Application of digital filters in the processing of magnetotelluric survey data

The Fast Fourier Transform (FFT) is widely used in the processing of magnetotelluric (MT) survey data. When the signal to noise ratio is high and the length of time series is long enough, the FFT processing gives us optimum electromagnetic response function. However, these conditions may fail from time to time. We must improve the data processing of MT data in order to obtain optimum response function independent on environmental conditions of exploration. One of the essential techniques in data processing is the digital filters. Various digital filters are developed and applied to the seismic data, and it is proven that applying digital filters is effective. On the other hand, digital filters have been rarely applied to MT data. In this study, we elucidate how effectively the application of digital filters improves the electrical response function. We focus on a digital filter called pole-on-pedestal (PP) which extracts or amplifies signals at a specific frequency. The PP filter is defined as the Z transform. It distorts the Fourier spectrum less than the FFT. We apply this filter to marine MT survey data. The data used in this study was recorded at the sea bottom near the Daitou islands, on the Philippine Sea plate, by OBEM (Ocean Bottom Electro Magnetometer). A very large magnetic storm occurred in the recorded term. A correlation between the orthogonal electric and magnetic fields can be observed in the raw time-series. When the PP filter extracts a signal at a specific frequency from each electric or magnetic field, there exists the strong correlation between the orthogonal filtered electric and magnetic fields. Next, we calculate the MT response function (e.g. apparent resistivity or phase). The phase difference between each electric and magnetic field should be treated, so the imaginary part of each field is important. We use the Hilbert transform in order to calculate imaginary part. The calculated apparent resistivity curves show very smooth changes, implying the successful estimations. The result demonstrates that the data processing of this study is accurate enough. We conclude that the application of PP filter in the data processing of MT survey data could be an alternative method to the other existing FFT based ones.