| Nuclear magnetic resonance (NMR) technology to detect underground water is a new method of geophysics, which is the only one can directly explore the aquifer. The artificially excited electromagnetic field is used, so that the hydrogen nucleuses of groundwater form into a macroscopic magnetic moment. In geomagnetic field, the precession movement is caused by the macroscopic magnetic moment. The electromagnetic signals generated from the precession movement are collected by coil, then whether the groundwater existed or not can be detected. In comparison with the traditional method of geophysical exploration for groundwater, MRS method has advantages with high-resolution, high efficiency, rich information and uniqueness of solution. The use of nuclear magnetic resonance detection system for underground water can be efficiently carried out the regional hydro-geological survey, determine the long-term water areas, delineate the groundwater in the distribution of three-dimensional space, and reliably locate the selected wells. The amplitude of Nuclear magnetic resonance response (MRS) signal is directly proportional to water content in the exploration space, however, due to nuclear magnetic resonance signal generated by hydrogen in water is of very low amplitude (of the order of nanovolts), and is very easily disturbed by the presence of electromagnetic noise of industrical or natural origin. It leads that the signal to noise ratio of collected signal is low, and cannot obtain accurate interpretation result. In this paper, a number of new data-processing methods have been studied to improve signal to noise ratio of the MRS signal, and respectively, for different types of noise, the de-noising strategies have been introduced, and through application of a field experiment with comparison of pumping drilling to verify that the proposed methods in this paper are effective and accurate.In this paper, the data-processing software is designed for JLMRS groundwater collection, which collects low signal to noise ratio MRS signal and various types of noise. Data-processing flow includes de-noising of MRS signal processing, MRS signal parameter extraction and hydro-geological parameters calculation is proposed.Stacking and filtering method are including in de-noising technology for MRS signal. Usually the signal-noise ratio is improved by using a simple stacking method, but it is difficult to eliminate it when there is spiky noise, nor does it apply to where the noise level is instability. In this paper, to solve these problems, three stacking methods are interposed: threshold stacking, weighted stacking and statistics stacking. Threshold stacking is applied to the less spiky noise; weighted stacking is applied to the instable noise level; statistics stacking is applied to both over spiky noise and instable noise level. Besides that, signal to noise ratio is improved to the best, number of stacking required is the least in statistics stacking. So measuring efficiency of MRS groundwater detection system can be improved. For different types of noise, three methods of filtering methods in this paper are present, such as ripple approximation low-pass FIR filter, moving average (MA) filter and the adaptive notch filter. For random noise, the signal to noise ratio is improved more using ripple approximation FIR low-pass filter than the moving average (MA) filter, but MA filtering algorithm is simple and easy to implement, and the dynamic response is small, less data will be lost. For the power-line harmonic noise, the adaptive notch filter can automatically track the frequency of power-line harmonics, and the transition zone is steeper. Three stacking methods and three types of filtering method are combined to three de-noising strategies.Strategy 1: Threshold Stacking + (Adaptive Notch Filter) + Moving Average FilterStrategy 2: Weighted Stacking + (Adaptive Notch Filter) + Low-pass FilterStrategy 3: Statistics Stacking + (Adaptive Notch Filter) + Low-pass FilterWhen the environmental noise level is stability, and spiky noise is less, the signal to noise ratio is high after a simple stacking, then strategy 1 may be used, that is the most easy-to-use algorithm. When the environmental noise level is instability, but noise spiky is less, the signal to noise ratio after a simple stacking is low, then the strategy 2 may be used, which effectively improve the signal to noise ratio. When the environmental noise level is instability, and noise spiky is much, the signal to noise ratio after simple stacking is still low, strategy 3 must be used, which can achieve the best signal to noise ratio improvement, while improving the measurement efficiency.Two MRS signal parameter extraction methods are proposed in this paper: linear fitting and nonlinear fitting. Compared through simulation under different signal to noise ratio of the fitting error, the result is that when signal to noise ratio is higher, the two methods both are smaller in fitting error; but when signal to noise ratio is low, non-linear fitting is superior to linear fitting in error. After MRS signal parameter extraction four key parameters are obtained: the initial amplitude, the average decay time, the initial phase and the received frequency. From this we can calculate the depth, thickness and water content of aquifer, at the same time to estimate the permeability and hydraulic conductivity of water, which are used to determine the water capacity of the quantitative reference.Finally, a measurement example is given in the paper in the northwest of Shaoguo Town in Changchun City. It shows that the use of the method proposed in this paper can effectively improve signal to noise ratio of MRS signal, obtain the credible aquifer vertical distribution results. These proved that this data processing is effective. Contrast with pumping and drilling results at the same location, the design of MRS groundwater detection data-processing software is verified to accurate.
|
PDF Full Text Request