Spectral acceleration at a given period, SA( T), is usually combined with other intensity measures (IMs) for vector-valued ground-motion selection and seismic hazard assessment. The correlations between SA( T) and other IMs have been extensively studied, based on the SA-related IM pairs with a damping ratio ( ξ) of 5%. Yet, ξ varies notably with various types of structures. This article thus investigates the correlations of SA( T, ξ) with 10 commonly used non-SA IMs (e.g. Arias intensity), and the correlations of damping-specific acceleration spectrum, spectrum, and displacement spectrum intensities, namely ASI( ξ), SI( ξ), and DSI( ξ), with the other IMs. Comprehensive correlation analyses are conducted to derive the correlations for the IM pairs considered based on the NGA-West2 database and compatible ground-motion models. The results indicate that increasing ξ generally yields stronger correlations, and the difference between the ξ-specific and 5%-damped correlation coefficients could be larger than 0.2. The neural network and polynomial regressions are subsequently used to develop parametric models for estimating the correlations of SA( T, ξ)-containing pairs and the other IM pairs, respectively. All the models developed exhibit good performance in terms of predictive accuracy. The application of the proposed models is demonstrated within the generalized conditional IM approach. The models proposed could be useful for ground-motion selection and seismic demand or risk assessment for structures with various ξ values.
Read full abstract