Site-condition map for Portugal, Western Iberia: methodology and constraints on the performance of Vs30 proxies for stable continental regions in Europe.

Information on the amplification characteristics of the near-surface formations in a regional sense is essential to adequately represent both seismic hazard maps and ground shaking maps. Due to the scarceness of shear-wave velocity data in most regions, several methods have been proposed in order to obtain first order representations of Vs30. These include the surface geology method and the topographic slope method. The latter method has become the standard way for incorporating site effects into regional studies worldwide given the convenience provided by the global Vs30 Internet server.

In the framework of project SCENE we developed a shear wave velocity database for Portugal. The database consists of 87 shear-wave velocity depth profiles from a variety of lithological and geological formations. We used an iterative three-step procedure to develop the Vs30 based site-condition map: 1) to define a preliminary set of geologically defined units based on the literature; 2) to calculate the distribution of Vs30 for each unit; and 3) to perform statistical tests in order to estimate the significance of the difference in the Vs30 distribution characteristics between the units. The units were merged according to the results of the statistical tests and the procedure was repeated. We started by classifying the sites into six generalized geological units. The final set consists of three units only: F1 (igneous, metamorphic and old sedimentary rocks); F2 (Neogene and Pleistocene formations); and F3 (Holocene deposits).

We used the database to evaluate the performance of Vs30 proxies. The use of proxies based either on geological units or on correlations with the topographic slope shows relatively unbiased total residual distributions of the logarithm of Vs30. However, the performance of the methods varies significantly with the generalized geological unit analyzed. Both methods are biased towards lower values of Vs30 for rock formations. The topographic-slope method is biased towards higher values of Vs30 for Holocene formations. Our results reinforce those published for SCR in Europe using a sparser database: Vs30 is poorly correlated with topographic slope. We conclude that in the absence of endogenous data the method based on analogue surface geology units should be preferred to that based on topographic slope.