Towards improved monitoring of changing permafrost by estimating soil characteristics from ground temperature time-series

Knowledge of subsurface liquid water content is important in permafrost but continuous measurements are rarely collected at monitoring sites. Two parameter estimation methods are used to estimate soil thermal properties and freezing characteristic curves (SFCC) from temperature time series in order to calculate changes in ground liquid water content. Tests with synthetic data show that even with the addition of noise, estimated SFCCs are visually similar to their true shape. Overall, saturation water content and freezing temperature are easiest to estimate, whereas heat capacity and the van Genuchten n parameter, which controls the curvature of the SFCC, are more difficult. Different calibration periods may result in high variability for estimates of low sensitivity parameters. Weighting model error by ground energy content underestimates saturation water content, but provides good estimates of freezing point temperature. Applying these techniques at monitoring sites encounters challenges when model structure is not well chosen.