Impact of Different Processing Methods on Extraction of Ambient Noise Cross-Correlation Functions

In ambient noise research, efficiently extracting stable and reliable Noise Cross-Correlation Functions (NCFs) is critical. Using one month of continuous broadband seismic data from Sichuan Earthquake Administration stations, this study systematically evaluates the impacts of data preprocessing, cross-correlation, and stacking algorithms on stacked NCFs based on routine workflows. For cross-correlation, when adopting Phase Cross-Correlation (PCC), time window length, window overlap rate, and 1-bit time-domain normalization have negligible effects on linearly stacked NCFs; its expansion Wavelet Phase Cross-Correlation (WPCC) further produces NCFs with a broader frequency band without explicit spectral whitening. For stacking, nonlinear stacking of short-time NCFs significantly improves signal-to-noise ratio (SNR) of coherent signal but causes waveform distortion. Phase-Weighted Stacking (PWS) achieves the highest SNR yet severe distortion, while time-scale Phase-Weighted Stacking (ts-PWS) balances high SNR with mild distortion. Theoretically, PWS extracts instantaneous phase via analytical signals, whereas ts-PWS directly converts time-domain signals to the time-frequency domain via discrete wavelet transform for more effective instantaneous phase extraction from non-stationary time series. Thus, combining WPCC and ts-PWS can generate stacked NCFs with both high SNR and broad frequency band, laying a solid foundation for subsequent research like subsurface structure imaging and medium velocity variation monitoring.