Study on eliminating noises in short-period data of geomagnetic intensity by using numerical multichannel predictive filtering

The effectiveness of eliminating the noises in short-period data of geomagnetic intensity recorded in a little seismo-geomagnetic array by using numerical multichannel predictive filtering has been studied. The result shows that this technique is effective to fit external magnetic disturbance to inner electromagnetic induced difference field and reduce the noise level of difference data successfully. The filter quality factor Q of two examples in this work are 0.86 and 0.68 respectively. The spectral analysis shows that during geomagnetic-calm days the fourfold-frequency harmonics of S _q in difference data are main components. The length of the optimum filter depends on not only the frequency of predicable energy in difference data but also maybe the phase difference between input and expected output data. It is difficult to obtain the filter fitting both the data during magnetic-disturbed days and calm days. The result shows that the conductivity in Yanqing-Huailai basin west to Beijing may be much non-uniform. Contribution No. 98A01013, Institute of Geophysics, China Seismological Bureau, China. This paper is supported by the Chinese Joint Seismological Science Foundation.