Estimation of time–frequency-dependent lagged coherence for ground motions in simulating multipoint ground motions

Abstract

Traditional coherence functions ignore the time-varying characteristics of ground motions, influencing the accuracy of simulating multipoint nonstationary ground motions. An effective process is proposed for building the time–frequency lagged coherence expression to describe the spatial variation in both the time and frequency domains. First, the advantage of wavelet packet transform (WPT) in recognizing the time–frequency characteristics of signals is leveraged. The WPT is used to analyze the time–frequency characteristics of ground motions and estimate the time–frequency lagged coherence using the observed recordings from SMART-1 array. Then, based on the frame of simulating multipoint ground motions, a blind test for verifying the rationality of time–frequency lagged coherence is conducted to simulate spatially correlated ground motions using time–frequency and traditional lagged coherence functions. The spatial coherence of the target region is observed to have clearly lost its inherent time-varying property. The comparative results confirm the significance of using the time–frequency lagged coherence in simulating multipoint ground motions.