Surface wave dispersion curves are extensively used to map the lithospheric shear‐wave velocity structures. However, traditional methods have difficulty extracting overtone dispersion curves from both ambient noise and seismic event data. Recently, Wang et al. (2019, https://doi.org/10.1029/2018JB016595) proposed a new array‐based method, the frequency‐Bessel transform (F‐J) method, to extract high frequency overtones from ambient noise data. However, because the Green's functions of earthquakes are complex, the F‐J method cannot be directly applied to extract overtones from earthquake event data. In this paper, we analyze the differences between ambient noise and earthquake event data and verify that the F‐J method can be efficiently applied to event data by controlling the azimuthal range and adopting time windows. For the F‐J method using earthquake records, the time windows calculated by group velocity intervals can improve the overtone extraction efficiency. We named this application of the F‐J method with time windows using event data the multiwindow F‐J (MWFJ) method to distinguish the two techniques. Further verifications were performed using real data from three earthquakes recorded by the U.S. Transportable Array. Compared to traditional methods for extracting overtones from earthquakes, the MWFJ method can effectively extract higher‐frequency (0.02–0.4 Hz, for the data used in this work) and higher‐resolution overtones from event data. Moreover, the overtones extracted from earthquakes are more sensitive to deeper structures than those extracted from ambient noise. The complementarity of the F‐J and MWFJ methods for the extraction of overtones indicates that better imaging results may be obtained by combining these two methods.

In the past two decades, seismic ambient-noise cross correlation (CC) has been one of the most important technologies in seismology. Usually, only the fundamental-mode surface-wave dispersion was extracted from the ambient noise. Recently, with the frequency–Bessel transform (F-J) method, overtone dispersion can also be extracted from the ambient noise and it adds significant value in inversion. This method has also been verified to be effective for array seismic records of earthquake events. In this article, we describe our algorithm and a Python package called CC-FJpy. For the F-J method, we use the Nvidia’s graphics processing unit to accelerate the computation, which can achieve a 100-fold computational efficiency. We have encapsulated our experiences and technologies into CC-FJpy and tested the CC-FJpy by ambient-noise and earthquake data to ensure its speed and ease of use. Our open-source package CC-FJpy can benefit the development of surface-wave studies using ambient noise and make it easier to start with high-mode surface waves.

With the advent of the ambient noise cross-correlation technique (Campillo & Paul, 2003;Shapiro & Campillo, 2004), surface wave imaging was freed from its dependence on the occurrence of earthquakes; as a consequence, surface wave imaging has undergone substantial development in the past two decades (e.g., Shapiro et al., 2005;Yao et al., 2006). By calculating the cross-correlation function, body wave impulse responses (which are much weaker than surface waves) can also be retrieved from earthquake coda waves and continuous ambient noise data (e.g.,

Rayleigh and Love wave dispersion curves are known as normal modes and are utilized for inversion. These two waves arrive after the S wave and account for the main energy in all waveforms. However, weaker signals arriving between the P and S waves also have dispersive properties, which are controlled by so-called leaking modes.Studies of leaking mode can be traced back to Somville (1930), who identified and named the PL phase. The PL phase is a train of long-period (>10 s) dispersed waveforms that arrive shortly after the P wave. Oliver and Major (1960) measured PL phase dispersion curves and linked the PL phase with the leaking mode of the crust-mantle waveguide. In addition to the PL phase, some other phases are also considered to have the same wave behavior as the PL phase, which can be explained by the leaking mode (e.g., Pg phase, Shaw & Orcutt, 1984; W phase, Furumura & Kennett, 2018). The leaking mode describes waveguides (leaking waves) whose radiation energy leaks into half-space and phase velocities are higher than the maximum shear velocity (Phinney, 1961). Subsequently, studies have mainly focused on how to calculate leaking modes and simulate leaking waves through leaking modes (e.g., Gilbert, 1964;Haddon, 1984Haddon, , 1986Watson, 1972). However, due to the complexity of leaking modes, limited research has been conducted on imaging underground structures by using the dispersion of leaking waves, and most of the extracted dispersion is below 0.1 Hz for lithospheric research (e.g.,

Ambient noise technology (Sabra et al., 2005;Shapiro et al., 2005) uses cross-correlation to extract the empirical Green's function between station pairs and has become a common tool to obtain underground structures. It is widely used in continental (