Magnetotelluric Noise Attenuation Using a Deep Residual Shrinkage Network

Magnetotelluric (MT) surveying is an essential geophysical method for mapping subsurface electrical conductivity structures. The MT signal susceptible to cultural noise, and the intensity of noise growing with urbanization. Cultural increasingly difficult be removed by conventional data processing methods. We propose a novel time-series editing based on deep residual shrinkage network (DRSN) address this issue. Firstly, are divided into small segments form dataset system. Secondly, we use system train denoising model. Finally, trained model used denoising. experiments using synthetic actual field collected in Qinghai Luzong, China, show that DRSN can effectively remove has better adaptability efficiency than traditional