Report GIRS 2022-02: An Efficient Computational Platform for Selecting and Scaling Ground Motion Records while Considering Multiple Target Spectra

By: Shokrabadi, Mehrdad; Bozorgnia, Yousef; Burton, Henry V.; Baker, Jack W.; Askari, Mohammad
DOI: 10.34948/N3K01N

Abstract: We developed an efficient computational platform to select and consistently scale recorded earthquake ground motions. Given a set of target response spectra for horizontal and vertical ground motion components, considering period-dependent record-to-record variabilities, a set of ground motions is selected and scaled such that: (1) the mean spectra of the selected and scaled horizontal motions would follow the target horizontal spectra; (2) the mean spectra of the selected and scaled vertical motions would follow the target vertical spectra; (3) the selected set of horizontal spectra would preserve the prescribed period-dependent record-to-record variability for the horizontal component; (4) the selected set of vertical spectra would match the prescribed period-dependent record-to-record variability for the vertical component; and (5) for each set of horizontal and vertical components, a single scaling factor is used; thus, preserving the relative amplitude and phasing of the original recorded horizontal and vertical components. Additionally, significant improvement in computational efficiency is achieved by employing a modified version of a greedy record selection algorithm. More specifically, the run time of the modified algorithm is significantly reduced by utilizing the parallelization capabilities that are present in most modern desktop and laptop computers. The process is demonstrated for sites where the hazard is dominated by shallow crustal earthquakes.