Improving the signal-to-noise ratio of ambient noise cross-correlation functions using the time-scale phase weighted stacking based on a root-mean-square ratio selection scheme

In ambient noise tomography, surface waves between stations are retrieved by cross-correlating continuous noise data recorded at station pairs, from which dispersion curves are measured to determine the Earth's subsurface velocity structure. Cross-Correlation Functions (CCFs) with a high Signal-to-Noise Ratio (SNR) are crucial for accurately measuring dispersion curves and ensuring reliable inversion results. To enhance the SNR of computed CCFs, this paper introduces a new stacking procedure, building on existing algorithms, that employs time-scale Phase-Weighted Stacking (ts-PWS) in combination with a Root-Mean-Square Ratio Selection Scheme (RMSR_SS). For a given CCF dataset, individual short-duration CCFs that make destructive contributions to the final stacked CCF are excluded by analyzing the variance of the root-mean-square ratio between the surface wave window and noise window. Subsequently, ts-PWS is applied to stacking the remaining CCFs to produce the final stacked CCF. We applied this new stacking method to the NECESSArray data, which was deployed in Northeast China through an international collaboration between China, the US, and Japan. The results show that our method can substantially improve the SNR of computed CCFs compared with linear stacking, especially when the ambient noise sources are largely located outside the stationary phase zones. In addition, our method can retrieve stable CCFs from ambient noise data with a very short duration (e.g., 10 days), significantly reducing the time required for seismic station deployments and thus improving the efficiency of ambient noise tomography.