Estimating the Coseismic Displacement of Ground Motions with Fling-Step Effect Based on an Intrinsic Timescale Decomposition Algorithm

ABSTRACT Strong ground motions resulting from the fling-step effect often exhibit significant static displacement. However, recorded ground-motion signals may display an unreasonable baseline offset because of contamination by low-frequency (LF) noise. Baseline correction is essential to eliminate this offset while preserving the coseismic displacement component. Many studies have generalized the two-stage offset model and proposed methods for empirically determining the starting times (t1 and t2) of the two stages with different levels of offset. To minimize errors caused by the subjective determination of t1 and t2, this article introduces a novel method for detecting optimal values of t1 and t2 and recovering the coseismic displacement of ground motions. The proposed approach uses an intrinsic timescale-decomposition-based filter to eliminate the LF offset component and introduces a quantitative index to identify the starting time points of the offset. After these time points are determined, the corresponding baseline offset is computed using the two-stage offset model. The effectiveness of this method is validated by comparing its results with the coseismic displacement data recorded by dense Global Positioning System networks during the 1999 Chi-Chi earthquake and the 2011 Tohoku earthquake. In addition, several existing empirical baseline correction methods are assessed. The results confirm that the proposed method effectively removes the baseline offset and accurately recovers coseismic displacement from strong-motion records.