Improving the Retrieval of High-Frequency Surface Waves Using Convolution-Based Three-Station Interferometry for Dense Linear Arrays

Abstract

The retrieval of surface waves from ambient noise is important for delineating the solid earth’s near-surface structures, especially in urban environments. Seismic interferometry (SI) with linear arrays is becoming popular in urban areas with abundant anthropogenic noise. However, retrieving the noise correlation functions (NCFs) is usually challenging for a dense linear array under the demand of short-time recordings and the limited number of stations in urban environments. We comprehensively compare the SI and three-station interferometry, and the results show that the convolution-based three-station interferometry can accurately retrieve the NCFs using short-time recordings for dense linear arrays from traffic-induced noise. A synthetic example demonstrates the superiority of the convolution-based three-station interferometry over the traditional SI and the correlation-based three-station interferometry. Results from two field examples validate the convolution-based three-station interferometry for linear arrays deployed synchronously and asynchronously and confirm its advantage for multi-component data. We conclude that the convolution-based three-station interferometry performs better because it makes better use of linear arrays with short-time recordings and retrieves higher-quality NCFs.