In-situ estimation of unsaturated hydraulic conductivity in freezing soil using improved field data and inverse numerical modeling

Hydraulic property determination in freezing soils continues to be a substantial challenge. Our overall objective was to present a novel method for in-situ estimation of unsaturated hydraulic conductivity in freezing soil using a combination of improved field data and a simplified inverse modeling approach. Dielectric sensor readings in the field were corrected using knowledge of ice permittivity, temperature and water redistribution to achieve higher accuracy in determination of liquid soil water content and soil ice content. The equation describing unsaturated hydraulic conductivity in freezing soil with an added impedance parameter was inversely estimated by firstly fitting soil freezing and thawing characteristic curves using the measured liquid soil water content and soil temperature, and then minimizing differences of the measured and simulated total soil water content with a coupled heat- and water-transfer model in frozen soil. Field measurements were made over two winters between 2011 and 2013 (year-1: winter of 2011-2012; year-2: winter of 2012-2013) in Beijing, China. Results suggested that (i) the relative error of liquid soil water content measurement in freezing soil was up to 53% if the knowledge of ice permittivity, temperature and water redistribution was neglected; (ii) the estimated impedance parameter in year-1 (1.152) was an order of magnitude higher than in year-2 (0.117), possibly because a faster freezing rate generated more fine ice in year-2, resulting in reduced tortuosity; (iii) the estimated impedance parameter uncertainty likely comes from the model assumptions, the measurement accuracies of model inputs and the inverse modeling parameter estimates. Based on these results and analysis, we conclude that the impedance parameter in frozen soil is strongly related to both the soil ice content and the size of formed ice, which is affected by the freezing rate and by the size of soil pores related to the soil texture.