INTRODUCTION
The reliability of low frequencies is a crucial question for geophone records. An example is the data recorded for passive seismic studies using geophones paired with TEXAN miniature recorders, where the high resonance frequency of the geophones degrades the low-frequency waveforms and increases the ambiguity in phase picking. We have evaluated the reliability of such data from a mobile array of geophone- TEXAN pairs in central China. The frequency response of the geophones can be easily established using the Power Spectral Density Ratio (PSDR) between records of the geophone- TEXAN pairs and a broadband seismometer occupying the same sites. This method allows a quantification of the retrievable frequencies of short-period seismic data and a reliable spectral extension. For instance, the retrievable frequency band of the data recorded by 4.5-Hz geophones can be extended down to 0.3 Hz for regional M2.0 earthquakes in our study area, and down to 0.04 Hz for surface wave of a M6.2 teleseismic event in Indonesia. After applying an inverse filter within the retrievable frequency band, the quality of the data is improved significantly and matched well with the records of a nearby broadband permanent station. The new method is useful for assessing and extracting low-frequency information from geophone data.
Geophones are widely used in both exploration seismology and solid earth seismology with a reliable quality and reasonable cost. The resonance frequencies of such short-period seismic sensors are typically around 10 Hz, and some are 4.5 Hz and 1 Hz. Those frequency ranges are sufficient for most activesource studies of sedimentary basins and upper crust where the main frequency is higher than 10 Hz. However, the targeted frequencies can be much lower for crustal-scale studies or monitoring microseismics. During the past decades the geophones have been used to study the regional seismic velocity structure using both active or passive sources (Nielsen and Thybo, 2009; Sroda, Czuba, and Grad, et al. 2006; Majdanski, Kozlovskaya, and Grad, et al., 2007; Malinowski, 2009; Nielsen and Thybo, 2009). In those studies only the high frequency information, such as 2 ~ 5 Hz (Sroda, Czuba, and Grad, etal. 2006), is used for the regional seismic structure study. This frequency range is insufficient for studying regional structure or low-frequency seismic sources. The analysis of the surface wave response of the regional quarry blast sources will need the frequency band from 0.2 Hz to 40 Hz (Yao and Dorman, 1992), and the source mechanism study needs the frequency band from 0.5 Hz to 2 Hz (Tan and Helmberger, 2007). Most importantly, acquiring the low frequencies is the key practical way to improve the frequency bandwidth and therefore seismic resolution (Knapp, 1990). The two practical ways to extend the frequency bandwidth of seismic data are: improving the low-frequency response of the sensor and balancing the low frequencies during the processing. The former can be achieved through adding better types of circuits to extending the frequency response (Barzilai, 2002; Webb, Deaton, and Lemire, 2001). But those processes are too complex for practical acquisition.
