基于相关建模检测的磁共振探水同频消噪方法

地面磁共振技术能够对地下水进行直接探测,具有定性定量的优点,是一种新兴的地球物理方法.然而,磁共振信号只有纳伏级,极其微弱,易受环境噪声干扰,尤其是具有拉莫尔频率的噪声,在时频域上均与信号重叠,难以有效去除,导致提取的信号参数准确度低、反演解释误差较大.本文针对同频噪声干扰问题,提出了相关建模检测(CMDT)方法,通过相关方法实现频谱迁移和低通滤波,结合信号和噪声特征建立数学模型,采用模型变换实现同频噪声的抑制,并利用最小二乘指数拟合方法提取高精度SNMR信号.为了对新方法进行定量分析,以验证其效果,对含有不同幅度的同频噪声和磁共振信号进行仿真实验,实验结果表明在信噪比为-31.17 dB的情况下,所有参数的最大提取误差不大于1.22%,验证了新方法能够在压制同频噪声的同时提取出高精度信号参数.为了模拟野外情况,在同频噪声和信号数据中加入随机噪声进行实验,结果表明当信噪比大于-10.12 dB时,CMDT方法仍可以获取有效的信号.因此,本文的研究为处理含有同频噪声干扰的实际SNMR信号数据提供了理论依据,为后期高精度反演提供了技术支撑.

Co-frequency harmonic attenuation from SNMR data based on correlation modeling detection technology

Surface nuclear magnetic resonance technology (SNMR) is a novel geophysical method that is used to detect groundwater directly without invasion. This method features both qualitative and quantitative advantages. However, the SNMR signal is extremely weak, and its magnitude may be lower than a few nonavolts. As a consequence, the signal is susceptible to ambient noise, especially the interference with the Larmor frequency, which overlaps with the signal in both time and frequency domains and is difficult to remove. In this case, errors will be introduced to the parameter extraction and subsequent inversion interpretation. To address this issue, we propose a correlation modeling detection technology (CMDT) to suppress the co-frequency noise. Specifically, we implement spectrum shift and low-pass filtering to the signal and establish a mathematical model referring to the signal and noise characteristics. The co-frequency harmonic is then removed by model transformation. Furthermore, we utilize least square fitting to extract high-precision SNMR signals. To validate the feasible of the proposed method, the synthetic signals are imposed on the co-frequency harmonics with different amplitudes. The experimental results indicate that, under the circumstance when the signal-to-noise ratio (SNR) is -31.17 dB, we can still extract accurate parameters after eliminating the co-frequency harmonic with an error not exceeding 1.22%. Random noise is then added to assess the performance of the method in the field. The results imply that the CMDT method can still obtain a precise signal when the SNR is greater than -10.12 dB. Therefore, the findings of this study provide a theoretical basis for processing the SNMR signal data with the co-frequency interference and technical support for the subsequent high-precision inversion.