Abstract:

The self-noise levels of different seismometer models directly affect the quality of seismic observations and further limit earth science research based on seismic records. We conducted a test of the self-noise level of seven types of seismometers at the Malingshan seismic station, in which the instrument types included short-period, broadband, very-broadband and ultra-broadband. Three seismometers of each type were set up, and the observation period was from November 22, 2018, to March 26, 2019. In this paper, based on continuous seismic waveforms from seven models of seismometers, the self-noise power spectral density(PSD)of the seismometers was calculated by using the three sensor correlation analysis method, and the probability density distribution of the self-noise PSD of the seven models of seismometers was obtained by using the PDF representation. Based on the mode values of the PDF distribution, the self-noise models of the three channels(UD, EW and NS)of the seven models of seismometers are given respectively.

For the ultra-broadband seismometer CMG-3T-360, in the microseism band(0.1Hz to 1Hz), the self-noise of the horizontal components(EW and NS)is higher than that of the vertical components(UD)and is consistent with the trend of the seismic noise, which may be attributed to misalignment of the horizontal direction between seismometers. In the low frequency band(<0.03Hz), the self-noise level of the horizontal component is higher than that of the vertical component, and small changes in the barometric pressure may lead to higher incoherent noise in the horizontal direction of the sensor at long periods. Compared with the vertical direction, the horizontal direction of the seismometer is more susceptible to air disturbances. At a frequency of 0.005Hz, the instrument self-noise of the horizontal component is close to the seismic background noise, and the instrument self-noise of the horizontal component is the main source of noise recording. Installing a heat and wind shield can effectively reduce the seismometer self-noise in the low frequency band. When using the CMG-3T-360 to observe long-period seismic signals, a shield with both thermal insulation and windproof function is required.

The self-noise level of the short-period seismometer JS-S02 is lower than that of TDV-33S and lower than that of the New Low Noise Model(NLNM)between 0.15Hz and 7Hz. In the UD and EW channels, the self-noise level of TDV-33S is lower than the NLNM model between 0.17Hz and 0.5Hz. The higher instrument self-noise further limits the extraction of long-period seismic signals in the digital recordings of short-period seismometers.

For the broadband seismometer TDV-60B and the very broadband seismometer TDV-120VB, the self-noise levels are basically consistent in the vertical direction and also are higher than those of the broadband seismometer JS-60 and the very broadband seismometer JS-120. In the horizontal direction, the two self-noise levels in the microseism band and the low frequency band show different characteristics, i.e., the self-noise levels of TDV-60B are lower than those of TDV-120VB in the microseism band and higher than those of TDV-120VB in the low frequency band. When the frequency is lower than 0.03Hz, the self-noise levels of TDV-60B and TDV-120VB on the horizontal channels are close to those of JS-60 and JS-120, respectively.

For JS-60 and JS-120, in the vertical direction, the self-noise levels of both are close to CMG-3T-360 in the microseism band. The self-noise level of JS-120 on the vertical channel is lower than 5dB away from CMG-3T-360 in the low frequency band and lies within the 68%confidence interval of the PSD; in the high frequency band(>2Hz), it is higher than CMG-3T-360 confidence interval of the PSD. In the horizontal direction, the self-noise levels of JS-60 and JS-120 are lower than those of CMG-3T-360 between 0.15Hz and 1Hz and in the microseism band, respectively. The self-noise levels of JS-60 on the horizontal channel show good agreement with those of CMG-3T-360 in the high frequency band. The self-noise of JS-120 on NS channel is higher than CMG-3T-360 confidence interval in the low frequency band. When extracting long-period seismic signals, a seismometer with lower noise level in the low frequency band should be selected as much as possible.

Key words: Seismometer, Self-noise, Power Spectral Density, Probability Density Function

摘要：

不同型号地震计的自噪声水平直接影响了地震观测数据的质量, 并进一步限制了利用地震数据解决地球科学问题的能力。长久以来, 由于受到观测条件的限制, 准确地测量和比较不同型号地震计的自噪声水平颇具挑战。文中利用马陵山地震台4个月的连续地震波形, 基于概率密度函数的表示方法计算了7个型号地震计的自噪声功率谱密度曲线。对于超宽频带地震计CMG-3T-360, 在微震频带(0.1~1Hz), 水平方向的自噪声明显高于垂直方向, 这可能是地震计水平方向相对方位未对齐导致的计算偏差; 在低频段(<0.03Hz), 水平方向显著偏高的自噪声可能源于大气压的变化。短周期地震计JS-S02的自噪声水平在频率为0.15~7Hz时低于全球新低噪声模型(NLNM)。宽频带地震计TDV-60B和甚宽频带地震计TDV-120VB在垂直方向的自噪声水平基本一致。宽频带地震计JS-60和甚宽频带地震计JS-120的自噪声水平在微震频段接近或低于CMG-3T-360。当频率为0.008~0.08Hz时, JS-120水平向NS通道的自噪声水平高于CMG-3T-360功率谱密度68%的置信区间。

关键词: 地震计, 自噪声, 功率谱密度, 概率密度函数