Surface and subsurface palaeoseismic records at the ancient Roman city of Baelo Claudia and the Bolonia Bay area, Cádiz (south Spain)

The Roman archaeological site of Baelo Claudia (Cadiz, south Spain) is located within the Gibraltar Arch, a region with no significant recent or historical seismicity. However, previous studies have emphasized the occurrence of repeated strong archaeoseismic damage (intensity IX MSK) at Baelo Claudia tentatively bracketed in this study around AD 40–60 and AD 260–290. A multidisciplinary study has been carried out including the detailed mapping of surface deformation and building damage, surface geology and geomorphology, collection of structural data, and an extensive ground penetrating radar (GPR) survey. The obtained data are not conclusive when considered separately, but evident links between archaeoseismic damage, structural and GPR data indicate that the destruction of the city was linked to seismic shaking. The analysis of the pattern and orientation of deformation clearly indicates SW–NE directed compression due to ground shaking. This analysis also focuses on localized landslides and liquefaction processes, which appear to be coeval with the earthquakes, but the poor geotechnical parameters of the clayey substratum were determinant to amplify the observed level of destruction. The application of the present Spanish seismic code (NCSE-02) indicates that intensity VIII MSK (0.24–0.26 g) can be reached in this zone for 500 year return periods. Previous studies reported on the occurrence of repeated strong archaeoseismic damage (intensity IX MSK) in the ancient Roman city of Baelo Claudia (first to fourth centuries AD), located at the axial zone of the Gibraltar Strait in southern Spain (Menanteau et al. 1983; Goy et al. 1994; Silleres 1997; Silva et al. 2005, 2006). The archaeological stratigraphy of the city evidences two major episodes of abrupt city destruction tentatively bracketed during AD 40–60 and AD 350–395, separated by an intervening horizon of demolition for city rebuilding (Silleres 1997; Silva et al. 2005), elsewhere interpreted as characteristic of many earthquake-damaged archaeological sites in the Mediterranean (i.e. Stiros 1996; Marco et al. 1997; Hancock & Altunel 1997; Stiros & Papageourgiu 2001; Altunel et al. 2003). The urban geology and general geomorphological features of the study area are described in the previous works of Borja et al. (1993), AlonsoVillalobos et al. (2003) and Silva et al. (2005), evidencing that the Roman city was differentially founded on plastic clayey substratum, Late Quaternary sandy and clayey materials thickening to the coast (up to 5 m thick), and a variable amount of poorly compacted Roman artificial fillings. In addition, all these materials are characterized by poor geotechnical parameters, including high swelling rates for the underlying clayey substratum (Borja et al. 1993). In detail, Silva et al. (2005) From: REICHERTER, K., MICHETTI, A. M. & SILVA, P. G. (eds) Palaeoseismology: Historical and Prehistorical Records of Earthquake Ground Effects for Seismic Hazard Assessment. The Geological Society, London, Special Publications, 316, 93–121. DOI: 10.1144/SP316.6 0305-8719/09/$15.00 # The Geological Society of London 2009. applied the existing seismic code of Spain (NCSE-94 1997) concluding that it will be necessary to invoke site effect amplification to link the observed damage (over VIII–IX MSK) with local seismic sources. On the other hand, these authors dismiss linking the observed ground deformations with the wellknown AD 365 Crete tsunami event as earlier proposed by other authors (Menanteau et al. 1983), because no evidence of tsunami-related damage is recorded in the ruins. However, recent findings in small fluvial outlets east of Baelo Claudia indicate probable tsunami occurrences in cal. 2150–1825 BP (Alonso-Villalobos et al. 2003) and again in the fifteenth and sixteenth centuries (Becker-Heidmann et al. 2007) in Bolonia Bay. The tsunami event during Roman times is geologically well documented along the Spanish Atlantic coast of the Gibraltar Strait from the Donana marshlands to Cadiz (Lario et al. 2001; Luque et al. 2002; Ruiz et al. 2004) and might correspond with one of the three historically documented tsunamis that occurred between 219 BC and 60 BC in this zone (Campos 1991; Luque et al. 2001). Nevertheless, strong seismic events occurred during the Roman period, presumably affecting southern Spain: the AD 33, AD 346 and AD 382 earthquakes are listed in the Spanish seismic catalogues (i.e. Galbis 1932; Martinez Solares & Mezcua 2002). All these events are poorly documented and supposedly related to the far-field Cape of San Vicente seismic source, responsible for the AD 1775 Lisbon event (Campos 1991; Luque et al. 2001). Dates of some of these historic events match with the assumed ages of AD 40–60 and AD 365–395 proposed for the two events of destruction recorded at Baelo Claudia (Menanteau et al. 1983; Silva et al. 2005). However, even these relatively strong earthquakes produced only ground motions of moderate intensities in the Gibraltar Strait region between VI–VII MSK (Martinez Solares et al. 1979) and V–VI EMS (Martinez Solares 2001) as recorded during the AD 1775 Lisbon event. Ground motions leading to these intensities are insufficient to explain the extensive destruction of the city, making it necessary to consider and check the specific site effect amplification at Baelo Claudia. In any case, an important limitation of archaeoseismological data is that they generally do not allow identification of the causative seismic source (i.e. capable fault) producing the observed architectural disruptions (Hancock & Altunel 1997; Altunel et al. 2003; Similox-Tohon et al. 2006). Additionally, in the particular case of Baelo Claudia, most of the observed archaeoseismological disruptions can be classified as secondary effects of ground shaking (Silva et al. 2005), but might have also been produced by other natural hazards (landslides, ground subsidence, soil creep, etc.) under specific climate and weather conditions. Therefore, thoughtful research is necessary to identify the true nature of the observed deformations by means of analysis of their surface and subsurface records. We tend to divide the observed archaeoseismological damage into two groups: (1) release of impulse-like high energy during seismic events and rupture-like processes; and (2) low energy events, mainly due to gravitational forces, producing ‘slow’ processes, like creep and landslides. This work focused on the detailed mapping of surface deformations and architectural disruptions within the ancient urban zone of Baelo Claudia, in order to determine the extension, nature and structural pattern of the recorded deformations as earlier performed for other archaeological sites in the Mediterranean zone and central Europe (Korjenkov & Marzor 1999; Hinzen & Schutte 2003; Monaco & Tortorici 2004; Similox-Tohon et al. 2006). Measurement of structural data on fractures, cracks, shocks, and pop-up structures affecting the ancient Roman pavements and walls, directions of collapse of columns, houses and city walls, offer enough data to discuss the directivity and pattern of the ground movement triggering the observed architectural destruction. Therefore, we use these kinematic indicators preserved in the ruins in order to probe the directional nature of the deformations to separate effects caused by the eventual progressive ruin and burying of the city from those pointing to seismic shaking. Geophysical analysis of subsurface archaeological remains by means of an extensive ground penetrating radar (GPR) survey has been carried out in order to determine the depth extension of the observed surface deformations. Additionally GPR data have been probed to offer quality information about the subsurface location of probable event horizons. Geodynamic setting of the Gibraltar