| Airborne electromagnetic survey is a geophysical exploration method which used the plane as the carrying platform and the electromagnetic induction theory of as the theoretical basis. The airborne electromagnetic survey method has great advantages, including the fast speed, low cost, good passage, and the large area coverage. Because the airborne electromagnetic detection equipment are carried on the mobile platform, the airborne electromagnetic data introduced the specific noise of the mobile platform and the background noise of aircraft, etc., except the human, environmental, geological and system noises. These noises affect the data quality and the detection ability of airborne electromagnetic system. Therefore, the complex multi-source noise suppression have the important practical significance for improving the detecting ability of airborne electromagnetic survey.This article rely on the subtopic of the national 863 Plan projects, “Study on the key technology on processing and mapping of podded time-domain helicopter airborne electromagnetic survey data”, and the subtopic of the national major scientific equipment projects, “airborne transient electromagnetic system data processing and inversion imaging”. For the effect on profile data of multi-source complex noise in airborne electromagnetic detection. The profile data is formed after preprocessing, including background field removing, stacking, and channel extracting. This paper studied the electromagnetic profile noise suppression methods based on the minimum noise fraction. The main research contents and conclusions of this paper are as follows:This paper analyzed the time-frequency characteristics of airborne electromagnetic profile data and the residual noise. To solve the problem that the residual noise in profile data is overlapping with the signal in both the time domain and frequency domain. The statistical analysis method is used to suppress profile residual noise.This paper introduced minimum noise fraction(MNF) into airborne electromagnetic profile noise suppression method study referenced the successful application on noise removal of multispectral data. Minimum noise fraction used the noise covariance for profile noise unitization and converted the electromagnetic profile data into MNF components arranged by SNR of MNF components. The L low-order MNF components with high SNR are used to reconstruct the profile data to suppress the residual noise. The unknown noise covariance estimation is one of the important calculations of minimum noise fraction. The noise removed through the adaptive width filter was used as noise estimation to calculate the noise covariance matrix is used in the paper. The results of simulation and field examples verified the validity of the minimum noise fraction to suppress the airborne electromagnetic profile noise. The signal to noise ratio of the latest 4-channel profiles is improved by 29.72 d B from the original data in the simulation example and the noise level reduced from ?60 n T/s to ?40 n T/s in the field result.Because of the MNF components with lower SNR in the L low-order reconstructed MNF components, this paper proposed the minimum noise fraction filter algorithm on the basis of minimum noise fraction to suppress the airborne electromagnetic profile noise. This algorithm reserved the first H(H<L) MNF components with enough high SNR, and processed the(L-H) MNF components with relatively low SNR using the wavelet threshold method to filtering the high-frequency spatial noise. Then the reserved MNF components and the filtered MNF components both participant the reconstruction of the minimum noise fraction. The method can not only suppress the noise in the electromagnetic profiles efficiently, but also can retain the anomaly information and avoid acting the detail characteristics of the data as noise completely. The results of simulation and field examples verified the validity of the algorithm to suppress the airborne electromagnetic profile noise. The signal to noise ratio of the latest 4-channel profiles is improved by 22.29 d B from the original data and by 4.37 d B from the result of minimum noise fraction in the simulation example. The noise level reduced from ?60 n T/s to ?25 n T/s in the field result.In the end, this paper studied the suppression method of multiplicative noise in the airborne electromagnetic profiles based on minimum noise fraction. The characteristics and the processing method of multiplicative noise in the airborne electromagnetic profiles were studied and the homomorphic logarithmic transformation was one of good methods to process the multiplicative noise. The homomorphic method transforms the multiplicative noise into additive noise and utilizing the minimum noise fraction to suppress the transformed noise in the logarithmic domain. The result of simulation example verified the validity of the minimum noise fraction to suppress the airborne electromagnetic profile multiplicative noise and the denoising result is even better than the adaptive width filter method. The signal to noise ratio is improved by 12.38 d B from the original data and by 2.77 d B from the result of the adaptive width filter method. The noise level reduced from ?25 n T/s to ?10 n T/s in the field result. |
PDF Full Text Request