Research On Seismometer Self-Noise Test Method

This report focuses on test method research on seismometer self-noise and its’ quantitative analysis of many influence factors on seismometer self-noise test result. There are a large number of seismometers in network operation of the earthquake monitoring system in China Earthquake Administration. The performance of these seismometers is the key point of keeping the efficiency of the earthquake monitoring system. So it is important to do network access license quality test and periodical running quality check. Now seismometer quality test method is nonstandard and incomplete. The quantitative research result about seismometer quality test lack of enough experimental verification. The research on seismometer self-noise test method is one of the most important parts of seismometer quality examination. Until now, no full research on the seismometer self-noise test method have been done.Based on seismometer self-noise simulation, actual seismometer noise observation and data analysis, this report mainly focus on impact studies on seismometer self-noise power spectrum density(PSD) testing result which caused by many different factors. These influence factors include two identical seismometer transfer functions, two different seismometer transfer functions(different damping characters, two different output sensitivities, two different frequency ranges). different ambient noise level, different continuous observing time-length, different analysis data-length, different standard seismometers with different noise level, different seismic data recorders with different specifications. These research conclusions will be the foundation of seismometer self-noise test.First, this report analyses the principle of electronic feed-back seismometer. The relationship between seismometer transfer functions and the amplitude-frequency characteristic, phase-frequency characteristic was given. With the shake-table calibration method, we checked amplitude-frequency characteristic, linearity and cross sensitivity of seismometers that are used in the self-noise experimental observation. These quality testing results indicate that these seismometers are qualified with the designed technical specifications. Second, this report introduced two seismometer self-noise computational methods, two-cross correlation method and three-cross correlation method. And two-cross correlation simulation analysis module and three-cross correlation simulation analysis module were created with these two methods in MATLAB language. Then many different self-noise simulation experiments were calculated with these two simulation modules. In these simulation experiments, actual seismometer noise observation data were used as a seismometer self-noise simulation input signal and ambient noise input signal. We got following conclusions:1)With two same transfer functions, seismometer self-noise simulation output signal is same with seismometer self-noise simulation input signal.2)With two different damping characters, seismometer self-noise simulation output signal is near to seismometer self-noise simulation input signal.3)With different frequency bandwidths, seismometer self-noise simulation output signal is much bigger than seismometer self-noise simulation input signal.4)With different output sensitivities, seismometer self-noise simulation output signal is same as the seismometer simulation input signal.5)A high seismometer self-noise PSD will be got from seismometer noise observation data in high ambient noise. A low seismometer self-noise PSD will be got from seismometer noise observation data in low ambient noise. So we should set a seismometer monitoring system at a low noise seismic station.6)Three different continuous noise observation data of seismometer were used to do seismometer self-noise simulation. The lowest self-noise PSD of vertical seismometer was produced when continuous observation time achieves to10days. At the same time, the horizontal seismometer self-noise PSD became lower as the observation time became longer.7)We did three kinds of simulation computation while different noise level seismometers worked as the standard seismometer. When the standard seismometer and the measured seismometer have the same noise level and the same transfer function, the simulation output signal is same as the simulation input signal. Otherwise, the simulation signal is much bigger than the simulation input signal.8)In the three-cross correlation self-noise simulation, the seismometer simulation output self-noise signal is same as the simulation input self-noise signal. Third, four kinds of power spectrum calculation method were introduced in the fourth part of the report. Two-cross correlation self-noise calculation program and three-cross correlation self-noise calculation program were written in MATLAB language. We use two-cross correlation self-noise computation program to do many kinds of observation data calculation. The data sources in these computations are from two observational experiments. One part of data is seismometer’s noise data from Tianjin Jixian seismographic station. Another part of data is seismometer’s noise data from Beijing Xibozi seismographic station. We got the following conclusions:1)With two same transfer functions, three axial self-noise PSD of two seismometers are same. And the vertical seismometer self-noise PSD achieves lowest value.2)With two different transfer functions, the seismometer self-noise PSD is much bigger than its’ actual PSD.3) A big seismometer self-noise PSD will be from seismometer’s noise observation data in high ambient noise. A small seismometer self-noise PSD will be from seismometer’s noise observation data in low ambient noise. So we should do seismometer’s self-noise observation in a low noise seismographic station.4)The broadband seismometer self-noise PSD achieves the lowest value when continuous running time achieved10days and covered with a protective cover. The short-period seismometer self-noise PSD achieves the lowest value in less than10continuous running days; While the very broadband seismometer self-noise PSD achieves the lowest value in more than20continuous running days;5)Six different lengths of noise observation data from broadband seismometer were used to do self-noise computation. The results manifest that sixty-minute length data is enough for a perfect broadband seismometer self-noise result.6)In one experiment, three seismometers with different noise level were separately used as a standard seismometer to do broadband seismometer self-noise observation. Only with a identical noise level and identical transfer function standard seismometer can get the seismometer true self-noise value. Otherwise the seismometer self-noise PSD is much bigger than the true PSD value.7)The broadband seismometer self-noise PSD of0.06Hz～1Hz is lower than NLNM in the three-cross correlation method. This seismometer self-noise PSD is lower than PSD in the two-cross correlation method under the same experimental circumstance.Then, four kinds of characteristic of two seismic data recorders, noise, amplitude frequency characteristic, linearity, cross sensitivity, were test in the last part of the report. Two proposed methods, FFT spectrum analysis method and DFT spectrum data integration method, were put forward to automatically compute amplitude frequency calibration data. Finally we do some simulation research about the effect of seismic data recorders’noise and time synchronization error which is between channels on the seismometer self-noise PSD. The conclusions are as follow:The higher noise the seismic data recorder has, the bigger the seismometer self-noise PSD will be. The bigger time synchronization error the seismic data recorder has, the bigger the seismometer self-noise PSD will be.