Analysis of earthquake-triggered landslides in the south of Iberiatesting the use of the Newmark's method at different scales

Resumen

In this Ph.D. Thesis, a comprehensive methodology has been developed and tested in order to assess the potential earthquake-triggered slope instabilities at different scales. This methodology was applied at a regional scale in order to obtain regional hazard maps in terms of Newmark displacement as a first step in studying earthquake-triggered slope instabilities at specific locations. The Newmark displacement represents the expected slope displacement due to strong ground motion. The construction of the Newmark displacement map is based on GIS technology (ArcGIS 9.3) and results from computing several sets of maps. The first aim of this calculation is to obtain a critical acceleration map -i.e., the minimum horizontal seismic acceleration to overcome shear resistance and initiate sliding, provided the static safety factor is known. The second aim is to estimate Newmark displacement combining the critical acceleration map with a peak ground acceleration map (PGA) by means of an empirical relationship. To produce the critical acceleration (ac) map, a lithological map is firstly arranged from digital geological maps from the Institute of Geology and Mines of Spain (IGME). Strength parameters -specific weight, cohesion and friction angle- are assigned to each lithological unit based on a database derived from geotechnical bibliography as well as from available geotechnical tests. Then, a map of static safety factors was estimated considering an infinite-slope limit equilibrium model. Finally, the safety factor map is combined with a slope map to produce a critical acceleration map, which can be regarded as a map of seismic landslide susceptibility. To estimate the Newmark displacements, different seismic input scenarios are considered: probabilistic, pseudo-probabilistic and deterministic. The former scenario considers hazard maps in terms of peak ground acceleration (PGA) corresponding to different return periods (or exceedance probability levels) commonly used in the engineering design of structures. Pseudo-probabilistic seismic scenarios are considered assuming the occurrence of the most probable earthquake for a specific return period at every location, while the deterministic scenarios are devised by considering the complete rupture of the main active faults in the study area. Input PGA values are originally referred to rock conditions, so it was necessary to consider the influence of local site effects (soil and topography factors) in the amplification of strong ground motion. Soil amplification factors are adopted from previous studies concerning seismic hazard in the south Spain (RISMUR and SISMOSAN projects), while the topographic factor has been particularly evaluated in this Ph.D. Thesis considering the slope and relative height of the ridges, following Eurocode-8 provisions. Finally, amplified PGA scenario maps are computed with the critical acceleration map by means of a regression equation that correlates Newmark displacement to the critical acceleration ratio (ac/PGA). The proposed methodology has been firstly used to perform a regional hazard assessment of earthquake-triggered slope instabilities in the Lorca and Granada basins and the Sierra Nevada Range. These regions are one of the most seismically active areas of the Iberian Peninsula. Three significant seismic series have recently struck the Lorca Basin: 1999 Mula (Mw=4.8, IEMS=VI), 2002 Bullas (Mw=5.0, IEMS=V) and 2005 La Paca (Mw=4.8, IEMS=VI-VII). Despite their small magnitude, these earthquakes produced significant damage on buildings, as well as they induced the well-known Bullas and La Paca rock slides. In the Granada Basin, several slope instabilities (mainly rock falls and landslides) can be related to historical earthquakes, such as the 1884 Arenas del Rey earthquake (Mw~6.5, IMSK=X), as well as to instrumental ones (e.g. 1956 Albolote earthquake). The resulting Newmark displacement maps show that seismically-induced landslide hazard in the Lorca and Granada Basins and Sierra Nevada Range can be considered as low. However, the occurrence of widespread slope instabilities across these areas is expected if a low-frequency but powerful earthquake (Mw>6.6) related to the rupture of one of the main active faults in the area takes place. In addition, this approach has allow to identify disrupted-type slides as the most likely earthquake-triggered slope instability in Lorca and Granada basins and Sierra Nevada Range. These instabilities seem to be related to a threshold Newmark displacement of 2 cm or even smaller. A detailed study of the 2002 Bullas and 2005 La Paca rock slides have been performed in order to evaluate the applicability of the developed method at different scales. In this sense, it has been showed in this Ph.D. Thesis that the evaluation of earthquake-triggered landslides at a regional scale can produce wrong estimates of Newmark displacements. Nevertheless, the regional scale maps are useful to show in a preliminarily way the areas with the highest susceptibility and hazard that can be interesting for subsequent site-specific studies at a larger scale. A critical Newmark displacement value of 3 cm has been obtained as the minimum threshold to trigger disrupted-type slope instabilities similar to the 2002 Bullas and 2005 La Paca rock slides. The results obtained at a sub-regional scale agreed with those obtained in these slides at a site scale. A new approach is also proposed in this Ph.D. Thesis to reassess the magnitude and epicentral location of pre-instrumental earthquakes by means of the implementation of the Newmark's method in the study of singular earthquake-triggered landslides. This methodology has been applied succesfully to the Güevéjar landslide that was triggered both by the 1755 Lisbon and 1884 Arenas del Rey historical earthquakes. A minimum Mw 8.5 and an epicentral distance from the Güevéjar landslide of 580 km have been estimated for the 1755 Lisbon earthquake. For the case of the 1884 Arenas del Rey earthquake, the estimated minimum magnitude was Mw 6.5 and the epicentral distance from the Güevéjar landslide was 55 km. In both cases, the results agree with the magnitude and epicentral location suggested by other authors. In addition, it has been confirmed that the 1884 Arenas del Rey earthquake was most likely related to the rupture of the Ventas de Zafarraya Fault. Finally, the Newmark's method is also applied successfully to consider the potencial seismic reactivation and testing the efficiency of slope stabilisation measures of present-day slope instabilities (e.g. Güevéjar and Diezma landslides). This approach provides minimum magnitudes at different epicentral distances that are required to overcome the critical acceleration value and thus to trigger the instability. It has been found that the Güevéjar and Diezma landslides would be very likely reactivated by an earthquake related to the rupture of some of the active faults in the Granada Basin (e.g. Granada, Atarfe and Santa Fe faults).