Noise removal in MRS applications: field cases and filtering strategies

Abstract : The usefulness and reliability of magnetic resonance information to characterize water bearing geological structures has been widely demonstrated these last two decades all over the world and many future applications just begin. The main limitation of MRS applicability is its sensitivity to the electromagnetic noise which results in a long and site dependent measuring duration, and generally prove to be impossible in urban conditions. Many improvements have been performed all along the development of MRS technology. Nowadays, numerous mono and multi-channel processing schemes have been published, but efficiency remains site and time dependent. We have reviewed data from various contexts and compared the noise removal efficiency and impact of the filtering on synthetic signal added to real noise data. We also used methods derived from magneto-telluric to study the structure of the noise and present a continuous EM field monitoring during a storm event in mountain where we performed a MRS survey. We observed that the reconstruction of natural noise is a percentage of the ambient noise, the ratio is almost stable. Despite this observation of stable removal performance, it means that when the level of noise is multiplied by 10 to 100 and more… it is better to stop measuring MRS and wait for a quiet period of time.