Previous work in frozen soils in North Dakota using Time Domain Reflectometry (TDR) has shown that significant amounts of unfrozen water remain in the soil at temperatures to -15 oC. Water transport in the vapor phase to the freezing-front in soils is an important transport phenomenon, but little is known about the importance of liquid water in water and solute transport in frozen soils. An important step in determining the liquid phase transport of water is to quantify how much water remains unfrozen and what effect, if any, initial water content before freezing has on the lower limit of unfrozen water content. The objectives of our study were to validate the amount of unfrozen water in frozen soil as determined by TDR using Nuclear Magnetic Resonance (NMR) and to determine the effect of soil texture and initial water content on the lower limit of unfrozen water content. Three soil textures consisting of a loamy sand (0.10, 0.20, and 0.30 m3/m3), a loam (0.20, 0.30, and 0.40 m3/m3), and a silty clay (0.20, 0.30, and 0.40 m3/m3) were frozen at three different initial water contents. Three replicates using TDR and four replicates using NMR of each soil at each water content were frozen to -15 oC and -26 oC, respectively. Each soil texture was found to have a much different freezing characteristic curve and similar shaped curves were measured by both TDR and NMR. Similar unfrozen water contents at -15 oC were found for the loam and the silty clay for both methods. However, the lower limit unfrozen water content in the loamy sand increased linearly with initial water content as measured with TDR (0.05 - 0.09 m3/m3) but remained constant (0.03 m3/m3) using NMR to measure the liquid water. Apparently the mixture of ice, air and water in the loamy sand has changed the dielectric properties of this soil enough to introduce error in the way water content is calculated in the TDR method. At -26 oC and an initial water content of 0.32 m3/m3, lower limit water contents as determined by NMR were 0.03, 0.09 and 0.11 m3/m3 for the loamy sand, loam and silty clay, respectively. Lower limit water contents increased by 0.02 m3/m3 for the two finer textured soil materials as the initial water content was increased from 0.32 to 0.44 m3/m3. TDR was found to overestimate unfrozen water at temperatures lower than -5 oC. If TDR is to be used successfully in frozen soils to estimate unfrozen water, refinements in calibration models must be made to correlate mixtures of ice, air and liquid water to the dielectric property of bulk soil. In order to predict the appropriate mixtures, freezing characteristic curves should be used to estimate the amount of liquid water at a given soil temperature and texture.

Ray Knighton
Dept. of Soil Science
North Dakota State Univ.
Fargo, ND  58105
Phone: (701) 231-8577
Fax: (701) 231-7861
E-mail: knighton@badlands.nodak.edu