Paleoseismic transect across the alhama de murcia fault and implications of a fault-based seismic hazard assessment for the eastern betics

Resumen

Resumen de la Tesis: The SE of Spain is one of the most seismically active regions of the Iberian Peninsula. In this framework, the Eastern Betics Shear Zone (EBSZ) is one of the most prominent fault systems absorbing an important part of the Eurasia-Africa convergence. The EBSZ has caused damaging earthquakes in historical times, the most recent being the 2011 Lorca earthquake (Mw 5.2). The unexpected damage associated to this moderate event evidenced that the hazard estimations for the region needed to be reviewed. This context has motivated the development of several paleoseismic studies within the EBSZ for over two decades aiming to characterize its faults for the seismic hazard assessments. Despite this, the available data is usually too local and, for some faults, reliable and representative parameters are still lacking. Such heterogeneity in the fault parameters of the EBSZ has usually hampered the development of seismic hazard models based on geological data, yet they could provide more realistic hazard estimations as proven in regions such as California, New Zealand or Italy. Based on the outlined issues, this thesis has been designed to be a comprehensive study that allows to: i) contribute to the completion of the paleoseismic record of one of the most active faults within the EBSZ, the Alhama de Murcia Fault (AMF), by obtaining more representative paleoseismic data in one of the fault transects, and ii) include faults and paleoseismic data as the primordial inputs to derive earthquake occurrence in a probabilistic seismic hazard assessment (PSHA) of the EBSZ. To achieve so, the study is divided in two main parts. The first part (A) is a local study and presents a structural characterization and integral paleoseismic study in four sites across a complete transect of the AMF, where previous studies focused only on a single fault site. The second part (B) is a regional study and presents a fault-based PSHA of the EBSZ where the main faults of the system and their available data are used to define scenarios of fault ruptures and to calculate their respective earthquake rates for the seismic source modelling. A) The structural characterization has been carried out in the two central segments of the AMF, Lorca-Totana and Totana-Alhama, by means of geomorphological mapping and field analysis. In the former segment, five subparallel fault branches (N50-65º) have been identified in a ~2 km wide transect with partitioning of the overall sinistral-reverse deformation. From N to S: N1-AMF (lateral), N2a and b-AMF (dip slip), S-AMF (lateral) and F-AMF (dip slip). Four of these structures bound two pressure ridges developed parallel to the main mountain front. The Totana-Alhama segment shows more diffuse deformation over a wider transect (>2km), where the most prominent structure is the Amarguillo Fault (AF), a ~N10º ramification of the AMF. The AF is composed by at least five branches forming a transtensional bend with left-lateral component. Based on these results, a paleoseismic survey has been carried out in the Lorca-Totana segment, which implied the excavation of eight paleoseismic trenches across four fault branches in a NW-SE transect. In each trench, we have performed i) a paleoseismic analysis to identify the paleoearthquake record and to derive the recurrence intervals (RI) of major events for the four branches, and ii) a structural analysis of the vertical and lateral slips to estimate the slip rates in each branch and for the whole segment. To do so, we relied on new radiocarbon and OSL dates, along with previous ones. The work allowed to obtain one of the completest paleoearthquake records in the Iberian Peninsula. This is especially relevant in the S-AMF, where seventeen events for the last ~100 ka are identified. The RI varies from 5.7±1.7 to 4.3±0.4 kyr for the last 73-31 ka, which yields a Poissonian to weakly periodic recurrence behavior based on the coefficient of variation (CV=0.8-1.0). This RI is reduced to 3.1±1.4 kyr if only the last five events for the last 18 ka in the F-AMF are accounted. Conversely, in the two N-AMF branches the paleoearthquake record is more unreliable due to the presence of large sedimentary gaps in the sequences at the apical areas of alluvial fans. The total net slip rate for the last 18 ka is 1.55 +0.14/-0.18 mm/yr (four branches), considerably higher than the previous estimations (0.9±0.1 mm/yr). In this segment, the S-AMF is the controlling fault, and its slip rate evolution since 73 ka shows fluctuations marked by two short accelerations and respective longer slower phases (super-cycles). The most marked acceleration is observed between 31 and 25 ka, coinciding in time and duration with an acceleration in the Carrascoy Fault (34-28 ka). This suggests a possible activity synchronicity between different faults for the first time in the EBSZ. The vertical slip rate evolution reveals that the S-AMF and F-AMF show increased activity with respect to northern parts of the fault system since ~18 ka. Furthermore, their nearly identical vertical slip rates in this period, along with the time compatibility of the last five events in both branches, suggests that they could rupture synchronously. B) The fault-based PSHA of the EBSZ relied, for the first time in Spain, on a particular source modelling with relaxed segmentation, in which different hypotheses of multi-fault ruptures have been envisaged in two models (FM and FM_bg). To do so, we used the SHERIFS code and geological fault data as inputs to compute earthquake probabilities. Each hypothesis has been weighted in a logic tree for PSHA based on their fit with the seismic catalogue and paleoearthquake rates. From this analysis, we suggest that multi-fault ruptures involving lengths up to single to several whole faults of the EBSZ are feasible, while the rupture of the whole system (~400 km) seems unfeasible based on the misfit with the available data. The hazard results of both models show a clear control of the EBSZ faults in the seismic hazard for all return periods, increasing drastically the hazard levels in the regions close to the fault traces and influencing up to 20 km farther with respect to an area source PSHA. The seismic hazard is dependent on the fault slip rates as peak ground accelerations and territorial extension of the fault influence appear higher around the AMF and Carboneras Fault, while lower slip rate faults (Palomares Fault; PF or the northeastern termination of the AMF) show much less contribution to the hazard. For the return period of 475 years and near-fault locations, our models are more consistent with the ground motion values reached in the 2011 Mw 5.2 Lorca event than the building code or national seismic hazard map. This suggests that our fault system-based model performs more accurate estimations for this return period. Paleoseismic parameters, mainly slip rates and its uncertainties, have a clear impact on the seismic hazard and, for some faults (PF), the lack of detailed paleoseismic studies can compromise the reliability of the hazard estimations. This is a key discussion point in the present study, marking the need for better constrained and reliable slip rates in the EBSZ. The integral paleoseismic study of the AMF performed in this thesis has revealed to be a crucial step towards a more representative characterization of the paleoseismic parameters of this fault (slip rates, recurrences, rupture behavior), and thus, of the EBSZ. The acquisition of refined and reviewed paleoseismic parameters is one of the key practices to improve seismic hazard evaluations. In this respect, further paleoseismic studies should focus on poorly researched faults and on performing integrative studies in other faults of the EBSZ, as we do for the AMF. The comprehensive approach followed here, from paleoseismic data collection to PSHA, contributes to perform more critical interpretations of the seismic hazard, and aims to serve as a case example for other low-to-moderate seismicity regions worldwide.