Motion noise remove for secondary-field semi-airborne transient electromagnetic data

Semi-airborne Transient Electromagnetics (SATEM) is a low-cost and productive method for quick geophysical surveys. The system includes a grounded line source on the ground and a receiver coil hanging in the air by a drone. Motion noise usually has the biggest impact on the acquired data because the receiver coil, which is suspended several meters below the drone by a soft rope, would maintain continuing random motion due to the movement of the drone, wind, etc. It also is a common problem in airborne electromagnetic (AEM) surveys, which typically use fixed-wing aircraft or helicopters. However, due to the secondary-field acquisition of SATEM, there are data deviations between the start and end of each motion period caused by continuous coil motion and discontinuous data acquisition. And this also brings difficulty in baseline elimination by using most de-noising methods, which require time or signal continuity on the baseline and are widely applied in full-time acquisition data (full-wave data). We proposed a simple but useful single-period polynomial-fitting method to remove motion noise from secondary-field data. And to prevent the obtained polynomial from being distorted from baseline by valid signals, a haar-based wavelet was employed to automatically recognize the start-point of pulses in the original signal so that the pulse can be eliminated before the polynomial-fitting process. Through a field SATEM survey data test, we found that the polynomial-fitting with 5 parameters (4th order) is good enough to describe the motion characteristics. We compared the results with those of the most commonly used wavelet methods at different bases and levels. The comparisons of de-noising results and apparent resistivity profiles of the original signal, data after wavelet de-nosing, and data after the polynomial-fitting de-noising show that this method is superior and much more reliable than wavelet de-noising of secondary-field SATEM data.