Comparative study of unfrozen water content measurement principles and calculation methods based on NMR

Unfrozen water content is a key concern in frozen soil. The measurement of unfrozen water using nuclear magnetic resonance (NMR) has gained significant popularity. However, the extraction methods of unfrozen water content from NMR data are still lack of a comprehensive evaluation. To overcome this challenge, a constant unfrozen water test (three representative soils, i.e., silty clay, bentonite clay, and silt sand, are selected) is proposed and used to study on the influence of temperature on NMR signal. Further, three extraction methods of unfrozen water content, such as the Curie law method (CLM), paramagnetic regression line (PRL) method, and resistivity-temperature method (RTM), are evaluated. The results demonstrate that: (1) As a theoretical method, CLM is the most convenient but with the highest error, as an average water content error of 0.74%. (2) PRL requires four calibration points and has an average water content error of 0.28%. (3) RTM needs a special calibration curve and yields the smallest water content error of 0.07%. Overall, RTM with a pre-calibrated lambda is recommended to obtain higher precision and PRL can be employed as a convenience choice.