强人文干扰环境的电磁数据小波去噪方法研究

大地电磁法(MT)以成本低廉,探测深度大、水平方向分辨能力高等优点,在矿产资源勘探方面得到广泛应用.然而,在老矿区或者矿区周围进行二次探矿时,强人文干扰严重影响观测数据的质量,导致反演结果出现偏差,甚至出现错误的解释结果.因此,需要对观测数据进行降噪处理.结合多分辨率分析算法和小波阈值算法的特点,本文提出了综合小波算法:采用db3小波基;基于多分辨率分析算法,去除长周期噪声;基于小波阈值算法,将Bayes估计配合改进型阈值函数去除短周期噪声干扰.对实测数据的处理结果显示,处理后的数据的时间序列以及视电阻率曲线质量都有了明显的改善,近源效应得到有效的抑制.

A combined wavelet transform algorithm used for de-noising magnetotellurics data in the strong human noise

Magnetotellurics (MT) is normally applied for mineral exploration, because of its low cost and deeper penetration, as well as high resolving power in horizontally. However, the quality of MT data is affected by the human noise seriously around the mine area, which result in a unreliable result of inversion and even a wrong explanation. Therefore, it is important to remove the various disturbance. The multi-resolution algorithm (MRA) has domination in terms of making a high resolution for the frequency domain of MT data, and the wavelet thresholding algorithm (WTA) is at an advantage when removing the high frequency noise. A combined wavelet transform algorithm based on MRA and WTA is proposed. db3 wavelet is adopted in processing. Baseline drift and periodic noise of square wave can be removed by multi-resolution algorithm. Bayes estimation combined with improved threshold function which based on algorithm of wavelet threshold value is used to eliminate the impulse noise, triangular waveform as well as other forms of high frequency noises. Processing case indicated that the quality of the processed data, including the time series data and apparent resistivity curves, is improved significantly. Meanwhile near-source effect, which is caused by human interference, is suppressed effectively.