关键词: 大容量气枪, 分布式光纤声波传感, 台阵技术

Abstract: The large-volume airgun system was introduced to excite highly repetitive seismic signal for medium change monitoring. Using dense seismic array to record the seismic wavefield will be helpful to high spatial resolution time-lapse tomography. However, most dense arrays employ the nodal short-period geophone with built-in battery that is not suitable for permanent monitoring. The novel distributed acoustic sensing(DAS)technology uses fiber-optic itself as sensor that providing small station spacing. The incident seismic wavefiled induces tiny strain of the fiber-optic that leads to phase change of the Rayleigh backscattered optical signal. Therefore, measuring the phase difference between two signals scattered at two nearby scatterers can be used to recover seismic signal. Since the scatter is randomly distributed in the fiber-optic, it is possible to record seismic wavefield with spacing down to sub-meters. Each optical signal is processed in the interrogator. Therefore, the DAS array is easily maintained as a permanent dense array for seismic monitoring. We conducted a pilot experiment to test feasibility of using DAS array to record airgun signal in Binchuan, Yunnan Province.
The Binchuan Fixed Airgun Signal Transmission Station built in 2011 is the first inland large-volume airgun in China. The airgun system consists of four Bolt LL 1500 airguns and fires at 10m depth in a reservoir. The seismic energy released by one airgun shot is close to the one of M_L0.7 earthquake. During this pilot experiment, the airgun was continuously shot after midnight with an interval of 15 minute. The DAS array is a micro-structured fiber-optic buried in an “L-shape” trench, which is about 9.8km away from the airgun. To enhance SNR of the optical signal used for recover seismic signal, a series of ultra-weak fiber Bragg gratings were built in the fiber with 2m spacing. The 180m fiber-optic is buried at about 20cm depth and the trench is backfilled with sand. The channel spacing is 4m and the interrogator continuously records at 1 000Hz.
The signal is barely visible on the record of single shot due to strong ambient seismic noise and optical noise. Since the seismic signal excited by the airgun is highly repetitive, we used the time-frequency phase weighted stacking method to stack records of multiple shots. The signals clearly emerge on the stack traces and the arrival time agrees well with the records of a co-located seismometer. Compared with the seismometer's record, the DAS records concentrate in a higher frequency band(5~8Hz). Since the DAS and seismometer record the seismic wavefield in dynamic strain and particle velocity, respectively, the frequency-wavenumber scaling algorithm was used to convert DAS's strain record to particle velocity record that shows clear phase difference from seismometer's record. The difference between records of DAS and seismometer was analyzed in time-frequency domain. The largest difference occurs between 3 and 6Hz in the airgun signal wave train, which may due to lower sensitivity in lower frequency band of DAS.
The bootstrapping resample method was used to evaluate the stacking converge rate of two datasets. Comparing to the reference trace that is stacked with 24 shots, the cross-correlation coefficient reaches 0.9 with only four shots for the seismometer dataset. At the meantime, the cross-correlation coefficient is only 0.8 with 20 shots for the DAS dataset. To improve the stacking efficiency, we also tried the array stacking method. The records of 26 channels on the X lag of the array were stacked. The one-shot stacking suppressed the traffic noise from a nearby street and the airgun signal clearly emerges on the one-shot stacking trace. The airgun signals on the stacking traces of multiple shots and multiple channels are comparable, which suggest the multiple channels stacking can be used to improve time resolution for time-lapse tomography/monitoring.
In summary, the airgun signal is successfully recorded by a DAS array with an engineered fiber-optic cable. Comparing with the seismometer, DAS dataset is strongly affected by the traffic noise and lower sensitive to lower frequency band. The dense spacing also provides opportunity to stack multiple channels’ records that improves SNR of airgun signal. Since the lack of reliable vertical component records, the phase identification cannot be done via particle motion analysis. The aperture of our DAS array is too small to estimate the apparent velocity to identify seismic phase too. In the future, it is worth to use telecom fiber-optic cables as sensor for time-lapse tomography, which have been widely deployed in urban area and significantly reduced deployment cost.
The clear variation of waveforms across one lag arising from un-uniform coupling was also observed. To comprehensively evaluate the monitor capability, it is important to deploy large aperture DAS array for seismic signal attenuation analysis. Our result suggests that the stronger lower frequency system noise of the DAS integrator reduces the sensitivity to seismic signal. More attention should be paid to approaches such as environmental vibration isolation and optical noise reduction. Another issue is accurate response function. Calibration with co-located seismometers and numerical modeling are helpful to provide accurate sensitivity and response function, which is important in seismology studies.

Key words: airgun, distributed acoustic sensing, array processing