, we are testing three complementary approaches: (1) GPR sounding (with B. de Voogd, Univ. of Pau, France) to localize shallow subsurface fault traces; (2) inversion of topographical signal to determine active fault parameters; and (3) calculating new focal mechanisms for the area, even for low magnitude earthquakes (3.3<M<4.2) to characterize the present stress field of the southeastern France (in collaboration with N. Béthoux, Univ. of Nice, France). Dating of some terraces of the Durance and Rhone Rivers with the cosmonucleid method (with D. Bourlès, CEREGE, France) is also in progress and will contribute to quantify displacements on the active faults that are in the area.
Sébrier M, Ghafiri A & Blès JL, J Geodynamic, 24, 1-4, (1997).
The Kinloch Hourn fault, located within the Kintail seismic zone of north-west Scotland is recognised as the UK's most striking postglacially-reactivated fault. Initiated as a major right-lateral basement fault during the Caledonian orogeny, the fault has recently been interpred as having undergone left-lateral reactivation during the Quaternary. Specifically, in the 13 kyr BP since deglaciation of the Devensian Scottish ice mass, drainage courses astride a 14 km long section of the fault are viewed to have been deflected left-laterally by circa 160 m.
Field and aerial photograph examination of the area indicates a more complex and widespread pattern of fault reactivation. Numerous sharply defined linear bedrock scarps, several kilometres long and one or more metres in height occur throughout the area, particularly along valley sides and near ridge tops where they are not obscured by Quaternary deposits. These features in general show a Caledonian trend although a number are aligned NNW-SSE and all display dip-slip displacement. Evidence from offset glacially smoothed bedrock surfaces clearly indicates that fault movement has occurred since deglaciation, some 10 kyrs BP. The Kinloch Hourn fault although a striking feature is thus not the isolated structure that was originally proposed. Similarly, field and air photograph examination do not support the extensive left-lateral displacement of the fault rather dip-slip movement.
The palaeoseismic history of the Kinloch Hourn Fault is vague, with sheared intra-fault-zone peat deposits and liquified sediments suggesting at least one palaeoseismic event in the last 2.4 kyr BP. Detailed biostratigraphical studies of peats which lie above liquefied sediments or within postglacial fault bounded basins, combined with radio-carbon dating, are being utilised to establish a palaeoenvironmental chronology of seismic events. Theory would tend to suggest that seismic activity was greater during and immediately following deglaciation (e.g. 13-9 kyr BP). The view that the Kinloch Hourne fault produced large siesmic events in the recent past perhaps suggests that contrasting tectonic processes are associated with this feature. The current work will provide a more rigourous framework within which the palaeoseismic activity can be assessed.
The Northern Jura Mountain Front is a tectonically active area. Analysis of maps derived from Digital Elevation Models, aerial photographs and satellite imagery allowed to identify marks of active tectonics associated to several hiden or visible faults with a best accuracy with high resolution DEM (10 m). Numerous morphological features of landslides, bended or irregular slopes, drag folds are pointed out using the shaded DEM. Most of these objects are not visible directly from aerial photographs nor satellite imagery and their detection and analysis on the field was guided through the DEM derived products.
These tools are powerful in order to identify potential structures which could be considered as the emergence of seismogenic sources of the Basel earthquake.
In the vicinity of the epicentral zone of Basel earthquake, the analysis of high resolution DEM reveals abnormal bended slopes and tilted terraces. This site is close to the Dieboldslöschli cave where possible co-seismic deformations and ruptures are observed and have been dated between 1000 and 1400 AD [Carbon et al., 1998]. Another potential source is located on the western side of the Birs river valley where the analysis of the DEM reveals a strong geomorphic slope which can be associated to a NS normal fault at least 3 km long in the Sundgau Pleistocene molasse. This fault was unrecorded until now and does not appear on any existing geological map. The northern extend of this fault runs towards the Basel city.
Considering the actual knowledge, as pointed previously [Meyer et al., 1995], the seismogenic source of the Basel earthquake is not identified. The use of DEM coupled with aerial and satellite imagery is helpful for such areas with moderate or low seismicity where dense and diverse landuses mask most of the tiny neotetonic morphologies produced by the most recent earthquakes. It allows to select potential sites in order to carry out further various field investigations in paleoseismology (geomorphic, geophysical and geological studies).
Carbon D, Cushing M, Lemeille F, Grellet B, Bitterli Th, Flehoc Chet Innocent Ch, Lesse, Tectonique, karst et séismes, Spéléochronos hors-série, 31-34, (1998).
Meyer B, Lacassin R, Brulhet J and Mouroux B, Terra Nova, 6, 54-63, (1995).
Index of EUG 10 Volume
Further EUG 10 Information
Index of the Journal of Conference Abstracts
Cambridge Publications Home Page
Last Updated on Wednesday, March 17, 1999.
© 1997 Cambridge Publications