Supervised stacking to improve the signal-to-noise ratio of AEM data

AEM method has proven to provide useful information on the subsurface for many applications. However, measured decays are affected by many noises, limiting its effectiveness and which may prevent to acquire usable data, especially in resistive environments. Stacking techniques are applied in an attempt to improve the signalto-noise ratio. However, stacking all decays falling within a stack interval can be ineffective, given the nature of noises that can affect the data from decay to decay. To a lesser extent, arbitrarily increasing the stack size may also be ineffective, especially in an anthropized environment. Stacking is generally done without any real control on the data taken into account. This paper introduces a supervised stacking method that stacks decays falling within a stack interval considering different combinations and estimates the signal-to-noise ratio of the resulting decays. The estimation of the signalto-noise ratio is performed using the singular value decomposition filtering which has proven to be effective in identifying and removing noise affecting an AEM dataset. The supervised stacking method is applied on the raw data. It has been tested on two AEM datasets, acquired in Reunion and Auvergne (France), where EM noise is high and resistivity can easily exceed 1000 ohm.m in some places. The results show that the presented method improves the signal-to-noise ratio and can reduce sferics and certain noises from man-made installations. It provides less noisy decays for post-processing and offers new possibilities for processing AEM data.