Database for design-compatible waveforms

Abstract

Abstract on selection of ground motions for microzonation. We have analyzed and compared the stations metadata of three strong motion databases: KiK-net (Japan), ESM (Europe and Middle East), and NGA-West2 (USA). For each data-base, we propose a subset of stations suitable for the purposes of the project. These sub-sets of stations comprise 772 sites classified in terms of SIA261 (2020) soil categories (A-E), inspecting the provided VS profiles. For these sites, the metadata of the stations were checked. To increase the number of recordings from earthquakes with large magnitude and short source-site distance, we added 870 three-component recordings, 28 related to the Chi-chi earthquake sequence recorded by 11 Taiwanese seismic stations and 842 from 69 seismic stations in ESM classified by using geologic criteria. This further selection was made to enlarge the dataset of waveforms mostly related to soil class A. The resulting subset of waveforms comprise 51,144 three-component recordings, that we subdivided in terms of SIA261 (2020) soil categories (A-E), inspecting their distribution in terms of magnitude and Joyner-Boore distance. From this database, we extract 15,303 three-component recordings based on the relevant scenarios from the disaggregation of the seismic hazard for Switzer-land. This collection of recordings was then supplemented with synthetic strong-motion waveforms from the BB-SPEED set to fill gaps in the magnitude-distance distribution. Once the disaggregation and the waveforms database are ready, we set a strategy for the selection of 11 waveforms, compatible with the elastic response spectrum defined in SIA261 for each seismic zone. Among the existing algorithms used to select compatible waveforms to a target, the most adaptable to our waveforms database is the one proposed by Baker and Lee (2018). The key steps of the algorithm are: the screening of the ground motion database for suitable motions, statistically simulating response spectra from a target distribution, finding scaled motions whose spectra match each statistically simulated re-sponse spectrum, and then performing an optimization to further improve the consistency of the selected motions with the target distribution (more details in Baker and Lee, 2018). For the selection of the 11 design-compatible waveforms, we further implement the Euro-code 8 (2022) criteria. Our selection in this report is made for microzonation purpose. We fit the target elastic response spectrum for a return period of 475 years, in the waveforms’ period range 0.02-2.0 s. The selection is initially attempted using unscaled waveforms, but if necessary, a scaling factor between 0.5-2 is then allowed. For each seismic zone, we selected all possible combinations of 11 scaled waveforms. Among these, we choose the set of waveforms with the lowest RMSE if most of the wave-forms are nearby or better inside the disaggregation borders, otherwise we check for an-other set of waveforms. Finally, we provide recommendations for the use of the selected and scaled waveforms in microzonation studies in Switzerland.

Abstract on selection of ground motions for structural analysis. We set a strategy for the selection of 11 two-component waveforms, compatible with the elastic re-sponse spectrum defined in SIA261 for buildings of importance class III and for each seismic zone and soil class. Our selection in this report is made for structural analysis purpose. We fit the target elastic response spectrum in SIA 261 multiplied with a factor 1.5, corresponding to about 975 years return period in the seismic hazard. Three period ranges were chosen to cover the fundamental periods of most of the Swiss building stock, in agreement with the definition of period range in the EC8 rules (0.2T0-1.5T0). In particular, the covered building periods are 0.10-0.66 s, 0.50-1.33 s and 1.25-2.66 s respectively. The selection is initially attempted using real waveforms scaled up to a factor of 2, but if necessary we add to the database synthetic waveforms and we scale the waveforms up to a factor of 3. For each seismic zone, we selected all suitable combinations of 11 two-components scaled wave-forms. The search is made using the geometric mean of the two horizontal components. For the verti-cal components, the same scaling is applied, although we do not expect that we match the vertical ground motion design spectrum defined in SIA261. Among the possible suitable combinations, we choose the set of waveforms with the lowest RMSE if most of its time histories are nearby or better inside the disaggregation magnitude-distance boundaries; otherwise we select another set of wave-forms satisfying the latter criterion. In particular cases, some search constrains had to be waived (e.g. selection for spectral period range 0.1-2.0 s zone Z2 soil class E we allow a scale factor up to 3).