Title: Reconstruction of Supernova Gravitational Waves Waveforms: Comparing Three Time-frequency Transform Methods

Abstract: For supernovae gravitational wave signal analysis which intend to reconstruct supernova gravitational waves waveforms, we compare the performance of short-time Fourier transform (STFT), the synchroextracting transform (SET) and multisynchrosqueezing transform (MSST) by a self-consistent time-frequency analysis based pipeline. The simulated supernovae waveforms injected into white noise are identified by a hierarchical clustering method in the time-frequency map and then reconstructed by the inverse time-frequency transforms. We find that in terms of signal reconstruction, the SET method performed the best, especially much better than traditional STFT method in reconstructing signals from data with white noise when valued the signal-to-noise ratio. While concerning the quality of time-frequency figures, the MSST method and SET method have less energy dispersion and were both better than STFT method. The higher energy dispersion in time-frequency figure of STFT is time consuming in the clustering process and reduce the accuracy of signal identification. Our preliminary conclusion is that the SET method is the suitable method for the supernovae gravitational wave signal analysis pipeline though more tests are stilled needed.