Study on Surface Nuclear Magnetic Resonance Inversion

摘要

Surface nuclear magnetic resonance is widely used for groundwater exploration and aquifercharacterization. Inversion of surface nuclear magnetic resonance provides important information about aquifers, such as their depths, thickness and water content. In this paper, we simply introduced the basic principles of nuclear magnetic resonance firstly, and then we discussed the relationship between surface nuclear magnetic resonance one dimensional kernel and pulse moment, medium conductivity, depth. It was found that surface nuclear magnetic resonance maximum depth is superimposed for one value of pulse moment and has smaller maximum depth in the higher conductive medium. A water content of layers lying below highly conductive layer is severely underestimated. It was also found that the condition number of one dimensional kernel matrix is small in the insulating medium. We improved simulated annealing algorithm from parameter restrictive condition, model disturbance, adaptive optimization method, circulate method, objective function, Markov chain and so on. Finally, we inverted MRS data using singular value decomposition and improved simulated annealing algorithm.The experimental result of synthetic data shows inverting MRS data using SVD in insulating medium and improved simulated annealing algorithm are desirable.