Reducing the Uncertainties in Ground Motion Prediction Equations by Incorporating the Frequency and Amplitude of the Fundamental Peak of the Earthquake Horizontal-to-Vertical Spectral Ratios
Loading...
Authors
Yazdi, Mohammad
Issue Date
2022
Type
Dissertation
Language
Keywords
ground motion model , ground motion prediction equation , horizontal-to-vertical spectral ratio , local site effects , site fundamental frequency , strong ground motion
Alternative Title
Abstract
In ground motion prediction equations (GMPEs), VS30 (time-averaged shear-wave velocity of the upper 30m) and/or ZX (depth to x km/s shear-wave velocity isosurface; X is equal to either 1.0 km/s or 2.5 km/s) are widely used to predict site effects. Past research showed that other site proxies such as site fundamental frequency (or site period) can also be an effective site parameter since VS30 is limited to the upper 30 m and does not incorporate the effect of deeper subsurface layers, and ZX is an inferred value from the Geophysical tests. This report describes the procedure implemented to estimated site fundamental frequency and its corresponding amplitude from earthquake horizontal-to-vertical spectral ratio and incorporate these two proxies into GMPEs. The work done is presented in three parts.In the first part, the common practices in obtaining HVSR and estimating the site fundamental frequency, which is the peak frequency in the HVSR curve. However, these practices can be very subjective. Therefore, a set of four fully automated methods are developed to estimate site fundamental frequency, which not only removes analysts’ subjectivity, but incorporate the uncertainty in the estimations. The outputs of these automated methods are (a) maximum likelihood estimate of site fundamental frequency (f_ml), (b) its corresponding amplitude (A_ml), and (c) the uncertainties associated with f_ml and A_ml estimations. After assuring the satisfactory performance of these automated methods, an article was published in the Seismological Research Letters (SRL) in March 2022 (https://doi.org/10.1785/0220210078).Second part aims to incorporate f_ml and A_ml into the NGA-West2 GMPEs. First, a subset of NGA-West2 database, which is eligible for HVSR method, is selected, and f_ml and A_ml are estimated for about 1,200 stations in this data set. Second, an HVSR-based model is developed to improve the prediction of site-effects without changing the coefficients of the NGA-West 2 GMPEs. In other words, the proposed model is added to the existing GMPEs and reduces the uncertainty in site term residuals (ϕ_S2S) by 13.5% on average, and the total uncertainty in the NGA-West2 GMPEs is reduced by 3.5% on average over the intensity measures (i.e., peak ground velocity, peak ground acceleration, and 21 spectral periods). These improvements are on top of the existing site proxies (i.e., VS30, Z1.0 and Z2.5) This part was submitted to Earthquake Spectra, and it is conditionally accepted with minor revision at the time of submission of this technical report.Third part includes (a) a new database for KiK-net in Japan, and (b) an HVSR-based model based on the KiK-net database. In this part, first, the entire KiK-net was downloaded and processed according to the NGA-West2 project guidelines. This database includes about 413,000 earthquake recordings on the ground surface from approximately 14,900 events and 699 stations. Using a subset of this database, a site-effects model based on f_ml and A_ml is proposed to be added to two NGA-West2 GMPEs. Supplementing the NGA-West2 GMPEs with this model reduced the uncertainty in site-term residuals (also called site-to-site variability, ϕ_S2S) by 12.6% on average over the intensity measures. The manuscript of this part is in preparation and will be submitted to a peer-reviewed journal in the near future.