Anibal Ojeda

Seismic Hazard in Istanbul following the 17 August 1999 İzmit and 12 November 1999 Düzce Earthquakes

Two recent destructive earthquakes that occurred along the western part of the North Anatolian Fault Zone (NAFZ), the 17 August 1999 ( M w 7.4) Izmit and 12 November 1999 Duzce ( M w 7.2) earthquakes, have caused major concern about future earthquake occurrences and their possible consequences in the Istanbul area. Probabilistic seismic hazard analyses are performed for the larger Istanbul area including the Gulf of Izmit and the Marmara Sea region. Hazard computations were done assuming different combinations of four attenuation relations and three alternative source models. The three models used are (1) Standard Poissonian earthquake occurrence with area sources, (2) renewal model (assuming characteristic earthquakes) with area and fault sources, and (3) renewal model (assuming characteristic earthquakes) with refined area and fault sources. Results are presented in twelve different maps of peak ground acceleration (PGA) with a 10% chance of exceedance in 50 yr. Among the earthquake recurrence models, the results assuming model 3, yield the highest PGA values, reaching greater than 0.3 g at the western end of the Gulf of Izmit. In general, PGA values decrease toward north and are reduced down to less than 0.2 g in central Istanbul in the Bosphorous area. The four attenuation relations examined display significant variations, and their effects become especially critical for distances less than 50 km. In addition, spectral hazard levels (corresponding to 475-yr return period) are computed and presented as uniform hazard response spectra for 5% damping. Two sites are selected, one in hard rock and the other in soft sediments. In general, these results correlate well with the Turkish Seismic Design code (TSDC) recommendations for the two site conditions. The estimates of the future earthquake hazard potential of the Istanbul area are sensitive to our present-day understanding of the behavior of the fault segments in the Marmara Sea region.

Crustal structure and local seismicity in Colombia

Using P-wave travel time data from local seismicity, the crustal structure ofthe central and southern part of Colombia was determined. A very stableand narrow range of possible velocity models for the region was obtainedusing travel time inversion. This range of models was tested with earthquakelocations to select the best velocity model. The 1D velocity modelproposed has five layers over a halfspace, with interfaces at depths of 4,25, 32, 40 and 100 km and P-wave velocities of 4.8, 6.6, 7.0, 8.0, 8.1and 8.2 km/sec, respectively. According to this model the Moho lies at32 km depth on average. For P-waves, the station corrections range from–0.62 to 0.44 sec and for S-wave they range from –1.17 to 0.62 sec.These low variations in station residuals indicate small lateral velocitychanges and therefore the velocity model found should be well suited forearthquake locations and future starting model for 3D tomography studies.Using this new velocity model, the local earthquakes were relocated. Theshallow seismicity, < 30 km, clearly shows the borders betweentectonic plates and also the main fault systems in the region. The deepseismicity, > 80 km, shows two subduction zones in the country: theCauca subduction zone with a strike of N120°E, dip of 35°and thickness of 35 km, and the Bucaramanga subduction zone which has,for the northern part, a strike of N103°E, dip of 27° andthickness undetermined and, for the southern part, a strike ofN115°E, dip of 40° and thickness of 20 km. Based ondifferences of thickness of brittle crust in the subducted slab and spatialdistribution of the seismicity, the Cauca and Bucaramanga subduction zonesseem to represent independent processes. The Cauca subduction seems tobe connected to the process of the Nazca plate being subducted under theNorth Andes Block. In the Bucaramanga subduction zone, the transitionbetween southern and northern parts and changes in geometry of the slabseem to be gradual and there is no evidence of a tear in the slab, howeverthe local seismicity does not allow us to determine which plate or plates arebeing subducted. The Bucaramanga nest appears to be included into thesubducted slab.

QLg tomography in Colombia

The crustal attenuation of Lg waves in Colombia was estimated and analyzed using local seismological data from the National Seismological Network of Colombia (RSNC). The selected dataset comprises 510 crustal earthquakes with a total of 2928 ray paths. This large dataset allowed us to invert for a regional average of QLg, tomographic images for Colombia, and simultaneously, source size for each event and site term at the seismological stations. The computed regional average for QLg in the frequency band 0.5–5.0 Hz was found to agree with the previously reported values in neighboring regions. In order to resolve the lateral variation in the attenuation of Lg waves in the crust, independent tomographic inversions for 26 frequencies between 0.5 and 5.0 Hz were conducted. The resulting maps confirm that heterogeneities in the crust exist and that they are related to the large-scale tectonic features in the country. The highest attenuation in the region is linked to the presence of active volcanic arcs, where the crust is weakened by partial melting. Relatively high attenuation is also found in zones where the crust is composed of accreted oceanic rocks and in regions with near-surface low-velocity sedimentary layers.

The new accelerograph network for Santa Fe De Bogota, Colombia and implications for microzonation

The new accelerographic network of Santa Fe de Bogota is composed of 29 three-component stations with sensors at the surface and three additional six-component borehole stations with three sensors at the surface and three at depth (115, 126 and 184 m). In total, 32 stations have been operative in the metropolitan area of Bogota since 1999. During this period of time, a significant number of weak motion are recorded and used for a preliminary analysis of local site effects. Using the SH-wave response spectra we verify the behavior of the different seismic zones proposed by the previous microzonation study of the city. A comparison between normalized SH-wave response spectra and the normalized design spectra for each zone clearly depicts that parts of the design spectra should be revised, as well as the boundaries between different zones may require some changes. The spectral amplification levels reach up to a factor of 5. The predominant periods obtained by the amplification spectra in different stations in the city, show variability from 0.3 to 3.0 s. A comparison is also made between the predominant periods obtained using H/V spectral ratios of microtremors and those using weak motion. In general, microtremors tend to predict slightly lower values of dominant periods than those calculated by the weak motion spectra. However, there is a general correlation between the two data sets. Using the data recorded by one of the borehole station, an equivalent linear seismic response analysis was conducted. The modeled and recorded response spectra show similarities in period peaks, however, the modeled soil amplification is underestimated for periods less than 0.8 s. Since the available record is weak motion which represents mostly the linear response of the soils, further analysis is required.

Integration and influence of paleoseismic and geologic data for the seismic hazard evaluation in the Catalan coastal ranges, Spain

Abstract In this study, the influence of paleoseismic and geologic data in the seismic hazard estimation for the Catalan coastal ranges is analysed. We computed the probabilistic seismic hazard using area seismic sources with a Poissonian assumption for the earthquake occurrence. For the computations, a previously published attenuation relationship based on European strong motion data was applied. The resulting hazard estimates show similarities to the previous assessments in the region. These results were then used as a reference for comparison with other new models. In order to analyse the influence of the paleoseismic data three different models were tested. Since the number of faults that are investigated in detail are few, the same area sources that were used in the Poissonian assumption were kept in all three new models. In addition, the new paleoseismic data with faults expressed as line sources were used. In this case, a cyclic earthquake occurrence was assumed. The three models were based on the paleoseismic data with different assumptions on the time elapsed since last event. The time elapsed was set to 0, 10 and 85% of the recurrence interval in each model. The results are presented as maps showing the difference between the three models and the reference model with the Poissonian assumption. The results are given in horizontal peak ground acceleration contour maps for different return periods, also taking into account large return periods as high as 25,000 years. This is done to demonstrate the effect of large recurrence intervals found for some of the active faults. In general, we observe that for short return periods ( 5000 years) the effects of the paleoseismic data become increasingly significant. In order to estimate the true seismic hazard potential of this apparently low seismicity area, long-term behaviour of the possible active faults in the region needs to be investigated systematically.